Generic graphs (common to directed/undirected)#

This module implements the base class for graphs and digraphs, and methods that can be applied on both. Here is what it can do:

Basic Graph operations:

networkx_graph()

Return a new NetworkX graph from the Sage graph

igraph_graph()

Return an igraph graph from the Sage graph

to_dictionary()

Create a dictionary encoding the graph.

copy()

Return a copy of the graph.

export_to_file()

Export the graph to a file.

adjacency_matrix()

Return the adjacency matrix of the (di)graph.

incidence_matrix()

Return an incidence matrix of the (di)graph

distance_matrix()

Return the distance matrix of the (strongly) connected (di)graph

weighted_adjacency_matrix()

Return the weighted adjacency matrix of the graph

kirchhoff_matrix()

Return the Kirchhoff matrix (a.k.a. the Laplacian) of the graph.

has_loops()

Return whether there are loops in the (di)graph

allows_loops()

Return whether loops are permitted in the (di)graph

allow_loops()

Change whether loops are permitted in the (di)graph

loops()

Return a list of all loops in the (di)graph

loop_edges()

Return a list of all loops in the (di)graph

number_of_loops()

Return the number of edges that are loops

loop_vertices()

Return a list of vertices with loops

remove_loops()

Remove loops on vertices in vertices.

has_multiple_edges()

Return whether there are multiple edges in the (di)graph.

allows_multiple_edges()

Return whether multiple edges are permitted in the (di)graph.

allow_multiple_edges()

Change whether multiple edges are permitted in the (di)graph.

multiple_edges()

Return any multiple edges in the (di)graph.

name()

Return or set the graph’s name.

is_immutable()

Return whether the graph is immutable.

weighted()

Whether the (di)graph is to be considered as a weighted (di)graph.

antisymmetric()

Test whether the graph is antisymmetric

density()

Return the density

order()

Return the number of vertices.

size()

Return the number of edges.

add_vertex()

Create an isolated vertex.

add_vertices()

Add vertices to the (di)graph from an iterable container of vertices

delete_vertex()

Delete vertex, removing all incident edges.

delete_vertices()

Delete vertices from the (di)graph taken from an iterable container of vertices.

has_vertex()

Check if vertex is one of the vertices of this graph.

random_vertex()

Return a random vertex of self.

random_vertex_iterator()

Return an iterator over random vertices of self.

random_edge()

Return a random edge of self.

random_edge_iterator()

Return an iterator over random edges of self.

vertex_boundary()

Return a list of all vertices in the external boundary of vertices1, intersected with vertices2.

set_vertices()

Associate arbitrary objects with each vertex

set_vertex()

Associate an arbitrary object with a vertex.

get_vertex()

Retrieve the object associated with a given vertex.

get_vertices()

Return a dictionary of the objects associated to each vertex.

vertex_iterator()

Return an iterator over the given vertices.

neighbor_iterator()

Return an iterator over neighbors of vertex.

vertices()

Return a list of the vertices.

neighbors()

Return a list of neighbors (in and out if directed) of vertex.

merge_vertices()

Merge vertices.

add_edge()

Add an edge from u to v.

add_edges()

Add edges from an iterable container.

subdivide_edge()

Subdivide an edge \(k\) times.

subdivide_edges()

Subdivide \(k\) times edges from an iterable container.

delete_edge()

Delete the edge from u to v

delete_edges()

Delete edges from an iterable container.

contract_edge()

Contract an edge from u to v.

contract_edges()

Contract edges from an iterable container.

delete_multiedge()

Delete all edges from u to v.

set_edge_label()

Set the edge label of a given edge.

has_edge()

Check whether (u, v) is an edge of the (di)graph.

edges()

Return a EdgesView of edges.

edge_boundary()

Return a list of edges (u,v,l) with u in vertices1

edge_iterator()

Return an iterator over edges.

edges_incident()

Return incident edges to some vertices.

edge_label()

Return the label of an edge.

edge_labels()

Return a list of the labels of all edges in self.

remove_multiple_edges()

Remove all multiple edges, retaining one edge for each.

clear()

Empty the graph of vertices and edges and removes name, associated objects, and position information.

degree()

Return the degree (in + out for digraphs) of a vertex or of vertices.

average_degree()

Return the average degree of the graph.

degree_histogram()

Return a list, whose ith entry is the frequency of degree i.

degree_iterator()

Return an iterator over the degrees of the (di)graph.

degree_sequence()

Return the degree sequence of this (di)graph.

random_subgraph()

Return a random subgraph containing each vertex with probability p.

add_clique()

Add a clique to the graph with the given vertices.

add_cycle()

Add a cycle to the graph with the given vertices.

add_path()

Add a path to the graph with the given vertices.

complement()

Return the complement of the (di)graph.

line_graph()

Return the line graph of the (di)graph.

to_simple()

Return a simple version of itself (i.e., undirected and loops and multiple edges are removed).

disjoint_union()

Return the disjoint union of self and other.

union()

Return the union of self and other.

relabel()

Relabel the vertices of self

degree_to_cell()

Return the number of edges from vertex to an edge in cell.

subgraph()

Return the subgraph containing the given vertices and edges.

is_subgraph()

Check whether self is a subgraph of other.

Graph products:

cartesian_product()

Return the Cartesian product of self and other.

tensor_product()

Return the tensor product, also called the categorical product, of self and other.

lexicographic_product()

Return the lexicographic product of self and other.

strong_product()

Return the strong product of self and other.

disjunctive_product()

Return the disjunctive product of self and other.

Paths and cycles:

eulerian_orientation()

Return a DiGraph which is an Eulerian orientation of the current graph.

eulerian_circuit()

Return a list of edges forming an Eulerian circuit if one exists.

minimum_cycle_basis()

Return a minimum weight cycle basis of the graph.

cycle_basis()

Return a list of cycles which form a basis of the cycle space of self.

all_paths()

Return a list of all paths (also lists) between a pair of vertices in the (di)graph.

triangles_count()

Return the number of triangles in the (di)graph.

shortest_simple_paths()

Return an iterator over the simple paths between a pair of vertices.

Linear algebra:

spectrum()

Return a list of the eigenvalues of the adjacency matrix.

eigenvectors()

Return the right eigenvectors of the adjacency matrix of the graph.

eigenspaces()

Return the right eigenspaces of the adjacency matrix of the graph.

Some metrics:

cluster_triangles()

Return the number of triangles for the set nbunch of vertices as a dictionary keyed by vertex.

clustering_average()

Return the average clustering coefficient.

clustering_coeff()

Return the clustering coefficient for each vertex in nbunch

cluster_transitivity()

Return the transitivity (fraction of transitive triangles) of the graph.

szeged_index()

Return the Szeged index of the graph.

katz_centrality()

Return the katz centrality of the vertex u of the graph.

katz_matrix()

Return the katz matrix of the graph.

pagerank()

Return the PageRank of the vertices of self.

Automorphism group:

coarsest_equitable_refinement()

Return the coarsest partition which is finer than the input partition, and equitable with respect to self.

automorphism_group()

Return the largest subgroup of the automorphism group of the (di)graph whose orbit partition is finer than the partition given.

is_vertex_transitive()

Return whether the automorphism group of self is transitive within the partition provided

is_isomorphic()

Test for isomorphism between self and other.

canonical_label()

Return the canonical graph.

is_cayley()

Check whether the graph is a Cayley graph.

Graph properties:

is_eulerian()

Return True if the graph has a (closed) tour that visits each edge exactly once.

is_planar()

Check whether the graph is planar.

is_circular_planar()

Check whether the graph is circular planar (outerplanar)

is_regular()

Return True if this graph is (\(k\)-)regular.

is_chordal()

Check whether the given graph is chordal.

is_bipartite()

Test whether the given graph is bipartite.

is_circulant()

Check whether the graph is a circulant graph.

is_interval()

Check whether the graph is an interval graph.

is_gallai_tree()

Return whether the current graph is a Gallai tree.

is_clique()

Check whether a set of vertices is a clique

is_cycle()

Check whether self is a (directed) cycle graph.

is_geodetic()

Check whether the input (di)graph is geodetic.

is_independent_set()

Check whether vertices is an independent set of self

is_transitively_reduced()

Test whether the digraph is transitively reduced.

is_equitable()

Check whether the given partition is equitable with respect to self.

is_self_complementary()

Check whether the graph is self-complementary.

Traversals:

breadth_first_search()

Return an iterator over the vertices in a breadth-first ordering.

depth_first_search()

Return an iterator over the vertices in a depth-first ordering.

lex_BFS()

Perform a lexicographic breadth first search (LexBFS) on the graph.

lex_UP()

Perform a lexicographic UP search (LexUP) on the graph.

lex_DFS()

Perform a lexicographic depth first search (LexDFS) on the graph.

lex_DOWN()

Perform a lexicographic DOWN search (LexDOWN) on the graph.

Distances:

centrality_betweenness()

Return the betweenness centrality

centrality_closeness()

Returns the closeness centrality (1/average distance to all vertices)

distance()

Return the (directed) distance from u to v in the (di)graph

distance_all_pairs()

Return the distances between all pairs of vertices.

distances_distribution()

Return the distances distribution of the (di)graph in a dictionary.

girth()

Return the girth of the graph.

odd_girth()

Return the odd girth of the graph.

shortest_path()

Return a list of vertices representing some shortest path from \(u\) to \(v\)

shortest_path_length()

Return the minimal length of paths from u to v

shortest_paths()

Return a dictionary associating to each vertex v a shortest path from u to v, if it exists.

shortest_path_lengths()

Return a dictionary of shortest path lengths keyed by targets that are connected by a path from u.

shortest_path_all_pairs()

Compute a shortest path between each pair of vertices.

wiener_index()

Return the Wiener index of the graph.

average_distance()

Return the average distance between vertices of the graph.

Flows, connectivity, trees:

is_connected()

Test whether the (di)graph is connected.

connected_components()

Return the list of connected components

connected_components_number()

Return the number of connected components.

connected_components_subgraphs()

Return a list of connected components as graph objects.

connected_component_containing_vertex()

Return a list of the vertices connected to vertex.

connected_components_sizes()

Return the sizes of the connected components as a list.

blocks_and_cut_vertices()

Compute the blocks and cut vertices of the graph.

blocks_and_cuts_tree()

Compute the blocks-and-cuts tree of the graph.

is_cut_edge()

Return True if the input edge is a cut-edge or a bridge.

is_cut_vertex()

Return True if the input vertex is a cut-vertex.

edge_cut()

Return a minimum edge cut between vertices \(s\) and \(t\)

vertex_cut()

Return a minimum vertex cut between non-adjacent vertices \(s\) and \(t\)

flow()

Return a maximum flow in the graph from x to y

nowhere_zero_flow()

Return a \(k\)-nowhere zero flow of the (di)graph.

edge_disjoint_paths()

Return a list of edge-disjoint paths between two vertices

vertex_disjoint_paths()

Return a list of vertex-disjoint paths between two vertices

edge_connectivity()

Return the edge connectivity of the graph.

vertex_connectivity()

Return the vertex connectivity of the graph.

transitive_closure()

Compute the transitive closure of a graph and returns it.

transitive_reduction()

Return a transitive reduction of a graph.

min_spanning_tree()

Return the edges of a minimum spanning tree.

spanning_trees_count()

Return the number of spanning trees in a graph.

dominator_tree()

Returns a dominator tree of the graph.

connected_subgraph_iterator()

Iterator over the induced connected subgraphs of order at most \(k\)

Plot/embedding-related methods:

set_embedding()

Set a combinatorial embedding dictionary to _embedding attribute.

get_embedding()

Return the attribute _embedding if it exists.

faces()

Return the faces of an embedded graph.

genus()

Return the number of faces of an embedded graph.

planar_dual()

Return the planar dual of an embedded graph.

get_pos()

Return the position dictionary

set_pos()

Set the position dictionary.

layout_planar()

Compute a planar layout of the graph using Schnyder’s algorithm.

is_drawn_free_of_edge_crossings()

Check whether the position dictionary gives a planar embedding.

latex_options()

Return an instance of GraphLatex for the graph.

set_latex_options()

Set multiple options for rendering a graph with LaTeX.

layout()

Return a layout for the vertices of this graph.

layout_spring()

Return a spring layout for this graph

layout_ranked()

Return a ranked layout for this graph

layout_extend_randomly()

Extend randomly a partial layout

layout_circular()

Return a circular layout for this graph

layout_tree()

Return an ordered tree layout for this graph

layout_forest()

Return an ordered forest layout for this graph

layout_graphviz()

Call graphviz to compute a layout of the vertices of this graph.

_circle_embedding()

Set some vertices on a circle in the embedding of this graph.

_line_embedding()

Set some vertices on a line in the embedding of this graph.

graphplot()

Return a GraphPlot object.

plot()

Return a Graphics object representing the (di)graph.

show()

Show the (di)graph.

plot3d()

Plot the graph in three dimensions.

show3d()

Plot the graph using Tachyon, and shows the resulting plot.

graphviz_string()

Return a representation in the dot language.

graphviz_to_file_named()

Write a representation in the dot language in a file.

Algorithmically hard stuff:

steiner_tree()

Return a tree of minimum weight connecting the given set of vertices.

edge_disjoint_spanning_trees()

Return the desired number of edge-disjoint spanning trees/arborescences.

feedback_vertex_set()

Compute the minimum feedback vertex set of a (di)graph.

multiway_cut()

Return a minimum edge multiway cut

max_cut()

Return a maximum edge cut of the graph.

longest_path()

Return a longest path of self.

traveling_salesman_problem()

Solve the traveling salesman problem (TSP)

is_hamiltonian()

Test whether the current graph is Hamiltonian.

hamiltonian_cycle()

Return a Hamiltonian cycle/circuit of the current graph/digraph

hamiltonian_path()

Return a Hamiltonian path of the current graph/digraph

multicommodity_flow()

Solve a multicommodity flow problem.

disjoint_routed_paths()

Return a set of disjoint routed paths.

dominating_set()

Return a minimum dominating set of the graph

greedy_dominating_set()

Return a greedy distance-\(k\) dominating set of the graph.

maximum_leaf_number()

Return the maximum leaf number of the graph.

subgraph_search()

Return a copy of G in self.

subgraph_search_count()

Return the number of labelled occurrences of G in self.

subgraph_search_iterator()

Return an iterator over the labelled copies of G in self.

characteristic_polynomial()

Return the characteristic polynomial of the adjacency matrix of the (di)graph.

genus()

Return the minimal genus of the graph.

crossing_number()

Return the crossing number of the graph.

Miscellaneous

edge_polytope()

Return the edge polytope of self.

symmetric_edge_polytope()

Return the symmetric edge polytope of self.

Methods#

class sage.graphs.generic_graph.GenericGraph#

Bases: GenericGraph_pyx

Base class for graphs and digraphs.

__eq__(other)#

Compare self and other for equality.

Do not call this method directly. That is, for G.__eq__(H) write G == H.

Two graphs are considered equal if the following hold:
  • they are either both directed, or both undirected;

  • they have the same settings for loops, multiedges, and weightedness;

  • they have the same set of vertices;

  • they have the same (multi)set of arrows/edges, where labels of arrows/edges are taken into account if and only if the graphs are considered weighted. See weighted().

Note that this is not an isomorphism test.

EXAMPLES:

sage: G = graphs.EmptyGraph()
sage: H = Graph()
sage: G == H
True
sage: G.to_directed() == H.to_directed()
True
sage: G = graphs.RandomGNP(8, .9999)
sage: H = graphs.CompleteGraph(8)
sage: G == H  # random - most often true
True
sage: G = Graph({0: [1, 2, 3, 4, 5, 6, 7]} )
sage: H = Graph({1: [0], 2: [0], 3: [0], 4: [0], 5: [0], 6: [0], 7: [0]} )
sage: G == H
True
sage: G.allow_loops(True)
sage: G == H
False
sage: G = graphs.RandomGNP(9, .3).to_directed()
sage: H = graphs.RandomGNP(9, .3).to_directed()
sage: G == H # most often false
False
sage: G = Graph(multiedges=True, sparse=True)
sage: G.add_edge(0, 1)
sage: H = copy(G)
sage: H.add_edge(0, 1)
sage: G == H
False

Note that graphs must be considered weighted, or Sage will not pay attention to edge label data in equality testing:

sage: foo = Graph(sparse=True)
sage: foo.add_edges([(0, 1, 1), (0, 2, 2)])
sage: bar = Graph(sparse=True)
sage: bar.add_edges([(0, 1, 2), (0, 2, 1)])
sage: foo == bar
True
sage: foo.weighted(True)
sage: foo == bar
False
sage: bar.weighted(True)
sage: foo == bar
False
add_clique(vertices, loops=False)#

Add a clique to the graph with the given vertices.

If the vertices are already present, only the edges are added.

INPUT:

  • vertices – an iterable container of vertices for the clique to be added, e.g. a list, set, graph, etc.

  • loops – boolean (default: False); whether to add edges from every given vertex to itself. This is allowed only if the (di)graph allows loops.

EXAMPLES:

sage: G = Graph()
sage: G.add_clique(range(4))
sage: G.is_isomorphic(graphs.CompleteGraph(4))
True
sage: D = DiGraph()
sage: D.add_clique(range(4))
sage: D.is_isomorphic(digraphs.Complete(4))
True
sage: D = DiGraph(loops=True)
sage: D.add_clique(range(4), loops=True)
sage: D.is_isomorphic(digraphs.Complete(4, loops=True))
True
sage: D = DiGraph(loops=False)
sage: D.add_clique(range(4), loops=True)
Traceback (most recent call last):
...
ValueError: cannot add edge from 0 to 0 in graph without loops

If the list of vertices contains repeated elements, a loop will be added at that vertex, even if loops=False:

sage: G = Graph(loops=True)
sage: G.add_clique([1, 1])
sage: G.edges(sort=True)
[(1, 1, None)]

This is equivalent to:

sage: G = Graph(loops=True)
sage: G.add_clique([1], loops=True)
sage: G.edges(sort=True)
[(1, 1, None)]
add_cycle(vertices)#

Add a cycle to the graph with the given vertices.

If the vertices are already present, only the edges are added.

For digraphs, adds the directed cycle, whose orientation is determined by the list. Adds edges (vertices[u], vertices[u+1]) and (vertices[-1], vertices[0]).

INPUT:

  • vertices – an ordered list of the vertices of the cycle to be added

EXAMPLES:

sage: G = Graph()
sage: G.add_vertices(range(10)); G
Graph on 10 vertices
sage: show(G)                                                               # needs sage.plot
sage: G.add_cycle(list(range(10, 20)))
sage: show(G)                                                               # needs sage.plot
sage: G.add_cycle(list(range(10)))
sage: show(G)                                                               # needs sage.plot
sage: D = DiGraph()
sage: D.add_cycle(list(range(4)))
sage: D.edges(sort=True)
[(0, 1, None), (1, 2, None), (2, 3, None), (3, 0, None)]
add_edge(u, v=None, label=None)#

Add an edge from u to v.

INPUT: The following forms are all accepted:

  • G.add_edge( 1, 2 )

  • G.add_edge( (1, 2) )

  • G.add_edges( [ (1, 2) ])

  • G.add_edge( 1, 2, ‘label’ )

  • G.add_edge( (1, 2, ‘label’) )

  • G.add_edges( [ (1, 2, ‘label’) ] )

WARNING: The following intuitive input results in nonintuitive output:

sage: G = Graph()
sage: G.add_edge((1, 2), 'label')
sage: G.edges(sort=False)
[('label', (1, 2), None)]

You must either use the label keyword:

sage: G = Graph()
sage: G.add_edge((1, 2), label="label")
sage: G.edges(sort=False)
[(1, 2, 'label')]

Or use one of these:

sage: G = Graph()
sage: G.add_edge(1, 2, 'label')
sage: G.edges(sort=False)
[(1, 2, 'label')]
sage: G = Graph()
sage: G.add_edge((1, 2, 'label'))
sage: G.edges(sort=False)
[(1, 2, 'label')]

Vertex name cannot be None, so:

sage: G = Graph()
sage: G.add_edge(None, 4)
sage: G.vertices(sort=True)
[0, 4]
add_edges(edges, loops=True)#

Add edges from an iterable container.

INPUT:

  • edges – an iterable of edges, given either as (u, v) or (u, v, label).

  • loops – boolean (default: True); if False, remove all loops (v, v) from the input iterator. If None, remove loops unless the graph allows loops.

EXAMPLES:

sage: G = graphs.DodecahedralGraph()
sage: H = Graph()
sage: H.add_edges(G.edge_iterator()); H
Graph on 20 vertices
sage: G = graphs.DodecahedralGraph().to_directed()
sage: H = DiGraph()
sage: H.add_edges(G.edge_iterator()); H
Digraph on 20 vertices
sage: H.add_edges(iter([]))

sage: H = Graph()
sage: H.add_edges([(0, 1), (0, 2, "label")])
sage: H.edges(sort=True)
[(0, 1, None), (0, 2, 'label')]

We demonstrate the loops argument:

sage: H = Graph()
sage: H.add_edges([(0, 0)], loops=False); H.edges(sort=True)
[]
sage: H.add_edges([(0, 0)], loops=None); H.edges(sort=True)
[]
sage: H.add_edges([(0, 0)]); H.edges(sort=True)
Traceback (most recent call last):
...
ValueError: cannot add edge from 0 to 0 in graph without loops
sage: H = Graph(loops=True)
sage: H.add_edges([(0, 0)], loops=False); H.edges(sort=True)
[]
sage: H.add_edges([(0, 0)], loops=None); H.edges(sort=True)
[(0, 0, None)]
sage: H.add_edges([(0, 0)]); H.edges(sort=True)
[(0, 0, None)]
add_path(vertices)#

Add a path to the graph with the given vertices.

If the vertices are already present, only the edges are added.

For digraphs, adds the directed path vertices[0], ..., vertices[-1].

INPUT:

  • vertices – an ordered list of the vertices of the path to be added

EXAMPLES:

sage: G = Graph()
sage: G.add_vertices(range(10)); G
Graph on 10 vertices
sage: show(G)                                                               # needs sage.plot
sage: G.add_path(list(range(10, 20)))
sage: show(G)                                                               # needs sage.plot
sage: G.add_path(list(range(10)))
sage: show(G)                                                               # needs sage.plot
sage: D = DiGraph()
sage: D.add_path(list(range(4)))
sage: D.edges(sort=True)
[(0, 1, None), (1, 2, None), (2, 3, None)]
add_vertex(name=None)#

Create an isolated vertex.

If the vertex already exists, then nothing is done.

INPUT:

  • name – an immutable object (default: None); when no name is specified (default), then the new vertex will be represented by the least integer not already representing a vertex. name must be an immutable object (e.g., an integer, a tuple, etc.).

As it is implemented now, if a graph \(G\) has a large number of vertices with numeric labels, then G.add_vertex() could potentially be slow, if name=None.

OUTPUT:

If name=None, the new vertex name is returned. None otherwise.

EXAMPLES:

sage: G = Graph(); G.add_vertex(); G
0
Graph on 1 vertex
sage: D = DiGraph(); D.add_vertex(); D
0
Digraph on 1 vertex
add_vertices(vertices)#

Add vertices to the (di)graph from an iterable container of vertices.

Vertices that already exist in the graph will not be added again.

INPUT:

  • vertices – iterator container of vertex labels. A new label is created, used and returned in the output list for all None values in vertices.

OUTPUT:

Generated names of new vertices if there is at least one None value present in vertices. None otherwise.

EXAMPLES:

sage: d = {0: [1,4,5], 1: [2,6], 2: [3,7], 3: [4,8], 4: [9], 5: [7,8], 6: [8,9], 7: [9]}
sage: G = Graph(d)
sage: G.add_vertices([10,11,12])
sage: G.vertices(sort=True)
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]
sage: G.add_vertices(graphs.CycleGraph(25).vertex_iterator())
sage: G.vertices(sort=True)
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24]
sage: G = Graph()
sage: G.add_vertices([1, 2, 3])
sage: G.add_vertices([4, None, None, 5])
[0, 6]
adjacency_matrix(sparse, vertices=None, base_ring=None, **kwds)#

Return the adjacency matrix of the (di)graph.

By default, the matrix returned is over the integers.

INPUT:

  • sparse – boolean (default: None); whether to represent with a sparse matrix

  • vertices – list (default: None); the ordering of the vertices defining how they should appear in the matrix. By default, the ordering given by GenericGraph.vertices() with sort=True is used. If the vertices are not comparable, the keyword vertices must be used to specify an ordering, or a TypeError exception will be raised.

  • base_ring – a ring (default: ZZ); the base ring of the matrix space to use.

  • **kwds – other keywords to pass to matrix().

EXAMPLES:

sage: G = graphs.CubeGraph(4)
sage: G.adjacency_matrix()                                                  # needs sage.modules
[0 1 1 0 1 0 0 0 1 0 0 0 0 0 0 0]
[1 0 0 1 0 1 0 0 0 1 0 0 0 0 0 0]
[1 0 0 1 0 0 1 0 0 0 1 0 0 0 0 0]
[0 1 1 0 0 0 0 1 0 0 0 1 0 0 0 0]
[1 0 0 0 0 1 1 0 0 0 0 0 1 0 0 0]
[0 1 0 0 1 0 0 1 0 0 0 0 0 1 0 0]
[0 0 1 0 1 0 0 1 0 0 0 0 0 0 1 0]
[0 0 0 1 0 1 1 0 0 0 0 0 0 0 0 1]
[1 0 0 0 0 0 0 0 0 1 1 0 1 0 0 0]
[0 1 0 0 0 0 0 0 1 0 0 1 0 1 0 0]
[0 0 1 0 0 0 0 0 1 0 0 1 0 0 1 0]
[0 0 0 1 0 0 0 0 0 1 1 0 0 0 0 1]
[0 0 0 0 1 0 0 0 1 0 0 0 0 1 1 0]
[0 0 0 0 0 1 0 0 0 1 0 0 1 0 0 1]
[0 0 0 0 0 0 1 0 0 0 1 0 1 0 0 1]
[0 0 0 0 0 0 0 1 0 0 0 1 0 1 1 0]
sage: matrix(GF(2), G)  # matrix over GF(2)                                 # needs sage.modules sage.rings.finite_rings
[0 1 1 0 1 0 0 0 1 0 0 0 0 0 0 0]
[1 0 0 1 0 1 0 0 0 1 0 0 0 0 0 0]
[1 0 0 1 0 0 1 0 0 0 1 0 0 0 0 0]
[0 1 1 0 0 0 0 1 0 0 0 1 0 0 0 0]
[1 0 0 0 0 1 1 0 0 0 0 0 1 0 0 0]
[0 1 0 0 1 0 0 1 0 0 0 0 0 1 0 0]
[0 0 1 0 1 0 0 1 0 0 0 0 0 0 1 0]
[0 0 0 1 0 1 1 0 0 0 0 0 0 0 0 1]
[1 0 0 0 0 0 0 0 0 1 1 0 1 0 0 0]
[0 1 0 0 0 0 0 0 1 0 0 1 0 1 0 0]
[0 0 1 0 0 0 0 0 1 0 0 1 0 0 1 0]
[0 0 0 1 0 0 0 0 0 1 1 0 0 0 0 1]
[0 0 0 0 1 0 0 0 1 0 0 0 0 1 1 0]
[0 0 0 0 0 1 0 0 0 1 0 0 1 0 0 1]
[0 0 0 0 0 0 1 0 0 0 1 0 1 0 0 1]
[0 0 0 0 0 0 0 1 0 0 0 1 0 1 1 0]
sage: D = DiGraph({0: [1, 2, 3], 1: [0, 2], 2: [3],
....:              3: [4], 4: [0, 5], 5: [1]})
sage: D.adjacency_matrix()                                                  # needs sage.modules
[0 1 1 1 0 0]
[1 0 1 0 0 0]
[0 0 0 1 0 0]
[0 0 0 0 1 0]
[1 0 0 0 0 1]
[0 1 0 0 0 0]

A different ordering of the vertices:

sage: graphs.PathGraph(5).adjacency_matrix(vertices=[2, 4, 1, 3, 0])        # needs sage.modules
[0 0 1 1 0]
[0 0 0 1 0]
[1 0 0 0 1]
[1 1 0 0 0]
[0 0 1 0 0]

A different base ring:

sage: graphs.PathGraph(5).adjacency_matrix(base_ring=RDF)                   # needs sage.modules
[0.0 1.0 0.0 0.0 0.0]
[1.0 0.0 1.0 0.0 0.0]
[0.0 1.0 0.0 1.0 0.0]
[0.0 0.0 1.0 0.0 1.0]
[0.0 0.0 0.0 1.0 0.0]
sage: type(_)                                                               # needs sage.modules
<class 'sage.matrix.matrix_real_double_dense.Matrix_real_double_dense'>

A different matrix implementation:

sage: graphs.PathGraph(5).adjacency_matrix(sparse=False,                    # needs numpy sage.modules
....:                                      implementation='numpy')
[0 1 0 0 0]
[1 0 1 0 0]
[0 1 0 1 0]
[0 0 1 0 1]
[0 0 0 1 0]
sage: type(_)
<class 'sage.matrix.matrix_numpy_integer_dense.Matrix_numpy_integer_dense'>

As an immutable matrix:

sage: M = graphs.PathGraph(5).adjacency_matrix(sparse=False,                # needs sage.modules
....:                                          immutable=True); M
[0 1 0 0 0]
[1 0 1 0 0]
[0 1 0 1 0]
[0 0 1 0 1]
[0 0 0 1 0]
sage: M[2, 2] = 1                                                           # needs sage.modules
Traceback (most recent call last):
...
ValueError: matrix is immutable; please change a copy instead
(i.e., use copy(M) to change a copy of M).
all_paths(G, start, end, use_multiedges=False, report_edges=False, labels=False)#

Return the list of all paths between a pair of vertices.

If start is the same vertex as end, then [[start]] is returned – a list containing the 1-vertex, 0-edge path “start”.

If G has multiple edges, a path will be returned as many times as the product of the multiplicity of the edges along that path depending on the value of the flag use_multiedges.

INPUT:

  • start – a vertex of a graph, where to start

  • end – a vertex of a graph, where to end

  • use_multiedges – boolean (default: False); this parameter is used only if the graph has multiple edges.

    • If False, the graph is considered as simple and an edge label is arbitrarily selected for each edge as in sage.graphs.generic_graph.GenericGraph.to_simple() if report_edges is True

    • If True, a path will be reported as many times as the edges multiplicities along that path (when report_edges = False or labels = False), or with all possible combinations of edge labels (when report_edges = True and labels = True)

  • report_edges – boolean (default: False); whether to report paths as list of vertices (default) or list of edges, if False then labels parameter is ignored

  • labels – boolean (default: False); if False, each edge is simply a pair (u, v) of vertices. Otherwise a list of edges along with its edge labels are used to represent the path.

EXAMPLES:

sage: eg1 = Graph({0:[1, 2], 1:[4], 2:[3, 4], 4:[5], 5:[6]})
sage: eg1.all_paths(0, 6)
[[0, 1, 4, 5, 6], [0, 2, 4, 5, 6]]
sage: eg2 = graphs.PetersenGraph()
sage: sorted(eg2.all_paths(1, 4))
[[1, 0, 4],
 [1, 0, 5, 7, 2, 3, 4],
 [1, 0, 5, 7, 2, 3, 8, 6, 9, 4],
 [1, 0, 5, 7, 9, 4],
 [1, 0, 5, 7, 9, 6, 8, 3, 4],
 [1, 0, 5, 8, 3, 2, 7, 9, 4],
 [1, 0, 5, 8, 3, 4],
 [1, 0, 5, 8, 6, 9, 4],
 [1, 0, 5, 8, 6, 9, 7, 2, 3, 4],
 [1, 2, 3, 4],
 [1, 2, 3, 8, 5, 0, 4],
 [1, 2, 3, 8, 5, 7, 9, 4],
 [1, 2, 3, 8, 6, 9, 4],
 [1, 2, 3, 8, 6, 9, 7, 5, 0, 4],
 [1, 2, 7, 5, 0, 4],
 [1, 2, 7, 5, 8, 3, 4],
 [1, 2, 7, 5, 8, 6, 9, 4],
 [1, 2, 7, 9, 4],
 [1, 2, 7, 9, 6, 8, 3, 4],
 [1, 2, 7, 9, 6, 8, 5, 0, 4],
 [1, 6, 8, 3, 2, 7, 5, 0, 4],
 [1, 6, 8, 3, 2, 7, 9, 4],
 [1, 6, 8, 3, 4],
 [1, 6, 8, 5, 0, 4],
 [1, 6, 8, 5, 7, 2, 3, 4],
 [1, 6, 8, 5, 7, 9, 4],
 [1, 6, 9, 4],
 [1, 6, 9, 7, 2, 3, 4],
 [1, 6, 9, 7, 2, 3, 8, 5, 0, 4],
 [1, 6, 9, 7, 5, 0, 4],
 [1, 6, 9, 7, 5, 8, 3, 4]]
sage: dg = DiGraph({0:[1, 3], 1:[3], 2:[0, 3]})
sage: sorted(dg.all_paths(0, 3))
[[0, 1, 3], [0, 3]]
sage: ug = dg.to_undirected()
sage: sorted(ug.all_paths(0, 3))
[[0, 1, 3], [0, 2, 3], [0, 3]]

sage: g = Graph([(0, 1), (0, 1), (1, 2), (1, 2)], multiedges=True)
sage: g.all_paths(0, 2, use_multiedges=True)
[[0, 1, 2], [0, 1, 2], [0, 1, 2], [0, 1, 2]]

sage: dg = DiGraph({0:[1, 2, 1], 3:[0, 0]}, multiedges=True)
sage: dg.all_paths(3, 1, use_multiedges=True)
[[3, 0, 1], [3, 0, 1], [3, 0, 1], [3, 0, 1]]

sage: g = Graph([(0, 1, 'a'), (0, 1, 'b'), (1, 2, 'c'), (1, 2, 'd')], multiedges=True)
sage: g.all_paths(0, 2, use_multiedges=False)
[[0, 1, 2]]
sage: g.all_paths(0, 2, use_multiedges=True)
[[0, 1, 2], [0, 1, 2], [0, 1, 2], [0, 1, 2]]
sage: g.all_paths(0, 2, use_multiedges=True, report_edges=True)
[[(0, 1), (1, 2)], [(0, 1), (1, 2)], [(0, 1), (1, 2)], [(0, 1), (1, 2)]]
sage: g.all_paths(0, 2, use_multiedges=True, report_edges=True, labels=True)
[((0, 1, 'b'), (1, 2, 'd')),
 ((0, 1, 'b'), (1, 2, 'c')),
 ((0, 1, 'a'), (1, 2, 'd')),
 ((0, 1, 'a'), (1, 2, 'c'))]
sage: g.all_paths(0, 2, use_multiedges=False, report_edges=True, labels=True)
[((0, 1, 'b'), (1, 2, 'd'))]
sage: g.all_paths(0, 2, use_multiedges=False, report_edges=False, labels=True)
[[0, 1, 2]]
sage: g.all_paths(0, 2, use_multiedges=True, report_edges=False, labels=True)
[[0, 1, 2], [0, 1, 2], [0, 1, 2], [0, 1, 2]]
allow_loops(new, check=True)#

Change whether loops are permitted in the (di)graph

INPUT:

  • new – boolean

  • check – boolean (default: True); whether to remove existing loops from the (di)graph when the new status is False

EXAMPLES:

sage: G = Graph(loops=True); G
Looped graph on 0 vertices
sage: G.has_loops()
False
sage: G.allows_loops()
True
sage: G.add_edge((0, 0))
sage: G.has_loops()
True
sage: G.loops()
[(0, 0, None)]
sage: G.allow_loops(False); G
Graph on 1 vertex
sage: G.has_loops()
False
sage: G.edges(sort=True)
[]

sage: D = DiGraph(loops=True); D
Looped digraph on 0 vertices
sage: D.has_loops()
False
sage: D.allows_loops()
True
sage: D.add_edge((0, 0))
sage: D.has_loops()
True
sage: D.loops()
[(0, 0, None)]
sage: D.allow_loops(False); D
Digraph on 1 vertex
sage: D.has_loops()
False
sage: D.edges(sort=True)
[]
allow_multiple_edges(new, check=True, keep_label='any')#

Change whether multiple edges are permitted in the (di)graph.

INPUT:

  • new – boolean; if True, the new graph will allow multiple edges

  • check – boolean (default: True); if True and new is False, we remove all multiple edges from the graph

  • keep_label – string (default: 'any'); used only if new is False and check is True. If there are multiple edges with different labels, this variable defines which label should be kept:

    • 'any' – any label

    • 'min' – the smallest label

    • 'max' – the largest label

Warning

'min' and 'max' only works if the labels can be compared. A TypeError might be raised when working with non-comparable objects in Python 3.

EXAMPLES:

The standard behavior with undirected graphs:

sage: G = Graph(multiedges=True, sparse=True); G
Multi-graph on 0 vertices
sage: G.has_multiple_edges()
False
sage: G.allows_multiple_edges()
True
sage: G.add_edges([(0, 1, 1), (0, 1, 2), (0, 1, 3)])
sage: G.has_multiple_edges()
True
sage: G.multiple_edges(sort=True)
[(0, 1, 1), (0, 1, 2), (0, 1, 3)]
sage: G.allow_multiple_edges(False); G
Graph on 2 vertices
sage: G.has_multiple_edges()
False
sage: G.edges(sort=True)
[(0, 1, 3)]

If we ask for the minimum label:

sage: G = Graph([(0, 1, 1), (0, 1, 2), (0, 1, 3)], multiedges=True, sparse=True)
sage: G.allow_multiple_edges(False, keep_label='min')
sage: G.edges(sort=True)
[(0, 1, 1)]

If we ask for the maximum label:

sage: G = Graph([(0, 1, 1), (0, 1, 2), (0, 1, 3)], multiedges=True, sparse=True)
sage: G.allow_multiple_edges(False, keep_label='max')
sage: G.edges(sort=True)
[(0, 1, 3)]

The standard behavior with digraphs:

sage: D = DiGraph(multiedges=True, sparse=True); D
Multi-digraph on 0 vertices
sage: D.has_multiple_edges()
False
sage: D.allows_multiple_edges()
True
sage: D.add_edges([(0, 1)] * 3)
sage: D.has_multiple_edges()
True
sage: D.multiple_edges()
[(0, 1, None), (0, 1, None), (0, 1, None)]
sage: D.allow_multiple_edges(False); D
Digraph on 2 vertices
sage: D.has_multiple_edges()
False
sage: D.edges(sort=True)
[(0, 1, None)]
allows_loops()#

Return whether loops are permitted in the (di)graph

EXAMPLES:

sage: G = Graph(loops=True); G
Looped graph on 0 vertices
sage: G.has_loops()
False
sage: G.allows_loops()
True
sage: G.add_edge((0, 0))
sage: G.has_loops()
True
sage: G.loops()
[(0, 0, None)]
sage: G.allow_loops(False); G
Graph on 1 vertex
sage: G.has_loops()
False
sage: G.edges(sort=True)
[]

sage: D = DiGraph(loops=True); D
Looped digraph on 0 vertices
sage: D.has_loops()
False
sage: D.allows_loops()
True
sage: D.add_edge((0, 0))
sage: D.has_loops()
True
sage: D.loops()
[(0, 0, None)]
sage: D.allow_loops(False); D
Digraph on 1 vertex
sage: D.has_loops()
False
sage: D.edges(sort=True)
[]
allows_multiple_edges()#

Return whether multiple edges are permitted in the (di)graph.

EXAMPLES:

sage: G = Graph(multiedges=True, sparse=True); G
Multi-graph on 0 vertices
sage: G.has_multiple_edges()
False
sage: G.allows_multiple_edges()
True
sage: G.add_edges([(0, 1)] * 3)
sage: G.has_multiple_edges()
True
sage: G.multiple_edges()
[(0, 1, None), (0, 1, None), (0, 1, None)]
sage: G.allow_multiple_edges(False); G
Graph on 2 vertices
sage: G.has_multiple_edges()
False
sage: G.edges(sort=True)
[(0, 1, None)]

sage: D = DiGraph(multiedges=True, sparse=True); D
Multi-digraph on 0 vertices
sage: D.has_multiple_edges()
False
sage: D.allows_multiple_edges()
True
sage: D.add_edges([(0, 1)] * 3)
sage: D.has_multiple_edges()
True
sage: D.multiple_edges()
[(0, 1, None), (0, 1, None), (0, 1, None)]
sage: D.allow_multiple_edges(False); D
Digraph on 2 vertices
sage: D.has_multiple_edges()
False
sage: D.edges(sort=True)
[(0, 1, None)]
am(sparse, vertices=None, base_ring=None, **kwds)#

Return the adjacency matrix of the (di)graph.

By default, the matrix returned is over the integers.

INPUT:

  • sparse – boolean (default: None); whether to represent with a sparse matrix

  • vertices – list (default: None); the ordering of the vertices defining how they should appear in the matrix. By default, the ordering given by GenericGraph.vertices() with sort=True is used. If the vertices are not comparable, the keyword vertices must be used to specify an ordering, or a TypeError exception will be raised.

  • base_ring – a ring (default: ZZ); the base ring of the matrix space to use.

  • **kwds – other keywords to pass to matrix().

EXAMPLES:

sage: G = graphs.CubeGraph(4)
sage: G.adjacency_matrix()                                                  # needs sage.modules
[0 1 1 0 1 0 0 0 1 0 0 0 0 0 0 0]
[1 0 0 1 0 1 0 0 0 1 0 0 0 0 0 0]
[1 0 0 1 0 0 1 0 0 0 1 0 0 0 0 0]
[0 1 1 0 0 0 0 1 0 0 0 1 0 0 0 0]
[1 0 0 0 0 1 1 0 0 0 0 0 1 0 0 0]
[0 1 0 0 1 0 0 1 0 0 0 0 0 1 0 0]
[0 0 1 0 1 0 0 1 0 0 0 0 0 0 1 0]
[0 0 0 1 0 1 1 0 0 0 0 0 0 0 0 1]
[1 0 0 0 0 0 0 0 0 1 1 0 1 0 0 0]
[0 1 0 0 0 0 0 0 1 0 0 1 0 1 0 0]
[0 0 1 0 0 0 0 0 1 0 0 1 0 0 1 0]
[0 0 0 1 0 0 0 0 0 1 1 0 0 0 0 1]
[0 0 0 0 1 0 0 0 1 0 0 0 0 1 1 0]
[0 0 0 0 0 1 0 0 0 1 0 0 1 0 0 1]
[0 0 0 0 0 0 1 0 0 0 1 0 1 0 0 1]
[0 0 0 0 0 0 0 1 0 0 0 1 0 1 1 0]
sage: matrix(GF(2), G)  # matrix over GF(2)                                 # needs sage.modules sage.rings.finite_rings
[0 1 1 0 1 0 0 0 1 0 0 0 0 0 0 0]
[1 0 0 1 0 1 0 0 0 1 0 0 0 0 0 0]
[1 0 0 1 0 0 1 0 0 0 1 0 0 0 0 0]
[0 1 1 0 0 0 0 1 0 0 0 1 0 0 0 0]
[1 0 0 0 0 1 1 0 0 0 0 0 1 0 0 0]
[0 1 0 0 1 0 0 1 0 0 0 0 0 1 0 0]
[0 0 1 0 1 0 0 1 0 0 0 0 0 0 1 0]
[0 0 0 1 0 1 1 0 0 0 0 0 0 0 0 1]
[1 0 0 0 0 0 0 0 0 1 1 0 1 0 0 0]
[0 1 0 0 0 0 0 0 1 0 0 1 0 1 0 0]
[0 0 1 0 0 0 0 0 1 0 0 1 0 0 1 0]
[0 0 0 1 0 0 0 0 0 1 1 0 0 0 0 1]
[0 0 0 0 1 0 0 0 1 0 0 0 0 1 1 0]
[0 0 0 0 0 1 0 0 0 1 0 0 1 0 0 1]
[0 0 0 0 0 0 1 0 0 0 1 0 1 0 0 1]
[0 0 0 0 0 0 0 1 0 0 0 1 0 1 1 0]
sage: D = DiGraph({0: [1, 2, 3], 1: [0, 2], 2: [3],
....:              3: [4], 4: [0, 5], 5: [1]})
sage: D.adjacency_matrix()                                                  # needs sage.modules
[0 1 1 1 0 0]
[1 0 1 0 0 0]
[0 0 0 1 0 0]
[0 0 0 0 1 0]
[1 0 0 0 0 1]
[0 1 0 0 0 0]

A different ordering of the vertices:

sage: graphs.PathGraph(5).adjacency_matrix(vertices=[2, 4, 1, 3, 0])        # needs sage.modules
[0 0 1 1 0]
[0 0 0 1 0]
[1 0 0 0 1]
[1 1 0 0 0]
[0 0 1 0 0]

A different base ring:

sage: graphs.PathGraph(5).adjacency_matrix(base_ring=RDF)                   # needs sage.modules
[0.0 1.0 0.0 0.0 0.0]
[1.0 0.0 1.0 0.0 0.0]
[0.0 1.0 0.0 1.0 0.0]
[0.0 0.0 1.0 0.0 1.0]
[0.0 0.0 0.0 1.0 0.0]
sage: type(_)                                                               # needs sage.modules
<class 'sage.matrix.matrix_real_double_dense.Matrix_real_double_dense'>

A different matrix implementation:

sage: graphs.PathGraph(5).adjacency_matrix(sparse=False,                    # needs numpy sage.modules
....:                                      implementation='numpy')
[0 1 0 0 0]
[1 0 1 0 0]
[0 1 0 1 0]
[0 0 1 0 1]
[0 0 0 1 0]
sage: type(_)
<class 'sage.matrix.matrix_numpy_integer_dense.Matrix_numpy_integer_dense'>

As an immutable matrix:

sage: M = graphs.PathGraph(5).adjacency_matrix(sparse=False,                # needs sage.modules
....:                                          immutable=True); M
[0 1 0 0 0]
[1 0 1 0 0]
[0 1 0 1 0]
[0 0 1 0 1]
[0 0 0 1 0]
sage: M[2, 2] = 1                                                           # needs sage.modules
Traceback (most recent call last):
...
ValueError: matrix is immutable; please change a copy instead
(i.e., use copy(M) to change a copy of M).
antisymmetric()#

Check whether the graph is antisymmetric.

A graph represents an antisymmetric relation if the existence of a path from a vertex \(x\) to a vertex \(y\) implies that there is not a path from \(y\) to \(x\) unless \(x = y\).

EXAMPLES:

A directed acyclic graph is antisymmetric:

sage: G = digraphs.RandomDirectedGNR(20, 0.5)                               # needs networkx
sage: G.antisymmetric()                                                     # needs networkx
True

Loops are allowed:

sage: G.allow_loops(True)                                                   # needs networkx
sage: G.add_edge(0, 0)                                                      # needs networkx
sage: G.antisymmetric()                                                     # needs networkx
True

An undirected graph is never antisymmetric unless it is just a union of isolated vertices (with possible loops):

sage: graphs.RandomGNP(20, 0.5).antisymmetric()                             # needs networkx
False
sage: Graph(3).antisymmetric()
True
sage: Graph([(i, i) for i in range(3)], loops=True).antisymmetric()
True
sage: DiGraph([(i, i) for i in range(3)], loops=True).antisymmetric()
True
automorphism_group(partition=None, verbosity=0, edge_labels=False, order=False, return_group=True, orbits=False, algorithm=None)#

Return the automorphism group of the graph.

With partition this can also return the largest subgroup of the automorphism group of the (di)graph whose orbit partition is finer than the partition given.

INPUT:

  • partition – default is the unit partition, otherwise computes the subgroup of the full automorphism group respecting the partition.

  • edge_labels – default False, otherwise allows only permutations respecting edge labels.

  • order – (default False) if True, compute the order of the automorphism group

  • return_group – default True

  • orbits – returns the orbits of the group acting on the vertices of the graph

  • algorithm – If algorithm = "bliss", the automorphism group is computed using the optional package bliss (http://www.tcs.tkk.fi/Software/bliss/index.html). Setting it to "sage" uses Sage’s implementation. If set to None (default), bliss is used when available.

OUTPUT: The order of the output is group, order, orbits. However, there are options to turn each of these on or off.

EXAMPLES:

Graphs:

sage: # needs sage.groups
sage: graphs_query = GraphQuery(display_cols=['graph6'],num_vertices=4)
sage: L = graphs_query.get_graphs_list()
sage: graphs_list.show_graphs(L)                                            # needs sage.plot
sage: for g in L:
....:     G = g.automorphism_group()
....:     G.order(), G.gens()
(24, ((2,3), (1,2), (0,1)))
(4, ((2,3), (0,1)))
(2, ((1,2),))
(6, ((1,2), (0,1)))
(6, ((2,3), (1,2)))
(8, ((1,2), (0,1)(2,3)))
(2, ((0,1)(2,3),))
(2, ((1,2),))
(8, ((2,3), (0,1), (0,2)(1,3)))
(4, ((2,3), (0,1)))
(24, ((2,3), (1,2), (0,1)))
sage: C = graphs.CubeGraph(4)
sage: G = C.automorphism_group()
sage: M = G.character_table() # random order of rows, thus abs() below
sage: QQ(M.determinant()).abs()
712483534798848
sage: G.order()
384
sage: # needs sage.groups
sage: D = graphs.DodecahedralGraph()
sage: G = D.automorphism_group()
sage: A5 = AlternatingGroup(5)
sage: Z2 = CyclicPermutationGroup(2)
sage: H = A5.direct_product(Z2)[0] #see documentation for direct_product to explain the [0]
sage: G.is_isomorphic(H)
True

Multigraphs:

sage: G = Graph(multiedges=True,sparse=True)
sage: G.add_edge(('a', 'b'))
sage: G.add_edge(('a', 'b'))
sage: G.add_edge(('a', 'b'))
sage: G.automorphism_group()                                                # needs sage.groups
Permutation Group with generators [('a','b')]

Digraphs:

sage: D = DiGraph( { 0:[1], 1:[2], 2:[3], 3:[4], 4:[0] } )
sage: D.automorphism_group()                                                # needs sage.groups
Permutation Group with generators [(0,1,2,3,4)]

Edge labeled graphs:

sage: G = Graph(sparse=True)
sage: G.add_edges( [(0,1,'a'),(1,2,'b'),(2,3,'c'),(3,4,'b'),(4,0,'a')] )
sage: G.automorphism_group(edge_labels=True)                                # needs sage.groups
Permutation Group with generators [(1,4)(2,3)]

sage: G.automorphism_group(edge_labels=True, algorithm="bliss") # optional - bliss
Permutation Group with generators [(1,4)(2,3)]

sage: G.automorphism_group(edge_labels=True, algorithm="sage")              # needs sage.groups
Permutation Group with generators [(1,4)(2,3)]
sage: G = Graph({0 : {1 : 7}})
sage: G.automorphism_group(edge_labels=True)                                # needs sage.groups
Permutation Group with generators [(0,1)]

sage: # needs sage.groups
sage: foo = Graph(sparse=True)
sage: bar = Graph(sparse=True)
sage: foo.add_edges([(0,1,1),(1,2,2), (2,3,3)])
sage: bar.add_edges([(0,1,1),(1,2,2), (2,3,3)])
sage: foo.automorphism_group(edge_labels=True)
Permutation Group with generators [()]
sage: foo.automorphism_group()
Permutation Group with generators [(0,3)(1,2)]
sage: bar.automorphism_group(edge_labels=True)
Permutation Group with generators [()]

You can also ask for just the order of the group:

sage: G = graphs.PetersenGraph()
sage: G.automorphism_group(return_group=False, order=True)                  # needs sage.groups
120

Or, just the orbits (note that each graph here is vertex transitive)

sage: # needs sage.groups
sage: G = graphs.PetersenGraph()
sage: G.automorphism_group(return_group=False, orbits=True, algorithm='sage')
[[0, 1, 2, 3, 4, 5, 6, 7, 8, 9]]
sage: orb = G.automorphism_group(partition=[[0],list(range(1,10))],
....:                            return_group=False, orbits=True, algorithm='sage')
sage: sorted([sorted(o) for o in orb], key=len)
[[0], [1, 4, 5], [2, 3, 6, 7, 8, 9]]
sage: C = graphs.CubeGraph(3)
sage: orb = C.automorphism_group(orbits=True, return_group=False, algorithm='sage')
sage: [sorted(o) for o in orb]
[['000', '001', '010', '011', '100', '101', '110', '111']]

One can also use the faster algorithm for computing the automorphism group of the graph - bliss:

sage: # optional - bliss
sage: G = graphs.HallJankoGraph()
sage: A1 = G.automorphism_group()                                           # needs sage.groups
sage: A2 = G.automorphism_group(algorithm='bliss')
sage: A1.is_isomorphic(A2)                                                  # needs sage.groups
True
average_degree()#

Return the average degree of the graph.

The average degree of a graph \(G=(V,E)\) is equal to \(\frac{2|E|}{|V|}\).

EXAMPLES:

The average degree of a regular graph is equal to the degree of any vertex:

sage: g = graphs.CompleteGraph(5)
sage: g.average_degree() == 4
True

The average degree of a tree is always strictly less than \(2\):

sage: tree = graphs.RandomTree(20)
sage: tree.average_degree() < 2
True

For any graph, it is equal to \(\frac{2|E|}{|V|}\):

sage: g = graphs.RandomGNP(20, .4)
sage: g.average_degree() == 2 * g.size() / g.order()
True
average_distance(by_weight=False, algorithm=None, weight_function=None, check_weight=True)#

Return the average distance between vertices of the graph.

Formally, for a graph \(G\) this value is equal to \(\frac 1 {n(n-1)} \sum_{u,v\in G} d(u,v)\) where \(d(u,v)\) denotes the distance between vertices \(u\) and \(v\) and \(n\) is the number of vertices in \(G\).

For more information on the input variables and more examples, we refer to wiener_index() and shortest_path_all_pairs(), which have very similar input variables.

INPUT:

  • by_weight – boolean (default: False); if True, the edges in the graph are weighted, otherwise all edges have weight 1

  • algorithm – string (default: None); one of the algorithms available for method wiener_index()

  • weight_function – function (default: None); a function that takes as input an edge (u, v, l) and outputs its weight. If not None, by_weight is automatically set to True. If None and by_weight is True, we use the edge label l, if l is not None, else 1 as a weight.

  • check_weight – boolean (default: True); if True, we check that the weight_function outputs a number for each edge

EXAMPLES:

From [GYLL1993]:

sage: g=graphs.PathGraph(10)
sage: w=lambda x: (x*(x*x -1)/6)/(x*(x-1)/2)
sage: g.average_distance()==w(10)
True

Average distance of a circuit:

sage: g = digraphs.Circuit(6)
sage: g.average_distance()
3
blocks_and_cut_vertices(G, algorithm='Tarjan_Boost', sort=False, key=None)#

Return the blocks and cut vertices of the graph.

In the case of a digraph, this computation is done on the underlying graph.

A cut vertex is one whose deletion increases the number of connected components. A block is a maximal induced subgraph which itself has no cut vertices. Two distinct blocks cannot overlap in more than a single cut vertex.

INPUT:

  • algorithm – string (default: "Tarjan_Boost"); the algorithm to use among:

    • "Tarjan_Boost" (default) – Tarjan’s algorithm (Boost implementation)

    • "Tarjan_Sage" – Tarjan’s algorithm (Sage implementation)

  • sort – boolean (default: False); whether to sort vertices inside the components and the list of cut vertices currently only available for ``”Tarjan_Sage”``

  • key – a function (default: None); a function that takes a vertex as its one argument and returns a value that can be used for comparisons in the sorting algorithm (we must have sort=True)

OUTPUT: (B, C), where B is a list of blocks - each is a list of vertices and the blocks are the corresponding induced subgraphs - and C is a list of cut vertices.

ALGORITHM:

We implement the algorithm proposed by Tarjan in [Tarjan72]. The original version is recursive. We emulate the recursion using a stack.

EXAMPLES:

We construct a trivial example of a graph with one cut vertex:

sage: from sage.graphs.connectivity import blocks_and_cut_vertices
sage: rings = graphs.CycleGraph(10)
sage: rings.merge_vertices([0, 5])
sage: blocks_and_cut_vertices(rings)
([[0, 1, 4, 2, 3], [0, 6, 9, 7, 8]], [0])
sage: rings.blocks_and_cut_vertices()
([[0, 1, 4, 2, 3], [0, 6, 9, 7, 8]], [0])
sage: B, C = blocks_and_cut_vertices(rings, algorithm="Tarjan_Sage", sort=True)
sage: B, C
([[0, 1, 2, 3, 4], [0, 6, 7, 8, 9]], [0])
sage: B2, C2 = blocks_and_cut_vertices(rings, algorithm="Tarjan_Sage", sort=False)
sage: Set(map(Set, B)) == Set(map(Set, B2)) and set(C) == set(C2)
True

The Petersen graph is biconnected, hence has no cut vertices:

sage: blocks_and_cut_vertices(graphs.PetersenGraph())
([[0, 1, 4, 5, 2, 6, 3, 7, 8, 9]], [])

Decomposing paths to pairs:

sage: g = graphs.PathGraph(4) + graphs.PathGraph(5)
sage: blocks_and_cut_vertices(g)
([[2, 3], [1, 2], [0, 1], [7, 8], [6, 7], [5, 6], [4, 5]], [1, 2, 5, 6, 7])

A disconnected graph:

sage: g = Graph({1: {2: 28, 3: 10}, 2: {1: 10, 3: 16}, 4: {}, 5: {6: 3, 7: 10, 8: 4}})
sage: blocks_and_cut_vertices(g)
([[1, 2, 3], [5, 6], [5, 7], [5, 8], [4]], [5])

A directed graph with Boost’s algorithm (github issue #25994):

sage: rings = graphs.CycleGraph(10)
sage: rings.merge_vertices([0, 5])
sage: rings = rings.to_directed()
sage: blocks_and_cut_vertices(rings, algorithm="Tarjan_Boost")
([[0, 1, 4, 2, 3], [0, 6, 9, 7, 8]], [0])
blocks_and_cuts_tree(G)#

Return the blocks-and-cuts tree of self.

This new graph has two different kinds of vertices, some representing the blocks (type B) and some other the cut vertices of the graph (type C).

There is an edge between a vertex \(u\) of type B and a vertex \(v\) of type C if the cut-vertex corresponding to \(v\) is in the block corresponding to \(u\).

The resulting graph is a tree, with the additional characteristic property that the distance between two leaves is even. When self is not connected, the resulting graph is a forest.

When self is biconnected, the tree is reduced to a single node of type \(B\).

We referred to [HarPri] and [Gallai] for blocks and cuts tree.

EXAMPLES:

sage: from sage.graphs.connectivity import blocks_and_cuts_tree
sage: T = blocks_and_cuts_tree(graphs.KrackhardtKiteGraph()); T
Graph on 5 vertices
sage: T.is_isomorphic(graphs.PathGraph(5))
True
sage: from sage.graphs.connectivity import blocks_and_cuts_tree
sage: T = graphs.KrackhardtKiteGraph().blocks_and_cuts_tree(); T
Graph on 5 vertices

The distance between two leaves is even:

sage: T = blocks_and_cuts_tree(graphs.RandomTree(40))
sage: T.is_tree()
True
sage: leaves = [v for v in T if T.degree(v) == 1]
sage: all(T.distance(u,v) % 2 == 0 for u in leaves for v in leaves)
True

The tree of a biconnected graph has a single vertex, of type \(B\):

sage: T = blocks_and_cuts_tree(graphs.PetersenGraph())
sage: T.vertices(sort=True)
[('B', (0, 1, 4, 5, 2, 6, 3, 7, 8, 9))]

Return an iterator over the vertices in a breadth-first ordering.

INPUT:

  • start – vertex or list of vertices from which to start the traversal

  • ignore_direction – boolean (default: False); only applies to directed graphs. If True, searches across edges in either direction.

  • distance – integer (default: None); the maximum distance from the start nodes to traverse. The start nodes are at distance zero from themselves.

  • neighbors – function (default: None); a function that inputs a vertex and return a list of vertices. For an undirected graph, neighbors is by default the neighbors() function. For a digraph, the neighbors function defaults to the neighbor_out_iterator() function of the graph.

  • report_distance – boolean (default: False); if True, reports pairs (vertex, distance) where distance is the distance from the start nodes. If False only the vertices are reported.

  • edges – boolean (default: False); whether to return the edges of the BFS tree in the order of visit or the vertices (default). Edges are directed in root to leaf orientation of the tree.

    Note that parameters edges and report_distance cannot be True simultaneously.

See also

EXAMPLES:

sage: G = Graph({0: [1], 1: [2], 2: [3], 3: [4], 4: [0]})
sage: list(G.breadth_first_search(0))
[0, 1, 4, 2, 3]

By default, the edge direction of a digraph is respected, but this can be overridden by the ignore_direction parameter:

sage: D = DiGraph({0: [1, 2, 3], 1: [4, 5], 2: [5], 3: [6], 5: [7], 6: [7], 7: [0]})
sage: list(D.breadth_first_search(0))
[0, 1, 2, 3, 4, 5, 6, 7]
sage: list(D.breadth_first_search(0, ignore_direction=True))
[0, 1, 2, 3, 7, 4, 5, 6]

You can specify a maximum distance in which to search. A distance of zero returns the start vertices:

sage: D = DiGraph({0: [1, 2, 3], 1: [4, 5], 2: [5], 3: [6], 5: [7], 6: [7], 7: [0]})
sage: list(D.breadth_first_search(0, distance=0))
[0]
sage: list(D.breadth_first_search(0, distance=1))
[0, 1, 2, 3]

Multiple starting vertices can be specified in a list:

sage: D = DiGraph({0: [1, 2, 3], 1: [4, 5], 2: [5], 3: [6], 5: [7], 6: [7], 7: [0]})
sage: list(D.breadth_first_search([0]))
[0, 1, 2, 3, 4, 5, 6, 7]
sage: list(D.breadth_first_search([0, 6]))
[0, 6, 1, 2, 3, 7, 4, 5]
sage: list(D.breadth_first_search([0, 6], distance=0))
[0, 6]
sage: list(D.breadth_first_search([0, 6], distance=1))
[0, 6, 1, 2, 3, 7]
sage: list(D.breadth_first_search(6, ignore_direction=True, distance=2))
[6, 3, 7, 0, 5]

More generally, you can specify a neighbors function. For example, you can traverse the graph backwards by setting neighbors to be the neighbors_in() function of the graph:

sage: D = DiGraph({0: [1, 2, 3], 1: [4, 5], 2: [5], 3: [6], 5: [7], 6: [7], 7: [0]})
sage: list(D.breadth_first_search(5, neighbors=D.neighbors_in, distance=2))
[5, 1, 2, 0]
sage: list(D.breadth_first_search(5, neighbors=D.neighbors_out, distance=2))
[5, 7, 0]
sage: list(D.breadth_first_search(5 ,neighbors=D.neighbors, distance=2))
[5, 1, 2, 7, 0, 4, 6]

It is possible (github issue #16470) using the keyword report_distance to get pairs (vertex, distance) encoding the distance from the starting vertices:

sage: G = graphs.PetersenGraph()
sage: list(G.breadth_first_search(0, report_distance=True))
[(0, 0), (1, 1), (4, 1), (5, 1), (2, 2), (6, 2), (3, 2), (9, 2),
(7, 2), (8, 2)]
sage: list(G.breadth_first_search(0, report_distance=False))
[0, 1, 4, 5, 2, 6, 3, 9, 7, 8]

sage: D = DiGraph({0: [1, 3], 1: [0, 2], 2: [0, 3], 3: [4]})
sage: D.show()                                                              # needs sage.plot
sage: list(D.breadth_first_search(4, neighbors=D.neighbor_in_iterator,
....:                             report_distance=True))
[(4, 0), (3, 1), (0, 2), (2, 2), (1, 3)]

sage: C = graphs.CycleGraph(4)
sage: list(C.breadth_first_search([0, 1], report_distance=True))
[(0, 0), (1, 0), (3, 1), (2, 1)]

You can get edges of the BFS tree instead of the vertices using the edges parameter:

sage: D = DiGraph({1:[2,3],2:[4],3:[4],4:[1],5:[2,6]})
sage: list(D.breadth_first_search(1, edges=True))
[(1, 2), (1, 3), (2, 4)]
canonical_label(partition=None, certificate=False, edge_labels=False, algorithm=None, return_graph=True)#

Return the canonical graph.

A canonical graph is the representative graph of an isomorphism class by some canonization function \(c\). If \(G\) and \(H\) are graphs, then \(G \cong c(G)\), and \(c(G) == c(H)\) if and only if \(G \cong H\).

See the Wikipedia article Graph_canonization for more information.

INPUT:

  • partition – if given, the canonical label with respect to this set partition will be computed. The default is the unit set partition.

  • certificate – boolean (default: False). When set to True, a dictionary mapping from the vertices of the (di)graph to its canonical label will also be returned.

  • edge_labels – boolean (default: False). When set to True, allows only permutations respecting edge labels.

  • algorithm – a string (default: None). The algorithm to use; currently available:

    • 'bliss': use the optional package bliss (http://www.tcs.tkk.fi/Software/bliss/index.html);

    • 'sage': always use Sage’s implementation.

    • None (default): use bliss when available and possible

      Note

      Make sure you always compare canonical forms obtained by the same algorithm.

  • return_graph – boolean (default: True). When set to False, returns the list of edges of the canonical graph instead of the canonical graph; only available when 'bliss' is explicitly set as algorithm.

EXAMPLES:

Canonization changes isomorphism to equality:

sage: g1 = graphs.GridGraph([2,3])
sage: g2 = Graph({1: [2, 4], 3: [2, 6], 5: [4, 2, 6]})
sage: g1 == g2
False
sage: g1.is_isomorphic(g2)
True
sage: g1.canonical_label() == g2.canonical_label()
True

We can get the relabeling used for canonization:

sage: g, c = g1.canonical_label(algorithm='sage', certificate=True)
sage: g
Grid Graph for [2, 3]: Graph on 6 vertices
sage: c
{(0, 0): 3, (0, 1): 4, (0, 2): 2, (1, 0): 0, (1, 1): 5, (1, 2): 1}

Multigraphs and directed graphs work too:

sage: G = Graph(multiedges=True,sparse=True)
sage: G.add_edge((0,1))
sage: G.add_edge((0,1))
sage: G.add_edge((0,1))
sage: G.canonical_label()
Multi-graph on 2 vertices
sage: Graph('A?').canonical_label()
Graph on 2 vertices

sage: P = graphs.PetersenGraph()
sage: DP = P.to_directed()
sage: DP.canonical_label(algorithm='sage').adjacency_matrix()               # needs sage.modules
[0 0 0 0 0 0 0 1 1 1]
[0 0 0 0 1 0 1 0 0 1]
[0 0 0 1 0 0 1 0 1 0]
[0 0 1 0 0 1 0 0 0 1]
[0 1 0 0 0 1 0 0 1 0]
[0 0 0 1 1 0 0 1 0 0]
[0 1 1 0 0 0 0 1 0 0]
[1 0 0 0 0 1 1 0 0 0]
[1 0 1 0 1 0 0 0 0 0]
[1 1 0 1 0 0 0 0 0 0]

Edge labeled graphs:

sage: G = Graph(sparse=True)
sage: G.add_edges( [(0,1,'a'),(1,2,'b'),(2,3,'c'),(3,4,'b'),(4,0,'a')] )
sage: G.canonical_label(edge_labels=True)
Graph on 5 vertices
sage: G.canonical_label(edge_labels=True, algorithm="bliss",        # optional - bliss
....:                   certificate=True)
(Graph on 5 vertices, {0: 4, 1: 3, 2: 1, 3: 0, 4: 2})

sage: G.canonical_label(edge_labels=True, algorithm="sage",
....:                   certificate=True)
(Graph on 5 vertices, {0: 4, 1: 3, 2: 0, 3: 1, 4: 2})

Another example where different canonization algorithms give different graphs:

sage: g = Graph({'a': ['b'], 'c': ['d']})
sage: g_sage = g.canonical_label(algorithm='sage')
sage: g_bliss = g.canonical_label(algorithm='bliss')                # optional - bliss
sage: g_sage.edges(sort=True, labels=False)
[(0, 3), (1, 2)]
sage: g_bliss.edges(sort=True, labels=False)                        # optional - bliss
[(0, 1), (2, 3)]
cartesian_product(other)#

Return the Cartesian product of self and other.

The Cartesian product of \(G\) and \(H\) is the graph \(L\) with vertex set \(V(L)\) equal to the Cartesian product of the vertices \(V(G)\) and \(V(H)\), and \(((u,v), (w,x))\) is an edge iff either - \((u, w)\) is an edge of self and \(v = x\), or - \((v, x)\) is an edge of other and \(u = w\).

See also

categorical_product(other)#

Return the tensor product of self and other.

The tensor product of \(G\) and \(H\) is the graph \(L\) with vertex set \(V(L)\) equal to the Cartesian product of the vertices \(V(G)\) and \(V(H)\), and \(((u,v), (w,x))\) is an edge iff - \((u, w)\) is an edge of self, and - \((v, x)\) is an edge of other.

The tensor product is also known as the categorical product and the Kronecker product (referring to the Kronecker matrix product). See the Wikipedia article Kronecker_product.

EXAMPLES:

sage: Z = graphs.CompleteGraph(2)
sage: C = graphs.CycleGraph(5)
sage: T = C.tensor_product(Z); T
Graph on 10 vertices
sage: T.size()
10
sage: T.plot()                      # long time                             # needs sage.plot
Graphics object consisting of 21 graphics primitives
sage: D = graphs.DodecahedralGraph()
sage: P = graphs.PetersenGraph()
sage: T = D.tensor_product(P); T
Graph on 200 vertices
sage: T.size()
900
sage: T.plot()                      # long time                             # needs sage.plot
Graphics object consisting of 1101 graphics primitives
centrality_betweenness(k=None, normalized=True, weight=None, endpoints=False, seed=None, exact=False, algorithm=None)#

Return the betweenness centrality.

The betweenness centrality of a vertex is the fraction of number of shortest paths that go through each vertex. The betweenness is normalized by default to be in range (0,1).

Measures of the centrality of a vertex within a graph determine the relative importance of that vertex to its graph. Vertices that occur on more shortest paths between other vertices have higher betweenness than vertices that occur on less.

INPUT:

  • normalized – boolean (default: True); if set to False, result is not normalized.

  • k – integer (default: None); if set to an integer, use k node samples to estimate betweenness. Higher values give better approximations. Not available when algorithm="Sage".

  • weight – string (default: None); if set to a string, use that attribute of the nodes as weight. weight = True is equivalent to weight = "weight". Not available when algorithm="Sage".

  • endpoints – boolean (default: False); if set to True it includes the endpoints in the shortest paths count. Not available when algorithm="Sage".

  • exact – boolean (default: False); whether to compute over rationals or on double C variables. Not available when algorithm="NetworkX".

  • algorithm – string (default: None); can be either "Sage" (see centrality), "NetworkX" or "None". In the latter case, Sage’s algorithm will be used whenever possible.

EXAMPLES:

sage: g = graphs.ChvatalGraph()
sage: g.centrality_betweenness() # abs tol 1e-10
{0: 0.06969696969696969, 1: 0.06969696969696969,
 2: 0.0606060606060606, 3: 0.0606060606060606,
 4: 0.06969696969696969, 5: 0.06969696969696969,
 6: 0.0606060606060606, 7: 0.0606060606060606,
 8: 0.0606060606060606, 9: 0.0606060606060606,
 10: 0.0606060606060606, 11: 0.0606060606060606}
sage: g.centrality_betweenness(normalized=False) # abs tol 1e-10
{0: 3.833333333333333, 1: 3.833333333333333, 2: 3.333333333333333,
 3: 3.333333333333333, 4: 3.833333333333333, 5: 3.833333333333333,
 6: 3.333333333333333, 7: 3.333333333333333, 8: 3.333333333333333,
 9: 3.333333333333333, 10: 3.333333333333333,
 11: 3.333333333333333}
sage: D = DiGraph({0:[1,2,3], 1:[2], 3:[0,1]})
sage: D.show(figsize=[2,2])                                                 # needs sage.plot
sage: D = D.to_undirected()
sage: D.show(figsize=[2,2])                                                 # needs sage.plot
sage: D.centrality_betweenness() # abs tol abs 1e-10
{0: 0.16666666666666666, 1: 0.16666666666666666, 2: 0.0, 3: 0.0}
centrality_closeness(vert=None, by_weight=False, algorithm=None, weight_function=None, check_weight=True)#

Return the closeness centrality of all vertices in vert.

In a (strongly) connected graph, the closeness centrality of a vertex \(v\) is equal to the inverse of the average distance between \(v\) and other vertices. If the graph is disconnected, the closeness centrality of \(v\) is multiplied by the fraction of reachable vertices in the graph: this way, central vertices should also reach several other vertices in the graph [OLJ2014]. In formulas,

\[c(v)=\frac{r(v)-1}{\sum_{w \in R(v)} d(v,w)}\frac{r(v)-1}{n-1}\]

where \(R(v)\) is the set of vertices reachable from \(v\), and \(r(v)\) is the cardinality of \(R(v)\).

‘Closeness centrality may be defined as the total graph-theoretic distance of a given vertex from all other vertices… Closeness is an inverse measure of centrality in that a larger value indicates a less central actor while a smaller value indicates a more central actor,’ [Bor1995].

For more information, see the Wikipedia article Centrality.

INPUT:

  • vert – the vertex or the list of vertices we want to analyze. If None (default), all vertices are considered.

  • by_weight – boolean (default: False); if True, the edges in the graph are weighted, and otherwise all edges have weight 1

  • algorithm – string (default: None); one of the following algorithms:

    • 'BFS': performs a BFS from each vertex that has to be analyzed. Does not work with edge weights.

    • 'NetworkX': the NetworkX algorithm (works only with positive weights).

    • 'Dijkstra_Boost': the Dijkstra algorithm, implemented in Boost (works only with positive weights).

    • 'Floyd-Warshall-Cython': the Cython implementation of the Floyd-Warshall algorithm. Works only if by_weight==False and all centralities are needed.

    • 'Floyd-Warshall-Python': the Python implementation of the Floyd-Warshall algorithm. Works only if all centralities are needed, but it can deal with weighted graphs, even with negative weights (but no negative cycle is allowed).

    • 'Johnson_Boost': the Johnson algorithm, implemented in Boost (works also with negative weights, if there is no negative cycle).

    • None (default): Sage chooses the best algorithm: 'BFS' if by_weight is False, 'Dijkstra_Boost' if all weights are positive, 'Johnson_Boost' otherwise.

  • weight_function – function (default: None); a function that takes as input an edge (u, v, l) and outputs its weight. If not None, by_weight is automatically set to True. If None and by_weight is True, we use the edge label l as a weight, if l is not None, else 1 as a weight.

  • check_weight – boolean (default: True); if True, we check that the weight_function outputs a number for each edge.

OUTPUT:

If vert is a vertex, the closeness centrality of that vertex. Otherwise, a dictionary associating to each vertex in vert its closeness centrality. If a vertex has (out)degree 0, its closeness centrality is not defined, and the vertex is not included in the output.

EXAMPLES:

Standard examples:

sage: (graphs.ChvatalGraph()).centrality_closeness()
{0: 0.61111111111111..., 1: 0.61111111111111...,
 2: 0.61111111111111..., 3: 0.61111111111111...,
 4: 0.61111111111111..., 5: 0.61111111111111...,
 6: 0.61111111111111..., 7: 0.61111111111111...,
 8: 0.61111111111111..., 9: 0.61111111111111...,
 10: 0.61111111111111..., 11: 0.61111111111111...}
sage: D = DiGraph({0:[1,2,3], 1:[2], 3:[0,1]})
sage: D.show(figsize=[2,2])                                                 # needs sage.plot
sage: D.centrality_closeness(vert=[0,1])
{0: 1.0, 1: 0.3333333333333333}
sage: D = D.to_undirected()
sage: D.show(figsize=[2,2])                                                 # needs sage.plot
sage: D.centrality_closeness()
{0: 1.0, 1: 1.0, 2: 0.75, 3: 0.75}

In a (strongly) connected (di)graph, the closeness centrality of \(v\) is inverse of the average distance between \(v\) and all other vertices:

sage: g = graphs.PathGraph(5)
sage: g.centrality_closeness(0)
0.4
sage: dist = g.shortest_path_lengths(0).values()
sage: float(len(dist)-1) / sum(dist)
0.4
sage: d = g.to_directed()
sage: d.centrality_closeness(0)
0.4
sage: dist = d.shortest_path_lengths(0).values()
sage: float(len(dist)-1) / sum(dist)
0.4

If a vertex has (out)degree 0, its closeness centrality is not defined:

sage: g = Graph(5)
sage: g.centrality_closeness()
{}
sage: print(g.centrality_closeness(0))
None

Weighted graphs:

sage: D = graphs.GridGraph([2,2])
sage: weight_function = lambda e:10
sage: D.centrality_closeness([(0,0),(0,1)])                          # tol abs 1e-12
{(0, 0): 0.75, (0, 1): 0.75}
sage: D.centrality_closeness((0,0), weight_function=weight_function) # tol abs 1e-12
0.075
characteristic_polynomial(var='x', laplacian=False)#

Return the characteristic polynomial of the adjacency matrix of the (di)graph.

Let \(G\) be a (simple) graph with adjacency matrix \(A\). Let \(I\) be the identity matrix of dimensions the same as \(A\). The characteristic polynomial of \(G\) is defined as the determinant \(\det(xI - A)\).

Note

characteristic_polynomial and charpoly are aliases and thus provide exactly the same method.

INPUT:

  • x – (default: 'x'); the variable of the characteristic polynomial

  • laplacian – boolean (default: False); if True, use the Laplacian matrix

EXAMPLES:

sage: P = graphs.PetersenGraph()
sage: P.characteristic_polynomial()                                         # needs sage.modules
x^10 - 15*x^8 + 75*x^6 - 24*x^5 - 165*x^4 + 120*x^3 + 120*x^2 - 160*x + 48
sage: P.charpoly()                                                          # needs sage.modules
x^10 - 15*x^8 + 75*x^6 - 24*x^5 - 165*x^4 + 120*x^3 + 120*x^2 - 160*x + 48
sage: P.characteristic_polynomial(laplacian=True)                           # needs sage.modules
x^10 - 30*x^9 + 390*x^8 - 2880*x^7 + 13305*x^6 -
39882*x^5 + 77640*x^4 - 94800*x^3 + 66000*x^2 - 20000*x
charpoly(var='x', laplacian=False)#

Return the characteristic polynomial of the adjacency matrix of the (di)graph.

Let \(G\) be a (simple) graph with adjacency matrix \(A\). Let \(I\) be the identity matrix of dimensions the same as \(A\). The characteristic polynomial of \(G\) is defined as the determinant \(\det(xI - A)\).

Note

characteristic_polynomial and charpoly are aliases and thus provide exactly the same method.

INPUT:

  • x – (default: 'x'); the variable of the characteristic polynomial

  • laplacian – boolean (default: False); if True, use the Laplacian matrix

EXAMPLES:

sage: P = graphs.PetersenGraph()
sage: P.characteristic_polynomial()                                         # needs sage.modules
x^10 - 15*x^8 + 75*x^6 - 24*x^5 - 165*x^4 + 120*x^3 + 120*x^2 - 160*x + 48
sage: P.charpoly()                                                          # needs sage.modules
x^10 - 15*x^8 + 75*x^6 - 24*x^5 - 165*x^4 + 120*x^3 + 120*x^2 - 160*x + 48
sage: P.characteristic_polynomial(laplacian=True)                           # needs sage.modules
x^10 - 30*x^9 + 390*x^8 - 2880*x^7 + 13305*x^6 -
39882*x^5 + 77640*x^4 - 94800*x^3 + 66000*x^2 - 20000*x
clear()#

Empties the graph of vertices and edges and removes name, associated objects, and position information.

EXAMPLES:

sage: G = graphs.CycleGraph(4)
sage: G.set_vertices({0:'vertex0'})
sage: print(G.order(), G.size())
4 4
sage: G.name()
'Cycle graph'
sage: G.get_vertex(0)
'vertex0'
sage: H = G.copy(sparse=True)
sage: H.clear()
sage: print(H.order(), H.size())
0 0
sage: H.name()
''
sage: H.get_vertex(0)
sage: H = G.copy(sparse=False)
sage: H.clear()
sage: print(H.order(), H.size())
0 0
sage: H.name()
''
sage: H.get_vertex(0)
cluster_transitivity()#

Return the transitivity (fraction of transitive triangles) of the graph.

Transitivity is the fraction of all existing triangles over all connected triples (triads), \(T = 3\times\frac{\text{triangles}}{\text{triads}}\).

See also section “Clustering” in chapter “Algorithms” of [HSS].

EXAMPLES:

sage: graphs.FruchtGraph().cluster_transitivity()                           # needs networkx
0.25
cluster_triangles(nbunch=None, implementation=None)#

Return the number of triangles for the set \(nbunch\) of vertices as a dictionary keyed by vertex.

See also section “Clustering” in chapter “Algorithms” of [HSS].

INPUT:

  • nbunch – a list of vertices (default: None); the vertices to inspect. If ``nbunch=None, returns data for all vertices in the graph.

  • implementation – string (default: None); one of 'sparse_copy', 'dense_copy', 'networkx' or None (default). In the latter case, the best algorithm available is used. Note that 'networkx' does not support directed graphs.

EXAMPLES:

sage: F = graphs.FruchtGraph()
sage: list(F.cluster_triangles().values())
[1, 1, 0, 1, 1, 1, 1, 1, 0, 1, 1, 0]
sage: F.cluster_triangles()
{0: 1, 1: 1, 2: 0, 3: 1, 4: 1, 5: 1, 6: 1, 7: 1, 8: 0, 9: 1, 10: 1, 11: 0}
sage: F.cluster_triangles(nbunch=[0, 1, 2])
{0: 1, 1: 1, 2: 0}
sage: G = graphs.RandomGNP(20, .3)
sage: d1 = G.cluster_triangles(implementation="networkx")                   # needs networkx
sage: d2 = G.cluster_triangles(implementation="dense_copy")
sage: d3 = G.cluster_triangles(implementation="sparse_copy")
sage: d1 == d2 and d1 == d3                                                 # needs networkx
True
clustering_average(implementation=None)#

Return the average clustering coefficient.

The clustering coefficient of a node \(i\) is the fraction of existing triangles containing node \(i\) over all possible triangles containing \(i\): \(c_i = T(i) / \binom {k_i} 2\) where \(T(i)\) is the number of existing triangles through \(i\), and \(k_i\) is the degree of vertex \(i\).

A coefficient for the whole graph is the average of the \(c_i\).

See also section “Clustering” in chapter “Algorithms” of [HSS].

INPUT:

  • implementation – string (default: None); one of 'boost', 'sparse_copy', 'dense_copy', 'networkx' or None (default). In the latter case, the best algorithm available is used. Note that only 'networkx' supports directed graphs.

EXAMPLES:

sage: (graphs.FruchtGraph()).clustering_average()
1/4
sage: (graphs.FruchtGraph()).clustering_average(implementation='networkx')  # needs networkx
0.25
clustering_coeff(nodes=None, weight=False, implementation=None)#

Return the clustering coefficient for each vertex in nodes as a dictionary keyed by vertex.

For an unweighted graph, the clustering coefficient of a node \(i\) is the fraction of existing triangles containing node \(i\) over all possible triangles containing \(i\): \(c_i = T(i) / \binom {k_i} 2\) where \(T(i)\) is the number of existing triangles through \(i\), and \(k_i\) is the degree of vertex \(i\).

For weighted graphs the clustering is defined as the geometric average of the subgraph edge weights, normalized by the maximum weight in the network.

The value of \(c_i\) is assigned \(0\) if \(k_i < 2\).

See also section “Clustering” in chapter “Algorithms” of [HSS].

INPUT:

  • nodes – an iterable container of vertices (default: None); the vertices to inspect. By default, returns data on all vertices in graph

  • weight – string or boolean (default: False); if it is a string it uses the indicated edge property as weight. weight = True is equivalent to weight = 'weight'

  • implementation – string (default: None); one of 'boost', 'sparse_copy', 'dense_copy', 'networkx' or None (default). In the latter case, the best algorithm available is used. Note that only 'networkx' supports directed or weighted graphs, and that 'sparse_copy' and 'dense_copy' do not support node different from None

EXAMPLES:

sage: graphs.FruchtGraph().clustering_coeff()
{0: 1/3, 1: 1/3, 2: 0, 3: 1/3, 4: 1/3, 5: 1/3,
 6: 1/3, 7: 1/3, 8: 0, 9: 1/3, 10: 1/3, 11: 0}

sage: (graphs.FruchtGraph()).clustering_coeff(weight=True)                  # needs networkx
{0: 0.3333333333333333, 1: 0.3333333333333333, 2: 0,
 3: 0.3333333333333333, 4: 0.3333333333333333,
 5: 0.3333333333333333, 6: 0.3333333333333333,
 7: 0.3333333333333333, 8: 0, 9: 0.3333333333333333,
 10: 0.3333333333333333, 11: 0}

sage: (graphs.FruchtGraph()).clustering_coeff(nodes=[0,1,2])
{0: 0.3333333333333333, 1: 0.3333333333333333, 2: 0.0}

sage: (graphs.FruchtGraph()).clustering_coeff(nodes=[0,1,2],                # needs networkx
....:                                         weight=True)
{0: 0.3333333333333333, 1: 0.3333333333333333, 2: 0}

sage: (graphs.GridGraph([5,5])).clustering_coeff(nodes=[(0,0),(0,1),(2,2)])
{(0, 0): 0.0, (0, 1): 0.0, (2, 2): 0.0}
coarsest_equitable_refinement(partition, sparse=True)#

Return the coarsest partition which is finer than the input partition, and equitable with respect to self.

A partition is equitable with respect to a graph if for every pair of cells \(C_1\), \(C_2\) of the partition, the number of edges from a vertex of \(C_1\) to \(C_2\) is the same, over all vertices in \(C_1\).

A partition \(P_1\) is finer than \(P_2\) (\(P_2\) is coarser than \(P_1\)) if every cell of \(P_1\) is a subset of a cell of \(P_2\).

INPUT:

  • partition – a list of lists

  • sparse – boolean (default: False); whether to use sparse or

    dense representation - for small graphs, use dense for speed

EXAMPLES:

sage: G = graphs.PetersenGraph()
sage: G.coarsest_equitable_refinement([[0],list(range(1,10))])
[[0], [2, 3, 6, 7, 8, 9], [1, 4, 5]]
sage: G = graphs.CubeGraph(3)
sage: verts = G.vertices(sort=True)
sage: Pi = [verts[:1], verts[1:]]
sage: Pi
[['000'], ['001', '010', '011', '100', '101', '110', '111']]
sage: [sorted(cell) for cell in G.coarsest_equitable_refinement(Pi)]
[['000'], ['011', '101', '110'], ['111'], ['001', '010', '100']]

Note that given an equitable partition, this function returns that partition:

sage: P = graphs.PetersenGraph()
sage: prt = [[0], [1, 4, 5], [2, 3, 6, 7, 8, 9]]
sage: P.coarsest_equitable_refinement(prt)
[[0], [1, 4, 5], [2, 3, 6, 7, 8, 9]]
sage: ss = (graphs.WheelGraph(6)).line_graph(labels=False)
sage: prt = [[(0, 1)], [(0, 2), (0, 3), (0, 4), (1, 2), (1, 4)], [(2, 3), (3, 4)]]
sage: ss.coarsest_equitable_refinement(prt)
Traceback (most recent call last):
...
TypeError: partition ([[(0, 1)], [(0, 2), (0, 3), (0, 4), (1, 2), (1, 4)],
[(2, 3), (3, 4)]]) is not valid for this graph: vertices are incorrect
sage: ss = (graphs.WheelGraph(5)).line_graph(labels=False)
sage: ss.coarsest_equitable_refinement(prt)
[[(0, 1)], [(1, 2), (1, 4)], [(0, 3)], [(0, 4), (0, 2)], [(2, 3), (3, 4)]]

ALGORITHM: Brendan D. McKay’s Master’s Thesis, University of Melbourne, 1976.

complement()#

Return the complement of the (di)graph.

The complement of a graph has the same vertices, but exactly those edges that are not in the original graph. This is not well defined for graphs with multiple edges.

EXAMPLES:

sage: P = graphs.PetersenGraph()
sage: P.plot()                      # long time                             # needs sage.plot
Graphics object consisting of 26 graphics primitives
sage: PC = P.complement()
sage: PC.plot()                     # long time                             # needs sage.plot
Graphics object consisting of 41 graphics primitives
sage: graphs.TetrahedralGraph().complement().size()
0
sage: graphs.CycleGraph(4).complement().edges(sort=True)
[(0, 2, None), (1, 3, None)]
sage: graphs.CycleGraph(4).complement()
complement(Cycle graph): Graph on 4 vertices
sage: G = Graph(multiedges=True, sparse=True)
sage: G.add_edges([(0, 1)] * 3)
sage: G.complement()
Traceback (most recent call last):
...
ValueError: This method is not known to work on graphs with
multiedges. Perhaps this method can be updated to handle them, but
in the meantime if you want to use it please disallow multiedges
using allow_multiple_edges().
connected_component_containing_vertex(G, vertex, sort=None, key=None)#

Return a list of the vertices connected to vertex.

INPUT:

  • G – the input graph

  • v – the vertex to search for

  • sort – boolean (default: None); if True, vertices inside the component are sorted according to the default ordering

    As of github issue #35889, this argument must be explicitly specified (unless a key is given); otherwise a warning is printed and sort=True is used. The default will eventually be changed to False.

  • key – a function (default: None); a function that takes a vertex as its one argument and returns a value that can be used for comparisons in the sorting algorithm (we must have sort=True)

EXAMPLES:

sage: from sage.graphs.connectivity import connected_component_containing_vertex
sage: G = Graph({0: [1, 3], 1: [2], 2: [3], 4: [5, 6], 5: [6]})
sage: connected_component_containing_vertex(G, 0, sort=True)
[0, 1, 2, 3]
sage: G.connected_component_containing_vertex(0, sort=True)
[0, 1, 2, 3]
sage: D = DiGraph({0: [1, 3], 1: [2], 2: [3], 4: [5, 6], 5: [6]})
sage: connected_component_containing_vertex(D, 0, sort=True)
[0, 1, 2, 3]
sage: connected_component_containing_vertex(D, 0, sort=True, key=lambda x: -x)
[3, 2, 1, 0]
connected_components(G, sort=None, key=None)#

Return the list of connected components.

This returns a list of lists of vertices, each list representing a connected component. The list is ordered from largest to smallest component.

INPUT:

  • G – the input graph

  • sort – boolean (default: None); if True, vertices inside each component are sorted according to the default ordering

    As of github issue #35889, this argument must be explicitly specified (unless a key is given); otherwise a warning is printed and sort=True is used. The default will eventually be changed to False.

  • key – a function (default: None); a function that takes a vertex as its one argument and returns a value that can be used for comparisons in the sorting algorithm (we must have sort=True)

EXAMPLES:

sage: from sage.graphs.connectivity import connected_components
sage: G = Graph({0: [1, 3], 1: [2], 2: [3], 4: [5, 6], 5: [6]})
sage: connected_components(G, sort=True)
[[0, 1, 2, 3], [4, 5, 6]]
sage: G.connected_components(sort=True)
[[0, 1, 2, 3], [4, 5, 6]]
sage: D = DiGraph({0: [1, 3], 1: [2], 2: [3], 4: [5, 6], 5: [6]})
sage: connected_components(D, sort=True)
[[0, 1, 2, 3], [4, 5, 6]]
sage: connected_components(D, sort=True, key=lambda x: -x)
[[3, 2, 1, 0], [6, 5, 4]]
connected_components_number(G)#

Return the number of connected components.

INPUT:

  • G – the input graph

EXAMPLES:

sage: from sage.graphs.connectivity import connected_components_number
sage: G = Graph({0: [1, 3], 1: [2], 2: [3], 4: [5, 6], 5: [6]})
sage: connected_components_number(G)
2
sage: G.connected_components_number()
2
sage: D = DiGraph({0: [1, 3], 1: [2], 2: [3], 4: [5, 6], 5: [6]})
sage: connected_components_number(D)
2
connected_components_sizes(G)#

Return the sizes of the connected components as a list.

The list is sorted from largest to lower values.

EXAMPLES:

sage: from sage.graphs.connectivity import connected_components_sizes
sage: for x in graphs(3):
....:     print(connected_components_sizes(x))
[1, 1, 1]
[2, 1]
[3]
[3]
sage: for x in graphs(3):
....:     print(x.connected_components_sizes())
[1, 1, 1]
[2, 1]
[3]
[3]
connected_components_subgraphs(G)#

Return a list of connected components as graph objects.

EXAMPLES:

sage: from sage.graphs.connectivity import connected_components_subgraphs
sage: G = Graph({0: [1, 3], 1: [2], 2: [3], 4: [5, 6], 5: [6]})
sage: L = connected_components_subgraphs(G)
sage: graphs_list.show_graphs(L)                                                # needs sage.plot
sage: D = DiGraph({0: [1, 3], 1: [2], 2: [3], 4: [5, 6], 5: [6]})
sage: L = connected_components_subgraphs(D)
sage: graphs_list.show_graphs(L)                                                # needs sage.plot
sage: L = D.connected_components_subgraphs()
sage: graphs_list.show_graphs(L)                                                # needs sage.plot
connected_subgraph_iterator(G, k=None, vertices_only=False, edges_only=False, labels=False, induced=True, exactly_k=False)#

Return an terator over the induced connected subgraphs of order at most \(k\).

This method implements a iterator over the induced connected subgraphs of the input (di)graph. An induced subgraph of a graph is another graph, formed from a subset of the vertices of the graph and all of the edges connecting pairs of vertices in that subset (Wikipedia article Induced_subgraph).

As for method sage.graphs.generic_graph.connected_components(), edge orientation is ignored. Hence, the directed graph with a single arc \(0 \to 1\) is considered connected.

INPUT:

  • G – a Graph or a DiGraph; loops and multiple edges are allowed

  • k – (optional) integer; maximum order of the connected subgraphs to report; by default, the method iterates over all connected subgraphs (equivalent to k == n)

  • vertices_only – boolean (default: False); whether to return (Di)Graph or list of vertices. This parameter is ignored when induced is True.

  • edges_only – boolean (default: False); whether to return (Di)Graph or list of edges. When vertices_only is True, this parameter is ignored.

  • labels – boolean (default: False); whether to return labelled edges or not. This parameter is used only when vertices_only is False and edges_only is True.

  • induced – boolean (default: True); whether to return induced connected sub(di)graph only or also non-induced sub(di)graphs. This parameter can be set to False for simple (di)graphs only.

  • exactly_k – boolean (default: False); True if we only return graphs of order \(k\), False if we return graphs of order at most \(k\).

EXAMPLES:

sage: G = DiGraph([(1, 2), (2, 3), (3, 4), (4, 2)])
sage: list(G.connected_subgraph_iterator())
[Subgraph of (): Digraph on 1 vertex,
 Subgraph of (): Digraph on 2 vertices,
 Subgraph of (): Digraph on 3 vertices,
 Subgraph of (): Digraph on 4 vertices,
 Subgraph of (): Digraph on 3 vertices,
 Subgraph of (): Digraph on 1 vertex,
 Subgraph of (): Digraph on 2 vertices,
 Subgraph of (): Digraph on 3 vertices,
 Subgraph of (): Digraph on 2 vertices,
 Subgraph of (): Digraph on 1 vertex,
 Subgraph of (): Digraph on 2 vertices,
 Subgraph of (): Digraph on 1 vertex]
sage: list(G.connected_subgraph_iterator(vertices_only=True))
[[1], [1, 2], [1, 2, 3], [1, 2, 3, 4], [1, 2, 4],
 [2], [2, 3], [2, 3, 4], [2, 4], [3], [3, 4], [4]]
sage: list(G.connected_subgraph_iterator(k=2))
[Subgraph of (): Digraph on 1 vertex,
 Subgraph of (): Digraph on 2 vertices,
 Subgraph of (): Digraph on 1 vertex,
 Subgraph of (): Digraph on 2 vertices,
 Subgraph of (): Digraph on 2 vertices,
 Subgraph of (): Digraph on 1 vertex,
 Subgraph of (): Digraph on 2 vertices,
 Subgraph of (): Digraph on 1 vertex]
sage: list(G.connected_subgraph_iterator(k=3, vertices_only=True, exactly_k=True))
[[1, 2, 3], [1, 2, 4], [2, 3, 4]]
sage: list(G.connected_subgraph_iterator(k=2, vertices_only=True))
[[1], [1, 2], [2], [2, 3], [2, 4], [3], [3, 4], [4]]

sage: G = DiGraph([(1, 2), (2, 1)])
sage: list(G.connected_subgraph_iterator())
[Subgraph of (): Digraph on 1 vertex,
 Subgraph of (): Digraph on 2 vertices,
 Subgraph of (): Digraph on 1 vertex]
sage: list(G.connected_subgraph_iterator(vertices_only=True))
[[1], [1, 2], [2]]

sage: G = graphs.CompleteGraph(3)
sage: len(list(G.connected_subgraph_iterator()))
7
sage: len(list(G.connected_subgraph_iterator(vertices_only=True)))
7
sage: len(list(G.connected_subgraph_iterator(edges_only=True)))
7
sage: len(list(G.connected_subgraph_iterator(induced=False)))
10

sage: G = DiGraph([(0, 1), (1, 0), (1, 2), (2, 1)])
sage: len(list(G.connected_subgraph_iterator()))
6
sage: len(list(G.connected_subgraph_iterator(vertices_only=True)))
6
sage: len(list(G.connected_subgraph_iterator(edges_only=True)))
6
sage: len(list(G.connected_subgraph_iterator(induced=False)))
18
contract_edge(u, v=None, label=None)#

Contract an edge from u to v.

This method returns silently if the edge does not exist.

INPUT: The following forms are all accepted:

  • G.contract_edge( 1, 2 )

  • G.contract_edge( (1, 2) )

  • G.contract_edge( [ (1, 2) ] )

  • G.contract_edge( 1, 2, ‘label’ )

  • G.contract_edge( (1, 2, ‘label’) )

  • G.contract_edge( [ (1, 2, ‘label’) ] )

EXAMPLES:

sage: G = graphs.CompleteGraph(4)
sage: G.contract_edge((0, 1)); G.edges(sort=True)
[(0, 2, None), (0, 3, None), (2, 3, None)]
sage: G = graphs.CompleteGraph(4)
sage: G.allow_loops(True); G.allow_multiple_edges(True)
sage: G.contract_edge((0, 1)); G.edges(sort=True)
[(0, 2, None), (0, 2, None), (0, 3, None), (0, 3, None), (2, 3, None)]
sage: G.contract_edge((0, 2)); G.edges(sort=True)
[(0, 0, None), (0, 3, None), (0, 3, None), (0, 3, None)]
sage: G = graphs.CompleteGraph(4).to_directed()
sage: G.allow_loops(True)
sage: G.contract_edge(0, 1); G.edges(sort=True)
[(0, 0, None),
 (0, 2, None),
 (0, 3, None),
 (2, 0, None),
 (2, 3, None),
 (3, 0, None),
 (3, 2, None)]
contract_edges(edges)#

Contract edges from an iterable container.

If \(e\) is an edge that is not contracted but the vertices of \(e\) are merged by contraction of other edges, then \(e\) will become a loop.

INPUT:

  • edges – a list containing 2-tuples or 3-tuples that represent edges

EXAMPLES:

sage: G = graphs.CompleteGraph(4)
sage: G.allow_loops(True); G.allow_multiple_edges(True)
sage: G.contract_edges([(0, 1), (1, 2), (0, 2)]); G.edges(sort=True)
[(0, 3, None), (0, 3, None), (0, 3, None)]
sage: G.contract_edges([(1, 3), (2, 3)]); G.edges(sort=True)
[(0, 3, None), (0, 3, None), (0, 3, None)]
sage: G = graphs.CompleteGraph(4)
sage: G.allow_loops(True); G.allow_multiple_edges(True)
sage: G.contract_edges([(0, 1), (1, 2), (0, 2), (1, 3), (2, 3)]); G.edges(sort=True)
[(0, 0, None)]
sage: D = digraphs.Complete(4)
sage: D.allow_loops(True); D.allow_multiple_edges(True)
sage: D.contract_edges([(0, 1), (1, 0), (0, 2)]); D.edges(sort=True)
[(0, 0, None),
 (0, 0, None),
 (0, 0, None),
 (0, 3, None),
 (0, 3, None),
 (0, 3, None),
 (3, 0, None),
 (3, 0, None),
 (3, 0, None)]
copy(weighted=None, data_structure=None, sparse=None, immutable=None, hash_labels=None)#

Change the graph implementation

INPUT:

  • weighted – boolean (default: None); weightedness for the copy. Might change the equality class if not None.

  • sparse – boolean (default: None); sparse=True is an alias for data_structure="sparse", and sparse=False is an alias for data_structure="dense". Only used when data_structure=None.

  • data_structure – string (default: None); one of "sparse", "static_sparse", or "dense". See the documentation of Graph or DiGraph.

  • immutable – boolean (default: None); whether to create a mutable/immutable copy. Only used when data_structure=None.

    • immutable=None (default) means that the graph and its copy will behave the same way.

    • immutable=True is a shortcut for data_structure='static_sparse'

    • immutable=False means that the created graph is mutable. When used to copy an immutable graph, the data structure used is "sparse" unless anything else is specified.

  • hash_labels – boolean (default: None); whether to include edge labels during hashing of the copy. This parameter defaults to True if the graph is weighted. This parameter is ignored when parameter immutable is not True. Beware that trying to hash unhashable labels will raise an error.

Note

If the graph uses StaticSparseBackend and the _immutable flag, then self is returned rather than a copy (unless one of the optional arguments is used).

OUTPUT:

A Graph object.

Warning

Please use this method only if you need to copy but change the underlying data structure or weightedness. Otherwise simply do copy(g) instead of g.copy().

Warning

If weighted is passed and is not the weightedness of the original, then the copy will not equal the original.

EXAMPLES:

sage: g = Graph({0: [0, 1, 1, 2]}, loops=True, multiedges=True, sparse=True)
sage: g == copy(g)
True
sage: g = DiGraph({0: [0, 1, 1, 2], 1: [0, 1]}, loops=True, multiedges=True, sparse=True)
sage: g == copy(g)
True

Note that vertex associations are also kept:

sage: d = {0: graphs.DodecahedralGraph(), 1: graphs.FlowerSnark(), 2: graphs.MoebiusKantorGraph(), 3: graphs.PetersenGraph()}
sage: T = graphs.TetrahedralGraph()
sage: T.set_vertices(d)
sage: T2 = copy(T)
sage: T2.get_vertex(0)
Dodecahedron: Graph on 20 vertices

Notice that the copy is at least as deep as the objects:

sage: T2.get_vertex(0) is T.get_vertex(0)
False

Examples of the keywords in use:

sage: G = graphs.CompleteGraph(9)
sage: H = G.copy()
sage: H == G; H is G
True
False
sage: G1 = G.copy(sparse=True)
sage: G1 == G
True
sage: G1 is G
False
sage: G2 = copy(G)
sage: G2 is G
False

Argument weighted affects the equality class:

sage: G = graphs.CompleteGraph(5)
sage: H1 = G.copy(weighted=False)
sage: H2 = G.copy(weighted=True)
sage: [G.weighted(), H1.weighted(), H2.weighted()]
[False, False, True]
sage: [G == H1, G == H2, H1 == H2]
[True, False, False]
sage: G.weighted(True)
sage: [G == H1, G == H2, H1 == H2]
[False, True, False]
crossing_number()#

Return the crossing number of the graph.

The crossing number of a graph is the minimum number of edge crossings needed to draw the graph on a plane. It can be seen as a measure of non-planarity; a planar graph has crossing number zero.

See the Wikipedia article Crossing_number for more information.

EXAMPLES:

sage: P = graphs.PetersenGraph()
sage: P.crossing_number()
2

ALGORITHM:

This is slow brute force implementation: for every \(k\) pairs of edges try adding a new vertex for a crossing point for them. If the result is not planar in any of those, try \(k+1\) pairs.

Computing the crossing number is NP-hard problem.

cycle_basis(output='vertex')#

Return a list of cycles which form a basis of the cycle space of self.

A basis of cycles of a graph is a minimal collection of cycles (considered as sets of edges) such that the edge set of any cycle in the graph can be written as a \(Z/2Z\) sum of the cycles in the basis.

See the Wikipedia article Cycle_basis for more information.

INPUT:

  • output – string (default: 'vertex'); whether every cycle is given as a list of vertices (output == 'vertex') or a list of edges (output == 'edge')

OUTPUT:

A list of lists, each of them representing the vertices (or the edges) of a cycle in a basis.

ALGORITHM:

Uses the NetworkX library for graphs without multiple edges.

Otherwise, by the standard algorithm using a spanning tree.

EXAMPLES:

A cycle basis in Petersen’s Graph

sage: g = graphs.PetersenGraph()
sage: g.cycle_basis()                                                       # needs networkx, random (changes in networkx 3.2)
[[1, 6, 8, 5, 0], [4, 9, 6, 8, 5, 0], [7, 9, 6, 8, 5],
 [4, 3, 8, 5, 0], [1, 2, 3, 8, 5, 0], [7, 2, 3, 8, 5]]

One can also get the result as a list of lists of edges:

sage: g.cycle_basis(output='edge')                                          # needs networkx, random (changes in networkx 3.2)
[[(1, 6, None), (6, 8, None), (8, 5, None), (5, 0, None),
 (0, 1, None)], [(4, 9, None), (9, 6, None), (6, 8, None),
 (8, 5, None), (5, 0, None), (0, 4, None)], [(7, 9, None),
 (9, 6, None), (6, 8, None), (8, 5, None), (5, 7, None)],
 [(4, 3, None), (3, 8, None), (8, 5, None), (5, 0, None),
 (0, 4, None)], [(1, 2, None), (2, 3, None), (3, 8, None),
 (8, 5, None), (5, 0, None), (0, 1, None)], [(7, 2, None),
 (2, 3, None), (3, 8, None), (8, 5, None), (5, 7, None)]]

Checking the given cycles are algebraically free:

sage: g = graphs.RandomGNP(30, .4)                                          # needs networkx
sage: basis = g.cycle_basis()                                               # needs networkx

Building the space of (directed) edges over \(Z/2Z\). On the way, building a dictionary associating a unique vector to each undirected edge:

sage: m = g.size()
sage: edge_space = VectorSpace(FiniteField(2), m)                           # needs sage.modules sage.rings.finite_rings
sage: edge_vector = dict(zip(g.edges(labels=False, sort=False),             # needs sage.modules sage.rings.finite_rings
....:                        edge_space.basis()))
sage: for (u, v), vec in list(edge_vector.items()):                         # needs sage.modules sage.rings.finite_rings
....:    edge_vector[(v, u)] = vec

Defining a lambda function associating a vector to the vertices of a cycle:

sage: vertices_to_edges = lambda x: zip(x, x[1:] + [x[0]])
sage: cycle_to_vector = lambda x: sum(edge_vector[e] for e in vertices_to_edges(x))

Finally checking the cycles are a free set:

sage: basis_as_vectors = [cycle_to_vector(_) for _ in basis]                # needs networkx sage.modules sage.rings.finite_rings
sage: edge_space.span(basis_as_vectors).rank() == len(basis)                # needs networkx sage.modules sage.rings.finite_rings
True

For undirected graphs with multiple edges:

sage: G = Graph([(0, 2, 'a'), (0, 2, 'b'), (0, 1, 'c'), (1, 2, 'd')],
....:           multiedges=True)
sage: G.cycle_basis()
[[2, 0], [2, 0, 1]]
sage: G.cycle_basis(output='edge')
[[(0, 2, 'b'), (2, 0, 'a')], [(1, 2, 'd'), (2, 0, 'a'), (0, 1, 'c')]]
sage: H = Graph([(1, 2), (2, 3), (2, 3), (3, 4), (1, 4),
....:            (1, 4), (4, 5), (5, 6), (4, 6), (6, 7)], multiedges=True)
sage: H.cycle_basis()
[[4, 1], [3, 2], [4, 1, 2, 3], [6, 4, 5]]

Disconnected graph:

sage: G.add_cycle(["Hey", "Wuuhuu", "Really ?"])
sage: [sorted(c) for c in G.cycle_basis()]                                  # needs networkx
[['Hey', 'Really ?', 'Wuuhuu'], [0, 2], [0, 1, 2]]
sage: [sorted(c) for c in G.cycle_basis(output='edge')]                     # needs networkx
[[('Hey', 'Really ?', None),
  ('Really ?', 'Wuuhuu', None),
  ('Wuuhuu', 'Hey', None)],
 [(0, 2, 'a'), (2, 0, 'b')],
 [(0, 1, 'c'), (1, 2, 'd'), (2, 0, 'b')]]

Graph that allows multiple edges but does not contain any:

sage: G = graphs.CycleGraph(3)
sage: G.allow_multiple_edges(True)
sage: G.cycle_basis()
[[2, 0, 1]]

Not yet implemented for directed graphs:

sage: G = DiGraph([(0, 2, 'a'), (0, 1, 'c'), (1, 2, 'd')])
sage: G.cycle_basis()
Traceback (most recent call last):
...
NotImplementedError: not implemented for directed graphs
degree(vertices=None, labels=False)#

Return the degree (in + out for digraphs) of a vertex or of vertices.

INPUT:

  • vertices – a vertex or an iterable container of vertices (default: None); if vertices is a single vertex, returns the number of neighbors of that vertex. If vertices is an iterable container of vertices, returns a list of degrees. If vertices is None, same as listing all vertices.

  • labels – boolean (default: False); when True, return a dictionary mapping each vertex in vertices to its degree. Otherwise, return the degree of a single vertex or a list of the degrees of each vertex in vertices

OUTPUT:

  • When vertices is a single vertex and labels is False, returns the degree of that vertex as an integer

  • When vertices is an iterable container of vertices (or None) and labels is False, returns a list of integers. The \(i\)-th value is the degree of the \(i\)-th vertex in the list vertices. When vertices is None, the \(i\)-th value is the degree of \(i\)-th vertex in the ordering list(self), which might be different from the ordering of the vertices given by g.vertices(sort=True).

  • When labels is True, returns a dictionary mapping each vertex in vertices to its degree

EXAMPLES:

sage: P = graphs.PetersenGraph()
sage: P.degree(5)
3
sage: K = graphs.CompleteGraph(9)
sage: K.degree()
[8, 8, 8, 8, 8, 8, 8, 8, 8]
sage: D = DiGraph({0: [1, 2, 3], 1: [0, 2], 2: [3], 3: [4], 4: [0,5], 5: [1]})
sage: D.degree(vertices=[0, 1, 2], labels=True)
{0: 5, 1: 4, 2: 3}
sage: D.degree()
[5, 4, 3, 3, 3, 2]

When vertices=None and labels=False, the \(i\)-th value of the returned list is the degree of the \(i\)-th vertex in the list list(self):

sage: # needs sage.combinat
sage: D = digraphs.DeBruijn(4, 2)
sage: D.delete_vertex('20')
sage: print(D.degree())
[7, 7, 6, 7, 8, 8, 7, 8, 8, 7, 8, 8, 8, 7, 8]
sage: print(D.degree(vertices=list(D)))
[7, 7, 6, 7, 8, 8, 7, 8, 8, 7, 8, 8, 8, 7, 8]
sage: print(D.degree(vertices=D.vertices(sort=False)))
[7, 7, 6, 7, 8, 8, 7, 8, 8, 7, 8, 8, 8, 7, 8]
degree_histogram()#

Return a list, whose \(i\)-th entry is the frequency of degree \(i\).

EXAMPLES:

sage: G = graphs.Grid2dGraph(9, 12)
sage: G.degree_histogram()
[0, 0, 4, 34, 70]
sage: G = graphs.Grid2dGraph(9, 12).to_directed()
sage: G.degree_histogram()
[0, 0, 0, 0, 4, 0, 34, 0, 70]
degree_iterator(vertices=None, labels=False)#

Return an iterator over the degrees of the (di)graph.

In the case of a digraph, the degree is defined as the sum of the in-degree and the out-degree, i.e. the total number of edges incident to a given vertex.

INPUT:

  • vertices – a vertex or an iterable container of vertices (default: None); if vertices is a single vertex, the iterator will yield the number of neighbors of that vertex. If vertices is an iterable container of vertices, return an iterator over the degrees of these vertices. If vertices is None, same as listing all vertices.

  • labels – boolean (default: False); whether to return an iterator over degrees (labels=False), or over tuples (vertex, degree)

Note

The returned iterator yields values in order specified by list(vertices). When vertices is None, it yields values in the same order as list(self), which might be different from the ordering of the vertices given by g.vertices(sort=True).

EXAMPLES:

sage: G = graphs.Grid2dGraph(3, 4)
sage: for i in G.degree_iterator():
....:     print(i)
2
3
3
...
3
2
sage: for i in G.degree_iterator(labels=True):
....:     print(i)
((0, 0), 2)
((0, 1), 3)
((0, 2), 3)
...
((2, 2), 3)
((2, 3), 2)
sage: D = graphs.Grid2dGraph(2,4).to_directed()
sage: for i in D.degree_iterator():
....:     print(i)
4
6
...
6
4
sage: for i in D.degree_iterator(labels=True):
....:     print(i)
((0, 0), 4)
((0, 1), 6)
...
((1, 2), 6)
((1, 3), 4)

When vertices=None yields values in the order of list(D):

sage: V = list(D)
sage: D = digraphs.DeBruijn(4, 2)                                           # needs sage.combinat
sage: D.delete_vertex('20')                                                 # needs sage.combinat
sage: print(list(D.degree_iterator()))                                      # needs sage.combinat
[7, 7, 6, 7, 8, 8, 7, 8, 8, 7, 8, 8, 8, 7, 8]
sage: print([D.degree(v) for v in D])                                       # needs sage.combinat
[7, 7, 6, 7, 8, 8, 7, 8, 8, 7, 8, 8, 8, 7, 8]
degree_sequence()#

Return the degree sequence of this (di)graph.

EXAMPLES:

The degree sequence of an undirected graph:

sage: g = Graph({1: [2, 5], 2: [1, 5, 3, 4], 3: [2, 5], 4: [3], 5: [2, 3]})
sage: g.degree_sequence()
[4, 3, 3, 2, 2]

The degree sequence of a digraph:

sage: g = DiGraph({1: [2, 5, 6], 2: [3, 6], 3: [4, 6], 4: [6], 5: [4, 6]})
sage: g.degree_sequence()
[5, 3, 3, 3, 3, 3]

Degree sequences of some common graphs:

sage: graphs.PetersenGraph().degree_sequence()
[3, 3, 3, 3, 3, 3, 3, 3, 3, 3]
sage: graphs.HouseGraph().degree_sequence()
[3, 3, 2, 2, 2]
sage: graphs.FlowerSnark().degree_sequence()
[3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3]
degree_to_cell(vertex, cell)#

Returns the number of edges from vertex to an edge in cell. In the case of a digraph, returns a tuple (in_degree, out_degree).

EXAMPLES:

sage: G = graphs.CubeGraph(3)
sage: cell = G.vertices(sort=True)[:3]
sage: G.degree_to_cell('011', cell)
2
sage: G.degree_to_cell('111', cell)
0
sage: D = DiGraph({ 0:[1,2,3], 1:[3,4], 3:[4,5]})
sage: cell = [0,1,2]
sage: D.degree_to_cell(5, cell)
(0, 0)
sage: D.degree_to_cell(3, cell)
(2, 0)
sage: D.degree_to_cell(0, cell)
(0, 2)
delete_edge(u, v=None, label=None)#

Delete the edge from u to v.

This method returns silently if vertices or edge does not exist.

INPUT: The following forms are all accepted:

  • G.delete_edge( 1, 2 )

  • G.delete_edge( (1, 2) )

  • G.delete_edges( [ (1, 2) ] )

  • G.delete_edge( 1, 2, ‘label’ )

  • G.delete_edge( (1, 2, ‘label’) )

  • G.delete_edges( [ (1, 2, ‘label’) ] )

EXAMPLES:

sage: G = graphs.CompleteGraph(9)
sage: G.size()
36
sage: G.delete_edge( 1, 2 )
sage: G.delete_edge( (3, 4) )
sage: G.delete_edges( [ (5, 6), (7, 8) ] )
sage: G.size()
32
sage: G.delete_edge( 2, 3, 'label' )
sage: G.delete_edge( (4, 5, 'label') )
sage: G.delete_edges( [ (6, 7, 'label') ] )
sage: G.size()
32
sage: G.has_edge( (4, 5) ) # correct!
True
sage: G.has_edge( (4, 5, 'label') ) # correct!
False
sage: C = digraphs.Complete(9)
sage: C.size()
72
sage: C.delete_edge( 1, 2 )
sage: C.delete_edge( (3, 4) )
sage: C.delete_edges( [ (5, 6), (7, 8) ] )
sage: C.size()
68
sage: C.delete_edge( 2, 3, 'label' )
sage: C.delete_edge( (4, 5, 'label') )
sage: C.delete_edges( [ (6, 7, 'label') ] )
sage: C.size() # correct!
68
sage: C.has_edge( (4, 5) ) # correct!
True
sage: C.has_edge( (4, 5, 'label') ) # correct!
False
delete_edges(edges)#

Delete edges from an iterable container.

EXAMPLES:

sage: K12 = graphs.CompleteGraph(12)
sage: K4 = graphs.CompleteGraph(4)
sage: K12.size()
66
sage: K12.delete_edges(K4.edge_iterator())
sage: K12.size()
60
sage: K12 = digraphs.Complete(12)
sage: K4 = digraphs.Complete(4)
sage: K12.size()
132
sage: K12.delete_edges(K4.edge_iterator())
sage: K12.size()
120
delete_multiedge(u, v)#

Delete all edges from u to v.

EXAMPLES:

sage: G = Graph(multiedges=True, sparse=True)
sage: G.add_edges([(0, 1), (0, 1), (0, 1), (1, 2), (2, 3)])
sage: G.edges(sort=True)
[(0, 1, None), (0, 1, None), (0, 1, None), (1, 2, None), (2, 3, None)]
sage: G.delete_multiedge(0, 1)
sage: G.edges(sort=True)
[(1, 2, None), (2, 3, None)]
sage: D = DiGraph(multiedges=True, sparse=True)
sage: D.add_edges([(0, 1, 1), (0, 1, 2), (0, 1, 3), (1, 0, None), (1, 2, None), (2, 3, None)])
sage: D.edges(sort=True)
[(0, 1, 1), (0, 1, 2), (0, 1, 3), (1, 0, None), (1, 2, None), (2, 3, None)]
sage: D.delete_multiedge(0, 1)
sage: D.edges(sort=True)
[(1, 0, None), (1, 2, None), (2, 3, None)]
delete_vertex(vertex, in_order=False)#

Delete vertex, removing all incident edges.

Deleting a non-existent vertex will raise an exception.

INPUT:

  • in_order – boolean (default: False); if True, this deletes the \(i\)-th vertex in the sorted list of vertices, i.e. G.vertices(sort=True)[i]

EXAMPLES:

sage: G = Graph(graphs.WheelGraph(9))
sage: G.delete_vertex(0)
sage: G.show()                                                              # needs sage.plot
sage: D = DiGraph({0: [1, 2, 3, 4, 5], 1: [2], 2: [3], 3: [4], 4: [5], 5: [1]})
sage: D.delete_vertex(0); D
Digraph on 5 vertices
sage: D.vertices(sort=True)
[1, 2, 3, 4, 5]
sage: D.delete_vertex(0)
Traceback (most recent call last):
...
ValueError: vertex (0) not in the graph
sage: G = graphs.CompleteGraph(4).line_graph(labels=False)
sage: G.vertices(sort=True)
[(0, 1), (0, 2), (0, 3), (1, 2), (1, 3), (2, 3)]
sage: G.delete_vertex(0, in_order=True)
sage: G.vertices(sort=True)
[(0, 2), (0, 3), (1, 2), (1, 3), (2, 3)]
sage: G = graphs.PathGraph(5)
sage: G.set_vertices({0: 'no delete', 1: 'delete'})
sage: G.delete_vertex(1)
sage: G.get_vertices()
{0: 'no delete', 2: None, 3: None, 4: None}
sage: G.get_pos()
{0: (0, 0), 2: (2, 0), 3: (3, 0), 4: (4, 0)}
delete_vertices(vertices)#

Delete vertices from the (di)graph taken from an iterable container of vertices.

Deleting a non-existent vertex will raise an exception, in which case none of the vertices in vertices is deleted.

EXAMPLES:

sage: D = DiGraph({0: [1, 2, 3, 4, 5], 1: [2], 2: [3], 3: [4], 4: [5], 5: [1]})
sage: D.delete_vertices([1, 2, 3, 4, 5]); D
Digraph on 1 vertex
sage: D.vertices(sort=False)
[0]
sage: D.delete_vertices([1])
Traceback (most recent call last):
...
ValueError: vertex (1) not in the graph
density()#

Return the density of the (di)graph.

The density of a (di)graph is defined as the number of edges divided by number of possible edges.

In the case of a multigraph, raises an error, since there is an infinite number of possible edges.

EXAMPLES:

sage: d = {0: [1,4,5], 1: [2,6], 2: [3,7], 3: [4,8], 4: [9], 5: [7, 8], 6: [8,9], 7: [9]}
sage: G = Graph(d); G.density()
1/3
sage: G = Graph({0: [1, 2], 1: [0]}); G.density()
2/3
sage: G = DiGraph({0: [1, 2], 1: [0]}); G.density()
1/2

Note that there are more possible edges on a looped graph:

sage: G.allow_loops(True)
sage: G.density()
1/3

Return an iterator over the vertices in a depth-first ordering.

INPUT:

  • start – vertex or list of vertices from which to start the traversal

  • ignore_direction – boolean (default: False); only applies to directed graphs. If True, searches across edges in either direction.

  • neighbors – function (default: None); a function that inputs a vertex and return a list of vertices. For an undirected graph, neighbors is by default the neighbors() function. For a digraph, the neighbors function defaults to the neighbor_out_iterator() function of the graph.

  • edges – boolean (default: False); whether to return the edges of the DFS tree in the order of visit or the vertices (default). Edges are directed in root to leaf orientation of the tree.

See also

EXAMPLES:

sage: G = Graph({0: [1], 1: [2], 2: [3], 3: [4], 4: [0]})
sage: list(G.depth_first_search(0))
[0, 4, 3, 2, 1]

By default, the edge direction of a digraph is respected, but this can be overridden by the ignore_direction parameter:

sage: D = DiGraph({0: [1, 2, 3], 1: [4, 5], 2: [5], 3: [6], 5: [7], 6: [7], 7: [0]})
sage: list(D.depth_first_search(0))
[0, 3, 6, 7, 2, 5, 1, 4]
sage: list(D.depth_first_search(0, ignore_direction=True))
[0, 7, 6, 3, 5, 2, 1, 4]

Multiple starting vertices can be specified in a list:

sage: D = DiGraph({0: [1, 2, 3], 1: [4, 5], 2: [5], 3: [6], 5: [7], 6: [7], 7: [0]})
sage: list(D.depth_first_search([0]))
[0, 3, 6, 7, 2, 5, 1, 4]
sage: list(D.depth_first_search([0, 6]))
[0, 3, 6, 7, 2, 5, 1, 4]

More generally, you can specify a neighbors function. For example, you can traverse the graph backwards by setting neighbors to be the neighbors_in() function of the graph:

sage: D = digraphs.Path(10)
sage: D.add_path([22, 23, 24, 5])
sage: D.add_path([5, 33, 34, 35])
sage: list(D.depth_first_search(5, neighbors=D.neighbors_in))
[5, 4, 3, 2, 1, 0, 24, 23, 22]
sage: list(D.breadth_first_search(5, neighbors=D.neighbors_in))
[5, 24, 4, 23, 3, 22, 2, 1, 0]
sage: list(D.depth_first_search(5, neighbors=D.neighbors_out))
[5, 6, 7, 8, 9, 33, 34, 35]
sage: list(D.breadth_first_search(5, neighbors=D.neighbors_out))
[5, 33, 6, 34, 7, 35, 8, 9]

You can get edges of the DFS tree instead of the vertices using the edges parameter:

sage: D = digraphs.Path(5)
sage: list(D.depth_first_search(2, edges=True))
[(2, 3), (3, 4)]
sage: list(D.depth_first_search(2, edges=True, ignore_direction=True))
[(2, 3), (3, 4), (2, 1), (1, 0)]
disjoint_routed_paths(pairs, solver, verbose=None, integrality_tolerance=0)#

Return a set of disjoint routed paths.

Given a set of pairs \((s_i,t_i)\), a set of disjoint routed paths is a set of \(s_i-t_i\) paths which can intersect at their endpoints and are vertex-disjoint otherwise.

INPUT:

  • pairs – list of pairs of vertices

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

EXAMPLES:

Given a grid, finding two vertex-disjoint paths, the first one from the top-left corner to the bottom-left corner, and the second from the top-right corner to the bottom-right corner is easy:

sage: g = graphs.Grid2dGraph(5, 5)
sage: p1,p2 = g.disjoint_routed_paths([((0, 0), (0, 4)), ((4, 4), (4, 0))])             # needs sage.numerical.mip

Though there is obviously no solution to the problem in which each corner is sending information to the opposite one:

sage: g = graphs.Grid2dGraph(5, 5)
sage: p1,p2 = g.disjoint_routed_paths([((0, 0), (4, 4)), ((0, 4), (4, 0))])             # needs sage.numerical.mip
Traceback (most recent call last):
...
EmptySetError: the disjoint routed paths do not exist
disjoint_union(other, labels='pairs', immutable=None)#

Return the disjoint union of self and other.

INPUT:

  • labels – string (default: 'pairs'); if set to 'pairs', each element v in the first graph will be named (0, v) and each element u in other will be named (1, u) in the result. If set to 'integers', the elements of the result will be relabeled with consecutive integers.

  • immutable – boolean (default: None); whether to create a mutable/immutable disjoint union. immutable=None (default) means that the graphs and their disjoint union will behave the same way.

See also

EXAMPLES:

sage: G = graphs.CycleGraph(3)
sage: H = graphs.CycleGraph(4)
sage: J = G.disjoint_union(H); J
Cycle graph disjoint_union Cycle graph: Graph on 7 vertices
sage: J.vertices(sort=True)
[(0, 0), (0, 1), (0, 2), (1, 0), (1, 1), (1, 2), (1, 3)]
sage: J = G.disjoint_union(H, labels='integers'); J
Cycle graph disjoint_union Cycle graph: Graph on 7 vertices
sage: J.vertices(sort=True)
[0, 1, 2, 3, 4, 5, 6]
sage: (G + H).vertices(sort=True)  # '+'-operator is a shortcut
[0, 1, 2, 3, 4, 5, 6]
sage: G = Graph({'a': ['b']})
sage: G.name("Custom path")
sage: G.name()
'Custom path'
sage: H = graphs.CycleGraph(3)
sage: J = G.disjoint_union(H); J
Custom path disjoint_union Cycle graph: Graph on 5 vertices
sage: J.vertices(sort=True)
[(0, 'a'), (0, 'b'), (1, 0), (1, 1), (1, 2)]
disjunctive_product(other)#

Return the disjunctive product of self and other.

The disjunctive product of \(G\) and \(H\) is the graph \(L\) with vertex set \(V(L)=V(G)\times V(H)\), and \(((u,v), (w,x))\) is an edge iff either :

  • \((u, w)\) is an edge of \(G\), or

  • \((v, x)\) is an edge of \(H\).

EXAMPLES:

sage: Z = graphs.CompleteGraph(2)
sage: D = Z.disjunctive_product(Z); D
Graph on 4 vertices
sage: D.plot()                      # long time                             # needs sage.plot
Graphics object consisting of 11 graphics primitives
sage: C = graphs.CycleGraph(5)
sage: D = C.disjunctive_product(Z); D
Graph on 10 vertices
sage: D.plot()                      # long time                             # needs sage.plot
Graphics object consisting of 46 graphics primitives
distance(u, v, by_weight=False, weight_function=None, check_weight=True)#

Return the (directed) distance from u to v in the (di)graph.

The distance is the length of the shortest path from u to v.

This method simply calls shortest_path_length(), with default arguments. For more information, and for more option, we refer to that method.

INPUT:

  • by_weight – boolean (default: False); if False, the graph is considered unweighted, and the distance is the number of edges in a shortest path. If True, the distance is the sum of edge labels (which are assumed to be numbers).

  • weight_function – function (default: None); a function that takes as input an edge (u, v, l) and outputs its weight. If not None, by_weight is automatically set to True. If None and by_weight is True, we use the edge label l, if l is not None, else 1 as a weight.

  • check_weight – boolean (default: True); whether to check that the weight_function outputs a number for each edge.

EXAMPLES:

sage: G = graphs.CycleGraph(9)
sage: G.distance(0,1)
1
sage: G.distance(0,4)
4
sage: G.distance(0,5)
4
sage: G = Graph({0:[], 1:[]})
sage: G.distance(0,1)
+Infinity
sage: G = Graph({ 0: {1: 1}, 1: {2: 1}, 2: {3: 1}, 3: {4: 2}, 4: {0: 2}}, sparse = True)
sage: G.plot(edge_labels=True).show()       # long time                     # needs sage.plot
sage: G.distance(0, 3)
2
sage: G.distance(0, 3, by_weight=True)
3
distance_all_pairs(by_weight=False, algorithm=None, weight_function=None, check_weight=True)#

Return the distances between all pairs of vertices.

INPUT:

  • by_weight boolean (default: \(False`\)); if True, the edges in the graph are weighted; if False, all edges have weight 1.

  • algorithm – string (default: None); one of the following algorithms:

    • 'BFS': the computation is done through a BFS centered on each vertex successively. Works only if by_weight==False.

    • 'Floyd-Warshall-Cython': the Cython implementation of the Floyd-Warshall algorithm. Works only if by_weight==False.

    • 'Floyd-Warshall-Python': the Python implementation of the Floyd-Warshall algorithm. Works also with weighted graphs, even with negative weights (but no negative cycle is allowed).

    • 'Dijkstra_NetworkX': the Dijkstra algorithm, implemented in NetworkX. It works with weighted graphs, but no negative weight is allowed.

    • 'Dijkstra_Boost': the Dijkstra algorithm, implemented in Boost (works only with positive weights).

    • 'Johnson_Boost': the Johnson algorithm, implemented in Boost (works also with negative weights, if there is no negative cycle).

    • None (default): Sage chooses the best algorithm: 'BFS' if by_weight is False, 'Dijkstra_Boost' if all weights are positive, 'Floyd-Warshall-Cython' otherwise.

  • weight_function – function (default: None); a function that takes as input an edge (u, v, l) and outputs its weight. If not None, by_weight is automatically set to True. If None and by_weight is True, we use the edge label l, if l is not None, else 1 as a weight.

  • check_weight – boolean (default: True); whether to check that the weight_function outputs a number for each edge.

OUTPUT:

A doubly indexed dictionary

Note

There is a Cython version of this method that is usually much faster for large graphs, as most of the time is actually spent building the final double dictionary. Everything on the subject is to be found in the distances_all_pairs module.

Note

This algorithm simply calls GenericGraph.shortest_path_all_pairs(), and we suggest to look at that method for more information and examples.

EXAMPLES:

The Petersen Graph:

sage: g = graphs.PetersenGraph()
sage: print(g.distance_all_pairs())
{0: {0: 0, 1: 1, 2: 2, 3: 2, 4: 1, 5: 1, 6: 2, 7: 2, 8: 2, 9: 2},
 1: {0: 1, 1: 0, 2: 1, 3: 2, 4: 2, 5: 2, 6: 1, 7: 2, 8: 2, 9: 2},
 2: {0: 2, 1: 1, 2: 0, 3: 1, 4: 2, 5: 2, 6: 2, 7: 1, 8: 2, 9: 2},
 3: {0: 2, 1: 2, 2: 1, 3: 0, 4: 1, 5: 2, 6: 2, 7: 2, 8: 1, 9: 2},
 4: {0: 1, 1: 2, 2: 2, 3: 1, 4: 0, 5: 2, 6: 2, 7: 2, 8: 2, 9: 1},
 5: {0: 1, 1: 2, 2: 2, 3: 2, 4: 2, 5: 0, 6: 2, 7: 1, 8: 1, 9: 2},
 6: {0: 2, 1: 1, 2: 2, 3: 2, 4: 2, 5: 2, 6: 0, 7: 2, 8: 1, 9: 1},
 7: {0: 2, 1: 2, 2: 1, 3: 2, 4: 2, 5: 1, 6: 2, 7: 0, 8: 2, 9: 1},
 8: {0: 2, 1: 2, 2: 2, 3: 1, 4: 2, 5: 1, 6: 1, 7: 2, 8: 0, 9: 2},
 9: {0: 2, 1: 2, 2: 2, 3: 2, 4: 1, 5: 2, 6: 1, 7: 1, 8: 2, 9: 0}}

Testing on Random Graphs:

sage: g = graphs.RandomGNP(20,.3)
sage: distances = g.distance_all_pairs()
sage: all((g.distance(0,v) == Infinity and v not in distances[0]) or
....:     g.distance(0,v) == distances[0][v] for v in g)
True
distance_matrix(vertices, base_ring=None, **kwds)#

Return the distance matrix of (di)graph.

The (di)graph is expected to be (strongly) connected.

The distance matrix of a (strongly) connected (di)graph is a matrix whose rows and columns are by default (vertices == None) indexed with the positions of the vertices of the (di)graph in the ordering vertices(). When vertices is set, the position of the vertices in this ordering is used. The intersection of row \(i\) and column \(j\) contains the shortest path distance from the vertex at the \(i\)-th position to the vertex at the \(j\)-th position.

Note that even when the vertices are consecutive integers starting from one, usually the vertex is not equal to its index.

INPUT:

  • vertices – list (default: None); the ordering of the vertices defining how they should appear in the matrix. By default, the ordering given by vertices() is used. Because vertices() only works if the vertices can be sorted, using vertices is useful when working with possibly non-sortable objects in Python 3.

  • base_ring – a ring (default: determined from the weights); the base ring of the matrix space to use.

  • **kwds – other keywords to pass to the subfunction distance_all_pairs() or to matrix()

EXAMPLES:

sage: d = DiGraph({1: [2, 3], 2: [3], 3: [4], 4: [1]})
sage: d.distance_matrix()                                                   # needs sage.modules
[0 1 1 2]
[3 0 1 2]
[2 3 0 1]
[1 2 2 0]
sage: d.distance_matrix(vertices=[4, 3, 2, 1])                              # needs sage.modules
[0 2 2 1]
[1 0 3 2]
[2 1 0 3]
[2 1 1 0]

sage: G = graphs.CubeGraph(3)
sage: G.distance_matrix()                                                   # needs sage.modules
[0 1 1 2 1 2 2 3]
[1 0 2 1 2 1 3 2]
[1 2 0 1 2 3 1 2]
[2 1 1 0 3 2 2 1]
[1 2 2 3 0 1 1 2]
[2 1 3 2 1 0 2 1]
[2 3 1 2 1 2 0 1]
[3 2 2 1 2 1 1 0]

The well known result of Graham and Pollak states that the determinant of the distance matrix of any tree of order \(n\) is \((-1)^{n-1}(n-1)2^{n-2}\):

sage: all(T.distance_matrix().det() == (-1)^9*(9)*2^8                       # needs sage.modules
....:     for T in graphs.trees(10))
True

See also

distances_distribution(G)#

Return the distances distribution of the (di)graph in a dictionary.

This method ignores all edge labels, so that the distance considered is the topological distance.

OUTPUT:

A dictionary d such that the number of pairs of vertices at distance k (if any) is equal to \(d[k] \cdot |V(G)| \cdot (|V(G)|-1)\).

Note

We consider that two vertices that do not belong to the same connected component are at infinite distance, and we do not take the trivial pairs of vertices \((v, v)\) at distance \(0\) into account. Empty (di)graphs and (di)graphs of order 1 have no paths and so we return the empty dictionary {}.

EXAMPLES:

An empty Graph:

sage: g = Graph()
sage: g.distances_distribution()
{}

A Graph of order 1:

sage: g = Graph()
sage: g.add_vertex(1)
sage: g.distances_distribution()
{}

A Graph of order 2 without edge:

sage: g = Graph()
sage: g.add_vertices([1,2])
sage: g.distances_distribution()
{+Infinity: 1}

The Petersen Graph:

sage: g = graphs.PetersenGraph()
sage: g.distances_distribution()
{1: 1/3, 2: 2/3}

A graph with multiple disconnected components:

sage: g = graphs.PetersenGraph()
sage: g.add_edge('good','wine')
sage: g.distances_distribution()
{1: 8/33, 2: 5/11, +Infinity: 10/33}

The de Bruijn digraph dB(2,3):

sage: D = digraphs.DeBruijn(2,3)                                                # needs sage.combinat
sage: D.distances_distribution()                                                # needs sage.combinat
{1: 1/4, 2: 11/28, 3: 5/14}
dominating_set(g, k, independent=1, total=False, connected=False, value_only=False, solver=False, verbose=None, integrality_tolerance=0)#

Return a minimum distance-\(k\) dominating set of the graph.

A minimum dominating set \(S\) of a graph \(G\) is a set of its vertices of minimal cardinality such that any vertex of \(G\) is in \(S\) or has one of its neighbors in \(S\). See the Wikipedia article Dominating_set.

A minimum distance-\(k\) dominating set is a set \(S\) of vertices of \(G\) of minimal cardinality such that any vertex of \(G\) is in \(S\) or at distance at most \(k\) from a vertex in \(S\). A distance-\(0\) dominating set is the set of vertices itself, and when \(k\) is the radius of the graph, any vertex dominates all the other vertices.

As an optimization problem, it can be expressed as follows, where \(N^k(u)\) denotes the set of vertices at distance at most \(k\) from \(u\) (the set of neighbors when \(k=1\)):

\[\begin{split}\mbox{Minimize : }&\sum_{v\in G} b_v\\ \mbox{Such that : }&\forall v \in G, b_v+\sum_{u \in N^k(v)} b_u\geq 1\\ &\forall x\in G, b_x\mbox{ is a binary variable}\end{split}\]

INPUT:

  • k – a non-negative integer (default: 1); the domination distance

  • independent – boolean (default: False); when True, computes a minimum independent dominating set, that is a minimum dominating set that is also an independent set (see also independent_set())

  • total – boolean (default: False); when True, computes a total dominating set (see the See the Wikipedia article Dominating_set)

  • connected – boolean (default: False); when True, computes a connected dominating set (see Wikipedia article Connected_dominating_set)

  • value_only – boolean (default: False); whether to only return the cardinality of the computed dominating set, or to return its list of vertices (default)

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

EXAMPLES:

A basic illustration on a PappusGraph:

sage: g = graphs.PappusGraph()
sage: g.dominating_set(value_only=True)                                         # needs sage.numerical.mip
5

If we build a graph from two disjoint stars, then link their centers we will find a difference between the cardinality of an independent set and a stable independent set:

sage: g = 2 * graphs.StarGraph(5)
sage: g.add_edge(0, 6)
sage: len(g.dominating_set())                                                   # needs sage.numerical.mip
2
sage: len(g.dominating_set(independent=True))                                   # needs sage.numerical.mip
6

The total dominating set of the Petersen graph has cardinality 4:

sage: G = graphs.PetersenGraph()
sage: G.dominating_set(total=True, value_only=True)                             # needs sage.numerical.mip
4

The dominating set is calculated for both the directed and undirected graphs (modification introduced in github issue #17905):

sage: g = digraphs.Path(3)
sage: g.dominating_set(value_only=True)                                         # needs sage.numerical.mip
2
sage: g = graphs.PathGraph(3)
sage: g.dominating_set(value_only=True)                                         # needs sage.numerical.mip
1

Cardinality of distance-\(k\) dominating sets:

sage: G = graphs.PetersenGraph()
sage: [G.dominating_set(k=k, value_only=True) for k in range(G.radius() + 1)]   # needs sage.numerical.mip
[10, 3, 1]
sage: G = graphs.PathGraph(5)
sage: [G.dominating_set(k=k, value_only=True) for k in range(G.radius() + 1)]   # needs sage.numerical.mip
[5, 2, 1]
dominating_sets(g, k, independent=1, total=False, connected=False, solver=False, verbose=None, integrality_tolerance=0)#

Return an iterator over the minimum distance-\(k\) dominating sets of the graph.

A minimum dominating set \(S\) of a graph \(G\) is a set of its vertices of minimal cardinality such that any vertex of \(G\) is in \(S\) or has one of its neighbors in \(S\). See the Wikipedia article Dominating_set.

A minimum distance-\(k\) dominating set is a set \(S\) of vertices of \(G\) of minimal cardinality such that any vertex of \(G\) is in \(S\) or at distance at most \(k\) from a vertex in \(S\). A distance-\(0\) dominating set is the set of vertices itself, and when \(k\) is the radius of the graph, any vertex dominates all the other vertices.

As an optimization problem, it can be expressed as follows, where \(N^k(u)\) denotes the set of vertices at distance at most \(k\) from \(u\) (the set of neighbors when \(k=1\)):

\[\begin{split}\mbox{Minimize : }&\sum_{v\in G} b_v\\ \mbox{Such that : }&\forall v \in G, b_v+\sum_{u \in N^k(v)} b_u\geq 1\\ &\forall x\in G, b_x\mbox{ is a binary variable}\end{split}\]

We use constraints generation to iterate over the minimum distance-\(k\) dominating sets. That is, after reporting a solution, we add a constraint to discard it and solve the problem again until no more solution can be found.

INPUT:

  • k – a non-negative integer (default: 1); the domination distance

  • independent – boolean (default: False); when True, computes minimum independent dominating sets, that is minimum dominating sets that are also independent sets (see also independent_set())

  • total – boolean (default: False); when True, computes total dominating sets (see the See the Wikipedia article Dominating_set)

  • connected – boolean (default: False); when True, computes connected dominating sets (see Wikipedia article Connected_dominating_set)

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

EXAMPLES:

Number of distance-\(k\) dominating sets of a Path graph of order 10:

sage: g = graphs.PathGraph(10)
sage: [sum(1 for _ in g.dominating_sets(k=k)) for k in range(11)]               # needs sage.numerical.mip
[1, 13, 1, 13, 25, 2, 4, 6, 8, 10, 10]

If we build a graph from two disjoint stars, then link their centers we will find a difference between the cardinality of an independent set and a stable independent set:

sage: g = 2 * graphs.StarGraph(5)
sage: g.add_edge(0, 6)
sage: [sum(1 for _ in g.dominating_sets(k=k)) for k in range(11)]               # needs sage.numerical.mip
[1, 1, 2, 12, 12, 12, 12, 12, 12, 12, 12]

The total dominating set of the Petersen graph has cardinality 4:

sage: G = graphs.PetersenGraph()
sage: G.dominating_set(total=True, value_only=True)                             # needs sage.numerical.mip
4
sage: sorted(G.dominating_sets(k=1))                                            # needs sage.numerical.mip
[[0, 2, 6],
 [0, 3, 9],
 [0, 7, 8],
 [1, 3, 7],
 [1, 4, 5],
 [1, 8, 9],
 [2, 4, 8],
 [2, 5, 9],
 [3, 5, 6],
 [4, 6, 7]]

Independent distance-\(k\) dominating sets of a Path graph:

sage: # needs sage.numerical.mip
sage: G = graphs.PathGraph(6)
sage: sorted(G.dominating_sets(k=1, independent=True))
[[1, 4]]
sage: sorted(G.dominating_sets(k=2, independent=True))
[[0, 3], [0, 4], [0, 5], [1, 3], [1, 4], [1, 5], [2, 4], [2, 5]]
sage: sorted(G.dominating_sets(k=3, independent=True))
[[2], [3]]

The dominating set is calculated for both the directed and undirected graphs (modification introduced in github issue #17905):

sage: # needs sage.numerical.mip
sage: g = digraphs.Path(3)
sage: g.dominating_set(value_only=True)
2
sage: list(g.dominating_sets())
[[0, 1], [0, 2]]
sage: list(g.dominating_sets(k=2))
[[0]]
sage: g = graphs.PathGraph(3)
sage: g.dominating_set(value_only=True)
1
sage: next(g.dominating_sets())
[1]

Minimum connected dominating sets of the Peterson graph:

sage: G = graphs.PetersenGraph()
sage: G.dominating_set(total=True, value_only=True)                             # needs sage.numerical.mip
4
sage: sorted(G.dominating_sets(k=1, connected=True))
[[0, 1, 2, 6],
 [0, 1, 4, 5],
 [0, 3, 4, 9],
 [0, 5, 7, 8],
 [1, 2, 3, 7],
 [1, 6, 8, 9],
 [2, 3, 4, 8],
 [2, 5, 7, 9],
 [3, 5, 6, 8],
 [4, 6, 7, 9]]

Subgraph induced by the dominating set is connected:

sage: G = graphs.PetersenGraph()
sage: all(G.subgraph(vertices=dom).is_connected()
....:     for dom in G.dominating_set(k=1, connected=True))
True

Minimum distance-k connected dominating sets of the Tietze graph:

sage: G = graphs.TietzeGraph()
sage: sorted(G.dominating_sets(k=2, connected=True))
[[0, 9], [1, 0], [2, 3], [4, 3], [5, 6], [7, 6], [8, 0], [10, 3], [11, 6]]
sage: sorted(G.dominating_sets(k=3, connected=True))
[[0], [1], [2], [3], [4], [5], [6], [7], [8], [9], [10], [11]]
dominator_tree(g, root, return_dict=False, reverse=False)#

Use Boost to compute the dominator tree of g, rooted at root.

A node \(d\) dominates a node \(n\) if every path from the entry node root to \(n\) must go through \(d\). The immediate dominator of a node \(n\) is the unique node that strictly dominates \(n\) but does not dominate any other node that dominates \(n\). A dominator tree is a tree where each node’s children are those nodes it immediately dominates. For more information, see the Wikipedia article Dominator_(graph_theory).

If the graph is connected and undirected, the parent of a vertex \(v\) is:

  • the root if \(v\) is in the same biconnected component as the root;

  • the first cut vertex in a path from \(v\) to the root, otherwise.

If the graph is not connected, the dominator tree of the whole graph is equal to the dominator tree of the connected component of the root.

If the graph is directed, computing a dominator tree is more complicated, and it needs time \(O(m\log m)\), where \(m\) is the number of edges. The implementation provided by Boost is the most general one, so it needs time \(O(m\log m)\) even for undirected graphs.

INPUT:

  • g – the input Sage (Di)Graph

  • root – the root of the dominator tree

  • return_dict – boolean (default: False); if True, the function returns a dictionary associating to each vertex its parent in the dominator tree. If False (default), it returns the whole tree, as a Graph or a DiGraph.

  • reverse – boolean (default: False); when set to True, computes the dominator tree in the reverse graph

OUTPUT:

The dominator tree, as a graph or as a dictionary, depending on the value of return_dict. If the output is a dictionary, it will contain None in correspondence of root and of vertices that are not reachable from root. If the output is a graph, it will not contain vertices that are not reachable from root.

EXAMPLES:

An undirected grid is biconnected, and its dominator tree is a star (everyone’s parent is the root):

sage: g = graphs.GridGraph([2,2]).dominator_tree((0,0))
sage: g.to_dictionary()
{(0, 0): [(0, 1), (1, 0), (1, 1)], (0, 1): [(0, 0)], (1, 0): [(0, 0)], (1, 1): [(0, 0)]}

If the graph is made by two 3-cycles \(C_1,C_2\) connected by an edge \((v,w)\), with \(v \in C_1\), \(w \in C_2\), the cut vertices are \(v\) and \(w\), the biconnected components are \(C_1\), \(C_2\), and the edge \((v,w)\). If the root is in \(C_1\), the parent of each vertex in \(C_1\) is the root, the parent of \(w\) is \(v\), and the parent of each vertex in \(C_2\) is \(w\):

sage: G = 2 * graphs.CycleGraph(3)
sage: v = 0
sage: w = 3
sage: G.add_edge(v,w)
sage: G.dominator_tree(1, return_dict=True)
{0: 1, 1: None, 2: 1, 3: 0, 4: 3, 5: 3}

An example with a directed graph:

sage: g = digraphs.Circuit(10).dominator_tree(5)
sage: g.to_dictionary()
{0: [1], 1: [2], 2: [3], 3: [4], 4: [], 5: [6], 6: [7], 7: [8], 8: [9], 9: [0]}
sage: g = digraphs.Circuit(10).dominator_tree(5, reverse=True)
sage: g.to_dictionary()
{0: [9], 1: [0], 2: [1], 3: [2], 4: [3], 5: [4], 6: [], 7: [6], 8: [7], 9: [8]}

If the output is a dictionary:

sage: graphs.GridGraph([2,2]).dominator_tree((0,0), return_dict=True)
{(0, 0): None, (0, 1): (0, 0), (1, 0): (0, 0), (1, 1): (0, 0)}
edge_boundary(vertices1, vertices2=None, labels=True, sort=False, key=None)#

Return a list of edges (u,v,l) with u in vertices1 and v in vertices2.

If vertices2 is None, then it is set to the complement of vertices1.

In a digraph, the external boundary of a vertex \(v\) are those vertices \(u\) with an arc \((v, u)\).

INPUT:

  • labels – boolean (default: True); if False, each edge is a tuple \((u,v)\) of vertices

  • sort – boolean (default: False); whether to sort the result

  • key – a function (default: None); a function that takes an edge as its one argument and returns a value that can be used for comparisons in the sorting algorithm (we must have sort=True)

EXAMPLES:

sage: K = graphs.CompleteBipartiteGraph(9, 3)
sage: len(K.edge_boundary([0, 1, 2, 3, 4, 5, 6, 7, 8], [9, 10, 11]))
27
sage: K.size()
27

Note that the edge boundary preserves direction:

sage: K = graphs.CompleteBipartiteGraph(9, 3).to_directed()
sage: len(K.edge_boundary([0, 1, 2, 3, 4, 5, 6, 7, 8], [9, 10, 11]))
27
sage: K.size()
54
sage: D = DiGraph({0: [1, 2], 3: [0]})
sage: D.edge_boundary([0], sort=True)
[(0, 1, None), (0, 2, None)]
sage: D.edge_boundary([0], labels=False, sort=True)
[(0, 1), (0, 2)]

Using the key argument to order multiple edges of incomparable types (see github issue #35903):

sage: G = Graph([(1, 'A', 4), (1, 2, 3)])
sage: G.edge_boundary([1], sort=True)
Traceback (most recent call last):
...
TypeError: unsupported operand parent(s) for <: 'Integer Ring' and '<class 'str'>'
sage: G.edge_boundary([1], sort=True, key=str)
[('A', 1, 4), (1, 2, 3)]
sage: G.edge_boundary([1], sort=True, key=lambda e:e[2])
[(1, 2, 3), ('A', 1, 4)]
sage: G.edge_boundary([1], labels=False, sort=True, key=lambda e:e[2])
Traceback (most recent call last):
...
IndexError: tuple index out of range
edge_connectivity(G, value_only=True, implementation=None, use_edge_labels=False, vertices=False, solver=None, verbose=0, integrality_tolerance=0.001)#

Return the edge connectivity of the graph.

For more information, see the Wikipedia article Connectivity_(graph_theory).

Note

When the graph is a directed graph, this method actually computes the strong connectivity, (i.e. a directed graph is strongly \(k\)-connected if there are \(k\) disjoint paths between any two vertices \(u, v\)). If you do not want to consider strong connectivity, the best is probably to convert your DiGraph object to a Graph object, and compute the connectivity of this other graph.

INPUT:

  • G – the input Sage (Di)Graph

  • value_only – boolean (default: True)

    • When set to True (default), only the value is returned.

    • When set to False, both the value and a minimum vertex cut are returned.

  • implementation – string (default: None); selects an implementation:

    • None (default) – selects the best implementation available

    • "boost" – use the Boost graph library (which is much more efficient). It is not available when edge_labels=True, and it is unreliable for directed graphs (see github issue #18753).

    -"Sage" – use Sage’s implementation based on integer linear

    programming

  • use_edge_labels – boolean (default: False)

    • When set to True, computes a weighted minimum cut where each edge has a weight defined by its label. (If an edge has no label, \(1\) is assumed.). Implies boost = False.

    • When set to False, each edge has weight \(1\).

  • vertices – boolean (default: False)

    • When set to True, also returns the two sets of vertices that are disconnected by the cut. Implies value_only=False.

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

EXAMPLES:

A basic application on the PappusGraph:

sage: from sage.graphs.connectivity import edge_connectivity
sage: g = graphs.PappusGraph()
sage: edge_connectivity(g)
3
sage: g.edge_connectivity()
3

The edge connectivity of a complete graph is its minimum degree, and one of the two parts of the bipartition is reduced to only one vertex. The graph of the cut edges is isomorphic to a Star graph:

sage: g = graphs.CompleteGraph(5)
sage: [ value, edges, [ setA, setB ]] = edge_connectivity(g,vertices=True)
sage: value
4
sage: len(setA) == 1 or len(setB) == 1
True
sage: cut = Graph()
sage: cut.add_edges(edges)
sage: cut.is_isomorphic(graphs.StarGraph(4))
True

Even if obviously in any graph we know that the edge connectivity is less than the minimum degree of the graph:

sage: g = graphs.RandomGNP(10,.3)
sage: min(g.degree()) >= edge_connectivity(g)
True

If we build a tree then assign to its edges a random value, the minimum cut will be the edge with minimum value:

sage: tree = graphs.RandomTree(10)
sage: for u,v in tree.edge_iterator(labels=None):
....:      tree.set_edge_label(u, v, random())
sage: minimum = min(tree.edge_labels())
sage: [_, [(_, _, l)]] = edge_connectivity(tree, value_only=False,              # needs sage.numerical.mip
....:                                      use_edge_labels=True)
sage: l == minimum                                                              # needs sage.numerical.mip
True

When value_only=True and implementation="sage", this function is optimized for small connectivity values and does not need to build a linear program.

It is the case for graphs which are not connected

sage: g = 2 * graphs.PetersenGraph()
sage: edge_connectivity(g, implementation="sage")
0.0

For directed graphs, the strong connectivity is tested through the dedicated function:

sage: g = digraphs.ButterflyGraph(3)
sage: edge_connectivity(g, implementation="sage")
0.0

We check that the result with Boost is the same as the result without Boost:

sage: g = graphs.RandomGNP(15, .3)
sage: (edge_connectivity(g, implementation="boost")                             # needs sage.numerical.mip
....:    == edge_connectivity(g, implementation="sage"))
True

Boost interface also works with directed graphs:

sage: edge_connectivity(digraphs.Circuit(10), implementation="boost",
....:                   vertices=True)
[1, [(0, 1)], [{0}, {1, 2, 3, 4, 5, 6, 7, 8, 9}]]

However, the Boost algorithm is not reliable if the input is directed (see github issue #18753):

sage: g = digraphs.Path(3)
sage: edge_connectivity(g)
0.0
sage: edge_connectivity(g, implementation="boost")
1
sage: g.add_edge(1, 0)
sage: edge_connectivity(g)
0.0
sage: edge_connectivity(g, implementation="boost")
0
edge_cut(s, t, value_only, use_edge_labels=True, vertices=False, algorithm=False, solver='FF', verbose=None, integrality_tolerance=0)#

Return a minimum edge cut between vertices \(s\) and \(t\).

A minimum edge cut between two vertices \(s\) and \(t\) of self is a set \(A\) of edges of minimum weight such that the graph obtained by removing \(A\) from the graph is disconnected. For more information, see the Wikipedia article Cut_(graph_theory).

INPUT:

  • s – source vertex

  • t – sink vertex

  • value_only – boolean (default: True); whether to return only the weight of a minimum cut (True) or a list of edges of a minimum cut (False)

  • use_edge_labels – boolean (default: False); whether to compute a weighted minimum edge cut where the weight of an edge is defined by its label (if an edge has no label, \(1\) is assumed), or to compute a cut of minimum cardinality (i.e., edge weights are set to 1)

  • vertices – boolean (default: False); whether set to True, return a list of edges in the edge cut and the two sets of vertices that are disconnected by the cut

    Note: vertices=True implies value_only=False.

  • algorithm – string (default: 'FF'); algorithm to use:

    • If algorithm = "FF", a Python implementation of the Ford-Fulkerson algorithm is used

    • If algorithm = "LP", the problem is solved using Linear Programming.

    • If algorithm = "igraph", the igraph implementation of the Goldberg-Tarjan algorithm is used (only available when igraph is installed)

    • If algorithm = None, the problem is solved using the default maximum flow algorithm (see flow())

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

Note

The use of Linear Programming for non-integer problems may possibly mean the presence of a (slight) numerical noise.

OUTPUT:

Real number or tuple, depending on the given arguments (examples are given below).

EXAMPLES:

A basic application in the Pappus graph:

sage: g = graphs.PappusGraph()
sage: g.edge_cut(1, 2, value_only=True)
3

Or on Petersen’s graph, with the corresponding bipartition of the vertex set:

sage: g = graphs.PetersenGraph()
sage: g.edge_cut(0, 3, vertices=True)
[3, [(0, 1, None), (0, 4, None), (0, 5, None)], [[0], [1, 2, 3, 4, 5, 6, 7, 8, 9]]]

If the graph is a path with randomly weighted edges:

sage: g = graphs.PathGraph(15)
sage: for u,v in g.edge_iterator(labels=None):
....:    g.set_edge_label(u, v, random())

The edge cut between the two ends is the edge of minimum weight:

sage: minimum = min(g.edge_labels())
sage: minimum == g.edge_cut(0, 14, use_edge_labels=True)
True
sage: [value, [e]] = g.edge_cut(0, 14, use_edge_labels=True, value_only=False)
sage: g.edge_label(e[0], e[1]) == minimum
True

The two sides of the edge cut are obviously shorter paths:

sage: value, edges, [set1, set2] = g.edge_cut(0, 14, use_edge_labels=True,
....:                                         vertices=True)
sage: g.subgraph(set1).is_isomorphic(graphs.PathGraph(len(set1)))
True
sage: g.subgraph(set2).is_isomorphic(graphs.PathGraph(len(set2)))
True
sage: len(set1) + len(set2) == g.order()
True
edge_disjoint_paths(s, t, algorithm, solver='FF', verbose=None, integrality_tolerance=False)#

Return a list of edge-disjoint paths between two vertices.

The edge version of Menger’s theorem asserts that the size of the minimum edge cut between two vertices \(s\) and`t` (the minimum number of edges whose removal disconnects \(s\) and \(t\)) is equal to the maximum number of pairwise edge-independent paths from \(s\) to \(t\).

This function returns a list of such paths.

INPUT:

  • algorithm – string (default: "FF"); the algorithm to use among:

    • "FF", a Python implementation of the Ford-Fulkerson algorithm

    • "LP", the flow problem is solved using Linear Programming

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

    Only used when \(àlgorithm`\) is "LP".

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

    Only used when \(àlgorithm`\) is "LP".

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

    Only used when \(àlgorithm`\) is "LP".

Note

This function is topological: it does not take the eventual weights of the edges into account.

EXAMPLES:

In a complete bipartite graph

sage: g = graphs.CompleteBipartiteGraph(2, 3)
sage: g.edge_disjoint_paths(0, 1)
[[0, 2, 1], [0, 3, 1], [0, 4, 1]]
edge_disjoint_spanning_trees(k, algorithm, root=None, solver=None, verbose=None, integrality_tolerance=0)#

Return the desired number of edge-disjoint spanning trees/arborescences.

INPUT:

  • k – integer; the required number of edge-disjoint spanning trees/arborescences

  • algorithm – string (default: None); specify the algorithm to use among:

    • "Roskind-Tarjan" – use the algorithm proposed by Roskind and Tarjan [RT1985] for finding edge-disjoint spanning-trees in undirected simple graphs in time \(O(m\log{m} + k^2n^2)\).

    • "MILP" – use a mixed integer linear programming formulation. This is the default method for directed graphs.

    • None – use "Roskind-Tarjan" for undirected graphs and "MILP" for directed graphs.

  • root – vertex (default: None); root of the disjoint arborescences when the graph is directed. If set to None, the first vertex in the graph is picked.

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

ALGORITHM:

Mixed Integer Linear Program.

There are at least two possible rewritings of this method which do not use Linear Programming:

  • The algorithm presented in the paper entitled “A short proof of the tree-packing theorem”, by Thomas Kaiser [Kai2012].

  • The implementation of a Matroid class and of the Matroid Union Theorem (see section 42.3 of [Sch2003]), applied to the cycle Matroid (see chapter 51 of [Sch2003]).

EXAMPLES:

The Petersen Graph does have a spanning tree (it is connected):

sage: g = graphs.PetersenGraph()
sage: [T] = g.edge_disjoint_spanning_trees(1)                               # needs sage.numerical.mip
sage: T.is_tree()                                                           # needs sage.numerical.mip
True

Though, it does not have 2 edge-disjoint trees (as it has less than \(2(|V|-1)\) edges):

sage: g.edge_disjoint_spanning_trees(2)                                     # needs sage.numerical.mip
Traceback (most recent call last):
...
EmptySetError: this graph does not contain the required number of trees/arborescences

By Edmonds’ theorem, a graph which is \(k\)-connected always has \(k\) edge-disjoint arborescences, regardless of the root we pick:

sage: # needs sage.numerical.mip
sage: g = digraphs.RandomDirectedGNP(11, .3)  # reduced from 30 to 11, cf. #32169
sage: k = Integer(g.edge_connectivity())
sage: while not k:
....:     g = digraphs.RandomDirectedGNP(11, .3)
....:     k = Integer(g.edge_connectivity())
sage: arborescences = g.edge_disjoint_spanning_trees(k)     # long time (up to 15s on sage.math, 2011)
sage: all(a.is_directed_acyclic() for a in arborescences)   # long time
True
sage: all(a.is_connected() for a in arborescences)  # long time
True

In the undirected case, we can only ensure half of it:

sage: # needs sage.numerical.mip
sage: g = graphs.RandomGNP(14, .3)  # reduced from 30 to 14, see #32169
sage: while not g.is_biconnected():
....:     g = graphs.RandomGNP(14, .3)
sage: k = Integer(g.edge_connectivity()) // 2
sage: trees = g.edge_disjoint_spanning_trees(k)
sage: all(t.is_tree() for t in trees)
True

Check the validity of the algorithms for undirected graphs:

sage: # needs sage.numerical.mip
sage: g = graphs.RandomGNP(12, .7)
sage: k = Integer(g.edge_connectivity()) // 2
sage: trees = g.edge_disjoint_spanning_trees(k, algorithm="MILP")
sage: all(t.is_tree() for t in trees)
True
sage: all(g.order() == t.size() + 1 for t in trees)
True
sage: trees = g.edge_disjoint_spanning_trees(k, algorithm="Roskind-Tarjan")
sage: all(t.is_tree() for t in trees)
True
sage: all(g.order() == t.size() + 1 for t in trees)
True

Example of github issue #32169:

sage: d6 = r'[E_S?_hKIH@eos[BSg???Q@FShGC?hTHUGM?IPug?JOEYCdOzdkQGo'
sage: d6 += r'@ADA@AAg?GAQW?[aIaSwHYcD@qQb@Dd?\hJTI@OHlJ_?C_OEIKoeC'
sage: d6 += r'R@_BC?Q??YBFosqITEA?IvCU_'
sage: G = DiGraph(d6, format='dig6')
sage: G.edge_connectivity()                                                 # needs sage.numerical.mip
5
sage: G.edge_disjoint_spanning_trees(5)     # long time                     # needs sage.numerical.mip
[Digraph on 28 vertices,
 Digraph on 28 vertices,
 Digraph on 28 vertices,
 Digraph on 28 vertices,
 Digraph on 28 vertices]

Small cases:

sage: # needs sage.numerical.mip
sage: Graph().edge_disjoint_spanning_trees(0)
[]
sage: Graph(1).edge_disjoint_spanning_trees(0)
[]
sage: Graph(2).edge_disjoint_spanning_trees(0)
[]
sage: Graph([(0, 1)]).edge_disjoint_spanning_trees(0)
[]
sage: Graph([(0, 1)]).edge_disjoint_spanning_trees(1)
[Graph on 2 vertices]
sage: Graph([(0, 1)]).edge_disjoint_spanning_trees(2)
Traceback (most recent call last):
...
EmptySetError: this graph does not contain the required number of trees/arborescences
edge_iterator(vertices=None, labels=True, ignore_direction=False, sort_vertices=True)#

Return an iterator over edges.

The iterator returned is over the edges incident with any vertex given in the parameter vertices. If the graph is directed, iterates over edges going out only. If vertices is None, then returns an iterator over all edges. If self is directed, returns outgoing edges only.

INPUT:

  • vertices – object (default: None); a vertex, a list of vertices or None

  • labels – boolean (default: True); if False, each edge is

    a tuple \((u,v)\) of vertices

  • ignore_direction – boolean (default: False); only applies to

    directed graphs. If True, searches across edges in either direction.

  • sort_vertices – boolean (default: True); only applies to undirected graphs. If True, sort the ends of the edges. Not sorting the ends is faster.

Note

It is somewhat safe to modify the graph during iterating.

vertices must be specified if modifying the vertices.

Without multiedges, you can safely use this graph to relabel edges or delete some edges. If you add edges, they might later appear in the iterator or not (depending on the internal order of vertices).

In case of multiedges, all arcs from one vertex to another are internally cached. So the iterator will yield them, even if you delete them all after seeing the first one.

EXAMPLES:

sage: for i in graphs.PetersenGraph().edge_iterator([0]):
....:  print(i)
(0, 1, None)
(0, 4, None)
(0, 5, None)
sage: D = DiGraph({0: [1, 2], 1: [0]})
sage: for i in D.edge_iterator([0]):
....:  print(i)
(0, 1, None)
(0, 2, None)
sage: G = graphs.TetrahedralGraph()
sage: list(G.edge_iterator(labels=False))
[(0, 1), (0, 2), (0, 3), (1, 2), (1, 3), (2, 3)]
sage: G = graphs.TetrahedralGraph()
sage: list(G.edge_iterator(labels=False, sort_vertices=False))
[(1, 0), (2, 0), (3, 0), (2, 1), (3, 1), (3, 2)]
sage: D = DiGraph({1: [0], 2: [0]})
sage: list(D.edge_iterator(0))
[]
sage: list(D.edge_iterator(0, ignore_direction=True))
[(1, 0, None), (2, 0, None)]
edge_label(u, v)#

Return the label of an edge.

If the graph allows multiple edges, then the list of labels on the edges is returned.

See also

EXAMPLES:

sage: G = Graph({0: {1: 'edgelabel'}})
sage: G.edge_label(0, 1)
'edgelabel'
sage: D = DiGraph({1: {2: 'up'}, 2: {1: 'down'}})
sage: D.edge_label(2, 1)
'down'
sage: G = Graph(multiedges=True)
sage: [G.add_edge(0, 1, i) for i in range(1, 6)]
[None, None, None, None, None]
sage: sorted(G.edge_label(0, 1))
[1, 2, 3, 4, 5]
edge_labels()#

Return a list of the labels of all edges in self.

The output list is not sorted.

EXAMPLES:

sage: G = Graph({0: {1: 'x', 2: 'z', 3: 'a'}, 2: {5: 'out'}}, sparse=True)
sage: G.edge_labels()
['x', 'z', 'a', 'out']
sage: G = DiGraph({0: {1: 'x', 2: 'z', 3: 'a'}, 2: {5: 'out'}}, sparse=True)
sage: G.edge_labels()
['x', 'z', 'a', 'out']
edge_polytope(backend=None)#

Return the edge polytope of self.

The edge polytope (EP) of a Graph on \(n\) vertices is the polytope in \(\ZZ^{n}\) defined as the convex hull of \(e_i + e_j\) for each edge \((i, j)\). Here \(e_1, \dots, e_n\) denotes the standard basis.

INPUT:

  • backend – string or None (default); the backend to use; see sage.geometry.polyhedron.constructor.Polyhedron()

EXAMPLES:

The EP of a \(4\)-cycle is a square:

sage: G = graphs.CycleGraph(4)
sage: P = G.edge_polytope(); P                                              # needs sage.geometry.polyhedron
A 2-dimensional polyhedron in ZZ^4 defined as the convex hull of 4 vertices

The EP of a complete graph on \(4\) vertices is cross polytope:

sage: G = graphs.CompleteGraph(4)
sage: P = G.edge_polytope(); P                                              # needs sage.geometry.polyhedron
A 3-dimensional polyhedron in ZZ^4 defined as the convex hull of 6 vertices
sage: P.is_combinatorially_isomorphic(polytopes.cross_polytope(3))          # needs sage.geometry.polyhedron
True

The EP of a graph is isomorphic to the subdirect sum of its connected components EPs:

sage: n = randint(3, 6)
sage: G1 = graphs.RandomGNP(n, 0.2)                                         # needs networkx
sage: n = randint(3, 6)
sage: G2 = graphs.RandomGNP(n, 0.2)                                         # needs networkx
sage: G = G1.disjoint_union(G2)                                             # needs networkx
sage: P = G.edge_polytope()                                                 # needs networkx sage.geometry.polyhedron
sage: P1 = G1.edge_polytope()                                               # needs networkx sage.geometry.polyhedron
sage: P2 = G2.edge_polytope()                                               # needs networkx sage.geometry.polyhedron
sage: P.is_combinatorially_isomorphic(P1.subdirect_sum(P2))                 # needs networkx sage.geometry.polyhedron
True

All trees on \(n\) vertices have isomorphic EPs:

sage: n = randint(4, 10)
sage: G1 = graphs.RandomTree(n)
sage: G2 = graphs.RandomTree(n)
sage: P1 = G1.edge_polytope()                                               # needs sage.geometry.polyhedron
sage: P2 = G2.edge_polytope()                                               # needs sage.geometry.polyhedron
sage: P1.is_combinatorially_isomorphic(P2)                                  # needs sage.geometry.polyhedron
True

However, there are still many different EPs:

sage: len(list(graphs(5)))
34
sage: polys = []
sage: for G in graphs(5):                                                   # needs sage.geometry.polyhedron
....:     P = G.edge_polytope()
....:     for P1 in polys:
....:         if P.is_combinatorially_isomorphic(P1):
....:             break
....:     else:
....:         polys.append(P)
sage: len(polys)                                                            # needs sage.geometry.polyhedron
19
edges(vertices=None, labels=True, sort=False, key=None, ignore_direction=False, sort_vertices=True)#

Return a EdgesView of edges.

Each edge is a triple (u, v, l) where u and v are vertices and l is a label. If the parameter labels is False then a list of couple (u, v) is returned where u and v are vertices.

The returned EdgesView is over the edges incident with any vertex given in the parameter vertices (all edges if None). If self is directed, iterates over outgoing edges only, unless parameter ignore_direction is True in which case it searches across edges in either direction.

INPUT:

  • vertices – object (default: None); a vertex, a list of vertices or None

  • labels – boolean (default: True); if False, each edge is simply a pair (u, v) of vertices

  • sort – boolean (default: False); whether to sort edges according the ordering specified with parameter key. If False (default), edges are not sorted. This is the fastest and less memory consuming method for iterating over edges.

  • key – a function (default: None); a function that takes an edge (a pair or a triple, according to the labels keyword) as its one argument and returns a value that can be used for comparisons in the sorting algorithm

  • ignore_direction – boolean (default: False); only applies to

    directed graphs. If True, searches across edges in either direction.

  • sort_vertices – boolean (default: True); only applies to undirected graphs. If True, sort the ends of the edges. Not sorting the ends is faster.

OUTPUT: A EdgesView.

Warning

Since any object may be a vertex, there is no guarantee that any two vertices will be comparable, and thus no guarantee how two edges may compare. With default objects for vertices (all integers), or when all the vertices are of the same simple type, then there should not be a problem with how the vertices will be sorted. However, if you need to guarantee a total order for the sorting of the edges, use the key argument, as illustrated in the examples below.

EXAMPLES:

sage: graphs.DodecahedralGraph().edges(sort=True)
[(0, 1, None), (0, 10, None), (0, 19, None), (1, 2, None),
 (1, 8, None), (2, 3, None), (2, 6, None), (3, 4, None),
 (3, 19, None), (4, 5, None), (4, 17, None), (5, 6, None),
 (5, 15, None), (6, 7, None), (7, 8, None), (7, 14, None),
 (8, 9, None), (9, 10, None), (9, 13, None), (10, 11, None),
 (11, 12, None), (11, 18, None), (12, 13, None), (12, 16, None),
 (13, 14, None), (14, 15, None), (15, 16, None), (16, 17, None),
 (17, 18, None), (18, 19, None)]
sage: graphs.DodecahedralGraph().edges(sort=True, labels=False)
[(0, 1), (0, 10), (0, 19), (1, 2), (1, 8), (2, 3), (2, 6), (3, 4),
 (3, 19), (4, 5), (4, 17), (5, 6), (5, 15), (6, 7), (7, 8), (7, 14),
 (8, 9), (9, 10), (9, 13), (10, 11), (11, 12), (11, 18), (12, 13),
 (12, 16), (13, 14), (14, 15), (15, 16), (16, 17), (17, 18),
 (18, 19)]
sage: D = graphs.DodecahedralGraph().to_directed()
sage: D.edges(sort=True)
[(0, 1, None), (0, 10, None), (0, 19, None), (1, 0, None),
 (1, 2, None), (1, 8, None), (2, 1, None), (2, 3, None),
 (2, 6, None), (3, 2, None), (3, 4, None), (3, 19, None),
 (4, 3, None), (4, 5, None), (4, 17, None), (5, 4, None),
 (5, 6, None), (5, 15, None), (6, 2, None), (6, 5, None),
 (6, 7, None), (7, 6, None), (7, 8, None), (7, 14, None),
 (8, 1, None), (8, 7, None), (8, 9, None), (9, 8, None),
 (9, 10, None), (9, 13, None), (10, 0, None), (10, 9, None),
 (10, 11, None), (11, 10, None), (11, 12, None), (11, 18, None),
 (12, 11, None), (12, 13, None), (12, 16, None), (13, 9, None),
 (13, 12, None), (13, 14, None), (14, 7, None), (14, 13, None),
 (14, 15, None), (15, 5, None), (15, 14, None), (15, 16, None),
 (16, 12, None), (16, 15, None), (16, 17, None), (17, 4, None),
 (17, 16, None), (17, 18, None), (18, 11, None), (18, 17, None),
 (18, 19, None), (19, 0, None), (19, 3, None), (19, 18, None)]
sage: D.edges(sort=True, labels=False)
[(0, 1), (0, 10), (0, 19), (1, 0), (1, 2), (1, 8), (2, 1), (2, 3),
 (2, 6), (3, 2), (3, 4), (3, 19), (4, 3), (4, 5), (4, 17), (5, 4),
 (5, 6), (5, 15), (6, 2), (6, 5), (6, 7), (7, 6), (7, 8), (7, 14),
 (8, 1), (8, 7), (8, 9), (9, 8), (9, 10), (9, 13), (10, 0), (10, 9),
 (10, 11), (11, 10), (11, 12), (11, 18), (12, 11), (12, 13),
 (12, 16), (13, 9), (13, 12), (13, 14), (14, 7), (14, 13), (14, 15),
 (15, 5), (15, 14), (15, 16), (16, 12), (16, 15), (16, 17), (17, 4),
 (17, 16), (17, 18), (18, 11), (18, 17), (18, 19), (19, 0), (19, 3),
 (19, 18)]

The default is to sort the returned list in the default fashion, as in the above examples. This can be overridden by specifying a key function. This first example just ignores the labels in the third component of the triple:

sage: G = graphs.CycleGraph(5)
sage: G.edges(sort=True, key=lambda x: (x[1], -x[0]))
[(0, 1, None), (1, 2, None), (2, 3, None), (3, 4, None), (0, 4, None)]

We set the labels to characters and then perform a default sort followed by a sort according to the labels:

sage: G = graphs.CycleGraph(5)
sage: for e in G.edges(sort=False):
....:   G.set_edge_label(e[0], e[1], chr(ord('A') + e[0] + 5 * e[1]))
sage: G.edges(sort=True)
[(0, 1, 'F'), (0, 4, 'U'), (1, 2, 'L'), (2, 3, 'R'), (3, 4, 'X')]
sage: G.edges(sort=True, key=lambda x: x[2])
[(0, 1, 'F'), (1, 2, 'L'), (2, 3, 'R'), (0, 4, 'U'), (3, 4, 'X')]

We can restrict considered edges to those incident to a given set:

sage: for i in graphs.PetersenGraph().edges(sort=True, vertices=[0]):
....:     print(i)
(0, 1, None)
(0, 4, None)
(0, 5, None)
sage: D = DiGraph({0: [1, 2], 1: [0]})
sage: for i in D.edges(sort=True, vertices=[0]):
....:     print(i)
(0, 1, None)
(0, 2, None)

Ignoring the direction of edges:

sage: D = DiGraph({1: [0], 2: [0]})
sage: D.edges(sort=True, vertices=0)
[]
sage: D.edges(sort=True, vertices=0, ignore_direction=True)
[(1, 0, None), (2, 0, None)]
sage: D.edges(sort=True, vertices=[0], ignore_direction=True)
[(1, 0, None), (2, 0, None)]

Not sorting the ends of the edges:

sage: G = Graph()
sage: G = Graph()
sage: G.add_edges([[1,2], [2,3], [0,3]])
sage: list(G.edge_iterator(sort_vertices=False))
[(3, 0, None), (2, 1, None), (3, 2, None)]
edges_incident(vertices=None, labels=True, sort=False)#

Return incident edges to some vertices.

If vertices is a vertex, then it returns the list of edges incident to that vertex. If vertices is a list of vertices then it returns the list of all edges adjacent to those vertices. If vertices is None, it returns a list of all edges in graph. For digraphs, only lists outward edges.

INPUT:

  • vertices – object (default: None); a vertex, a list of vertices or None

  • labels – boolean (default: True); if False, each edge is

    a tuple \((u,v)\) of vertices

  • sort – boolean (default: False); if True the returned list is sorted

EXAMPLES:

sage: graphs.PetersenGraph().edges_incident([0, 9], labels=False)
[(0, 1), (0, 4), (0, 5), (4, 9), (6, 9), (7, 9)]
sage: D = DiGraph({0: [1]})
sage: D.edges_incident([0])
[(0, 1, None)]
sage: D.edges_incident([1])
[]
eigenspaces(laplacian=False)#

Return the right eigenspaces of the adjacency matrix of the graph.

INPUT:

  • laplacian – boolean (default: False); if True, use the

    Laplacian matrix (see kirchhoff_matrix())

OUTPUT:

A list of pairs. Each pair is an eigenvalue of the adjacency matrix of the graph, followed by the vector space that is the eigenspace for that eigenvalue, when the eigenvectors are placed on the right of the matrix.

For some graphs, some of the eigenspaces are described exactly by vector spaces over a NumberField(). For numerical eigenvectors use eigenvectors().

EXAMPLES:

sage: P = graphs.PetersenGraph()
sage: P.eigenspaces()                                                       # needs sage.modules sage.rings.number_field
[
(3,  Vector space of degree 10 and dimension 1 over Rational Field
     User basis matrix:
     [1 1 1 1 1 1 1 1 1 1]),
(-2, Vector space of degree 10 and dimension 4 over Rational Field
     User basis matrix:
     [ 1  0  0  0 -1 -1 -1  0  1  1]
     [ 0  1  0  0 -1  0 -2 -1  1  2]
     [ 0  0  1  0 -1  1 -1 -2  0  2]
     [ 0  0  0  1 -1  1  0 -1 -1  1]),
(1,  Vector space of degree 10 and dimension 5 over Rational Field
     User basis matrix:
     [ 1  0  0  0  0  1 -1  0  0 -1]
     [ 0  1  0  0  0 -1  1 -1  0  0]
     [ 0  0  1  0  0  0 -1  1 -1  0]
     [ 0  0  0  1  0  0  0 -1  1 -1]
     [ 0  0  0  0  1 -1  0  0 -1  1])
]

Eigenspaces for the Laplacian should be identical since the Petersen graph is regular. However, since the output also contains the eigenvalues, the two outputs are slightly different:

sage: P.eigenspaces(laplacian=True)                                         # needs sage.modules sage.rings.number_field
[
(0, Vector space of degree 10 and dimension 1 over Rational Field
    User basis matrix:
    [1 1 1 1 1 1 1 1 1 1]),
(5, Vector space of degree 10 and dimension 4 over Rational Field
    User basis matrix:
    [ 1  0  0  0 -1 -1 -1  0  1  1]
    [ 0  1  0  0 -1  0 -2 -1  1  2]
    [ 0  0  1  0 -1  1 -1 -2  0  2]
    [ 0  0  0  1 -1  1  0 -1 -1  1]),
(2, Vector space of degree 10 and dimension 5 over Rational Field
    User basis matrix:
    [ 1  0  0  0  0  1 -1  0  0 -1]
    [ 0  1  0  0  0 -1  1 -1  0  0]
    [ 0  0  1  0  0  0 -1  1 -1  0]
    [ 0  0  0  1  0  0  0 -1  1 -1]
    [ 0  0  0  0  1 -1  0  0 -1  1])
]

Notice how one eigenspace below is described with a square root of 2. For the two possible values (positive and negative) there is a corresponding eigenspace:

sage: C = graphs.CycleGraph(8)
sage: C.eigenspaces()                                                       # needs sage.modules sage.rings.number_field
[
(2,  Vector space of degree 8 and dimension 1 over Rational Field
     User basis matrix:
     [1 1 1 1 1 1 1 1]),
(-2, Vector space of degree 8 and dimension 1 over Rational Field
     User basis matrix:
     [ 1 -1  1 -1  1 -1  1 -1]),
(0,  Vector space of degree 8 and dimension 2 over Rational Field
     User basis matrix:
     [ 1  0 -1  0  1  0 -1  0]
     [ 0  1  0 -1  0  1  0 -1]),
(a3, Vector space of degree 8 and dimension 2 over
      Number Field in a3 with defining polynomial x^2 - 2
     User basis matrix:
     [  1   0  -1 -a3  -1   0   1  a3]
     [  0   1  a3   1   0  -1 -a3  -1])
]

A digraph may have complex eigenvalues and eigenvectors. For a 3-cycle, we have:

sage: T = DiGraph({0: [1], 1: [2], 2: [0]})
sage: T.eigenspaces()                                                       # needs sage.modules sage.rings.number_field
[
(1,  Vector space of degree 3 and dimension 1 over Rational Field
     User basis matrix:
     [1 1 1]),
(a1, Vector space of degree 3 and dimension 1 over Number Field in a1
      with defining polynomial x^2 + x + 1
     User basis matrix:
     [      1      a1 -a1 - 1])
]
eigenvectors(laplacian=False)#

Return the right eigenvectors of the adjacency matrix of the graph.

INPUT:

  • laplacian – boolean (default: False); if True, use the Laplacian matrix (see kirchhoff_matrix())

OUTPUT:

A list of triples. Each triple begins with an eigenvalue of the adjacency matrix of the graph. This is followed by a list of eigenvectors for the eigenvalue, when the eigenvectors are placed on the right side of the matrix. Together, the eigenvectors form a basis for the eigenspace. The triple concludes with the algebraic multiplicity of the eigenvalue.

For some graphs, the exact eigenspaces provided by eigenspaces() provide additional insight into the structure of the eigenspaces.

EXAMPLES:

sage: P = graphs.PetersenGraph()
sage: P.eigenvectors()                                                      # needs sage.modules sage.rings.number_field
[(3, [
(1, 1, 1, 1, 1, 1, 1, 1, 1, 1)
], 1), (-2, [
(1, 0, 0, 0, -1, -1, -1, 0, 1, 1),
(0, 1, 0, 0, -1, 0, -2, -1, 1, 2),
(0, 0, 1, 0, -1, 1, -1, -2, 0, 2),
(0, 0, 0, 1, -1, 1, 0, -1, -1, 1)
], 4), (1, [
(1, 0, 0, 0, 0, 1, -1, 0, 0, -1),
(0, 1, 0, 0, 0, -1, 1, -1, 0, 0),
(0, 0, 1, 0, 0, 0, -1, 1, -1, 0),
(0, 0, 0, 1, 0, 0, 0, -1, 1, -1),
(0, 0, 0, 0, 1, -1, 0, 0, -1, 1)
], 5)]

Eigenspaces for the Laplacian should be identical since the Petersen graph is regular. However, since the output also contains the eigenvalues, the two outputs are slightly different:

sage: P.eigenvectors(laplacian=True)                                        # needs sage.modules sage.rings.number_field
[(0, [
(1, 1, 1, 1, 1, 1, 1, 1, 1, 1)
], 1), (5, [
(1, 0, 0, 0, -1, -1, -1, 0, 1, 1),
(0, 1, 0, 0, -1, 0, -2, -1, 1, 2),
(0, 0, 1, 0, -1, 1, -1, -2, 0, 2),
(0, 0, 0, 1, -1, 1, 0, -1, -1, 1)
], 4), (2, [
(1, 0, 0, 0, 0, 1, -1, 0, 0, -1),
(0, 1, 0, 0, 0, -1, 1, -1, 0, 0),
(0, 0, 1, 0, 0, 0, -1, 1, -1, 0),
(0, 0, 0, 1, 0, 0, 0, -1, 1, -1),
(0, 0, 0, 0, 1, -1, 0, 0, -1, 1)
], 5)]
sage: C = graphs.CycleGraph(8)
sage: C.eigenvectors()                                                      # needs sage.modules sage.rings.number_field
[(2,
  [
  (1, 1, 1, 1, 1, 1, 1, 1)
  ],
  1),
 (-2,
  [
  (1, -1, 1, -1, 1, -1, 1, -1)
  ],
  1),
 (0,
  [
  (1, 0, -1, 0, 1, 0, -1, 0),
  (0, 1, 0, -1, 0, 1, 0, -1)
  ],
  2),
 (-1.4142135623...,
  [(1, 0, -1, 1.4142135623..., -1, 0, 1, -1.4142135623...),
   (0, 1, -1.4142135623..., 1, 0, -1, 1.4142135623..., -1)],
  2),
 (1.4142135623...,
  [(1, 0, -1, -1.4142135623..., -1, 0, 1, 1.4142135623...),
   (0, 1, 1.4142135623..., 1, 0, -1, -1.4142135623..., -1)],
  2)]

A digraph may have complex eigenvalues. Previously, the complex parts of graph eigenvalues were being dropped. For a 3-cycle, we have:

sage: T = DiGraph({0:[1], 1:[2], 2:[0]})
sage: T.eigenvectors()                                                      # needs sage.modules sage.rings.number_field
[(1,
  [
  (1, 1, 1)
  ],
  1),
 (-0.5000000000... - 0.8660254037...*I,
  [(1, -0.5000000000... - 0.8660254037...*I, -0.5000000000... + 0.8660254037...*I)],
  1),
 (-0.5000000000... + 0.8660254037...*I,
  [(1, -0.5000000000... + 0.8660254037...*I, -0.5000000000... - 0.8660254037...*I)],
  1)]
eulerian_circuit(return_vertices=False, labels=True, path=False)#

Return a list of edges forming an Eulerian circuit if one exists.

If no Eulerian circuit is found, the method returns False.

This is implemented using Hierholzer’s algorithm.

INPUT:

  • return_vertices – boolean (default: False); optionally

    provide a list of vertices for the path

  • labels – boolean (default: True); whether to return edges

    with labels (3-tuples)

  • path – boolean (default: False); find an Eulerian path

    instead

OUTPUT:

either ([edges], [vertices]) or [edges] of an Eulerian circuit (or path)

EXAMPLES:

sage: g = graphs.CycleGraph(5)
sage: g.eulerian_circuit()
[(0, 4, None), (4, 3, None), (3, 2, None), (2, 1, None), (1, 0, None)]
sage: g.eulerian_circuit(labels=False)
[(0, 4), (4, 3), (3, 2), (2, 1), (1, 0)]
sage: g = graphs.CompleteGraph(7)
sage: edges, vertices = g.eulerian_circuit(return_vertices=True)
sage: vertices
[0, 6, 5, 4, 6, 3, 5, 2, 4, 3, 2, 6, 1, 5, 0, 4, 1, 3, 0, 2, 1, 0]
sage: graphs.CompleteGraph(4).eulerian_circuit()
False

A disconnected graph can be Eulerian:

sage: g = Graph({0: [], 1: [2], 2: [3], 3: [1], 4: []})
sage: g.eulerian_circuit(labels=False)
[(1, 3), (3, 2), (2, 1)]
sage: g = DiGraph({0: [1], 1: [2, 4], 2:[3], 3:[1]})
sage: g.eulerian_circuit(labels=False, path=True)
[(0, 1), (1, 2), (2, 3), (3, 1), (1, 4)]
sage: g = Graph({0:[1,2,3], 1:[2,3], 2:[3,4], 3:[4]})
sage: g.is_eulerian(path=True)
(0, 1)
sage: g.eulerian_circuit(labels=False, path=True)
[(1, 3), (3, 4), (4, 2), (2, 3), (3, 0), (0, 2), (2, 1), (1, 0)]
eulerian_orientation()#

Return a DiGraph which is an Eulerian orientation of the current graph.

An Eulerian graph being a graph such that any vertex has an even degree, an Eulerian orientation of a graph is an orientation of its edges such that each vertex \(v\) verifies \(d^+(v)=d^-(v)=d(v)/2\), where \(d^+\) and \(d^-\) respectively represent the out-degree and the in-degree of a vertex.

If the graph is not Eulerian, the orientation verifies for any vertex \(v\) that \(| d^+(v)-d^-(v) | \leq 1\).

ALGORITHM:

This algorithm is a random walk through the edges of the graph, which orients the edges according to the walk. When a vertex is reached which has no non-oriented edge (this vertex must have odd degree), the walk resumes at another vertex of odd degree, if any.

This algorithm has complexity \(O(m)\), where \(m\) is the number of edges in the graph.

EXAMPLES:

The CubeGraph with parameter 4, which is regular of even degree, has an Eulerian orientation such that \(d^+ = d^-\):

sage: g = graphs.CubeGraph(4)
sage: g.degree()
[4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4]
sage: o = g.eulerian_orientation()
sage: o.in_degree()
[2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2]
sage: o.out_degree()
[2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2]

Secondly, the Petersen Graph, which is 3 regular has an orientation such that the difference between \(d^+\) and \(d^-\) is at most 1:

sage: g = graphs.PetersenGraph()
sage: o = g.eulerian_orientation()
sage: o.in_degree()
[2, 2, 2, 2, 2, 1, 1, 1, 1, 1]
sage: o.out_degree()
[1, 1, 1, 1, 1, 2, 2, 2, 2, 2]
export_to_file(filename, format=None, **kwds)#

Export the graph to a file.

INPUT:

See also

  • save() – save a Sage object to a ‘sobj’ file (preserves all its attributes)

Note

This functions uses the write_* functions defined in NetworkX (see http://networkx.lanl.gov/reference/readwrite.html).

EXAMPLES:

sage: g = graphs.PetersenGraph()
sage: filename = tmp_filename(ext=".pajek")
sage: g.export_to_file(filename)                                            # needs networkx
sage: import networkx                                                       # needs networkx
sage: G_networkx = networkx.read_pajek(filename)                            # needs networkx
sage: Graph(G_networkx).is_isomorphic(g)                                    # needs networkx
True
sage: filename = tmp_filename(ext=".edgelist")
sage: g.export_to_file(filename, data=False)                                # needs networkx
sage: h = Graph(networkx.read_edgelist(filename))                           # needs networkx
sage: g.is_isomorphic(h)                                                    # needs networkx
True
faces(embedding=None)#

Return the faces of an embedded graph.

A combinatorial embedding of a graph is a clockwise ordering of the neighbors of each vertex. From this information one can define the faces of the embedding, which is what this method returns.

If no embedding is provided or stored as self._embedding, this method will compute the set of faces from the embedding returned by is_planar() (if the graph is, of course, planar).

Warning

This method is not well defined when the graph is not connected. Indeed, the result may contain several faces corresponding to the external face.

INPUT:

  • embedding – dictionary (default: None); a combinatorial embedding dictionary. Format: {v1: [v2,v3], v2: [v1], v3: [v1]} (clockwise ordering of neighbors at each vertex). If set to None (default) the method will use the embedding stored as self._embedding. If none is stored, the method will compute the set of faces from the embedding returned by is_planar() (if the graph is, of course, planar).

Note

embedding is an ordered list based on the hash order of the vertices of graph. To avoid confusion, it might be best to set the rot_sys based on a ‘nice_copy’ of the graph.

EXAMPLES:

Providing an embedding:

sage: T = graphs.TetrahedralGraph()
sage: T.faces({0: [1, 3, 2], 1: [0, 2, 3], 2: [0, 3, 1], 3: [0, 1, 2]})
[[(0, 1), (1, 2), (2, 0)],
 [(0, 2), (2, 3), (3, 0)],
 [(0, 3), (3, 1), (1, 0)],
 [(1, 3), (3, 2), (2, 1)]]

With no embedding provided:

sage: graphs.TetrahedralGraph().faces()
[[(0, 1), (1, 2), (2, 0)],
 [(0, 2), (2, 3), (3, 0)],
 [(0, 3), (3, 1), (1, 0)],
 [(1, 3), (3, 2), (2, 1)]]

With no embedding provided (non-planar graph):

sage: graphs.PetersenGraph().faces()
Traceback (most recent call last):
...
ValueError: no embedding is provided and the graph is not planar
feedback_vertex_set(value_only, solver=False, verbose=None, constraint_generation=0, integrality_tolerance=True)#

Return the minimum feedback vertex set of a (di)graph.

The minimum feedback vertex set of a (di)graph is a set of vertices that intersect all of its cycles. Equivalently, a minimum feedback vertex set of a (di)graph is a set \(S\) of vertices such that the digraph \(G-S\) is acyclic. For more information, see the Wikipedia article Feedback_vertex_set.

INPUT:

  • value_only – boolean (default: False); whether to return only the minimum cardinal of a minimum vertex set, or the Set of vertices of a minimal feedback vertex set

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

  • constraint_generation – boolean (default: True); whether to use constraint generation when solving the Mixed Integer Linear Program

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

ALGORITHMS:

(Constraints generation)

When the parameter constraint_generation is enabled (default) the following MILP formulation is used to solve the problem:

\[\begin{split}\mbox{Minimize : }&\sum_{v\in G} b_{v}\\ \mbox{Such that : }&\\ &\forall C\text{ circuits }\subseteq G, \sum_{v\in C}b_{v}\geq 1\\\end{split}\]

As the number of circuits contained in a graph is exponential, this LP is solved through constraint generation. This means that the solver is sequentially asked to solve the problem, knowing only a portion of the circuits contained in \(G\), each time adding to the list of its constraints the circuit which its last answer had left intact.

(Another formulation based on an ordering of the vertices)

When the graph is directed, a second (and very slow) formulation is available, which should only be used to check the result of the first implementation in case of doubt.

\[\begin{split}\mbox{Minimize : }&\sum_{v\in G} b_v\\ \mbox{Such that : }&\\ &\forall (u,v)\in G, d_u-d_v+nb_u+nb_v\geq 0\\ &\forall u\in G, 0\leq d_u\leq |G|\\\end{split}\]

A brief explanation:

An acyclic digraph can be seen as a poset, and every poset has a linear extension. This means that in any acyclic digraph the vertices can be ordered with a total order \(<\) in such a way that if \((u,v)\in G\), then \(u<v\). Thus, this linear program is built in order to assign to each vertex \(v\) a number \(d_v\in [0,\dots,n-1]\) such that if there exists an edge \((u,v)\in G\) then either \(d_v<d_u\) or one of \(u\) or \(v\) is removed. The number of vertices removed is then minimized, which is the objective.

EXAMPLES:

The necessary example:

sage: # needs sage.numerical.mip
sage: g = graphs.PetersenGraph()
sage: fvs = g.feedback_vertex_set()
sage: len(fvs)
3
sage: g.delete_vertices(fvs)
sage: g.is_forest()
True

In a digraph built from a graph, any edge is replaced by arcs going in the two opposite directions, thus creating a cycle of length two. Hence, to remove all the cycles from the graph, each edge must see one of its neighbors removed: a feedback vertex set is in this situation a vertex cover:

sage: # needs sage.numerical.mip
sage: cycle = graphs.CycleGraph(5)
sage: dcycle = DiGraph(cycle)
sage: cycle.vertex_cover(value_only=True)
3
sage: feedback = dcycle.feedback_vertex_set()
sage: len(feedback)
3
sage: u,v = next(cycle.edge_iterator(labels=None))
sage: u in feedback or v in feedback
True

For a circuit, the minimum feedback arc set is clearly \(1\):

sage: circuit = digraphs.Circuit(5)
sage: circuit.feedback_vertex_set(value_only=True) == 1                     # needs sage.numerical.mip
True
flow(x, y, value_only, integer=True, use_edge_labels=False, vertex_bound=True, algorithm=False, solver=None, verbose=None, integrality_tolerance=0)#

Return a maximum flow in the graph from x to y.

The returned flow is represented by an optimal valuation of the edges. For more information, see the Wikipedia article Max_flow.

As an optimization problem, is can be expressed this way :

\[\begin{split}\mbox{Maximize : }&\sum_{e\in G.edges()} w_e b_e\\ \mbox{Such that : }&\forall v \in G, \sum_{(u,v)\in G.edges()} b_{(u,v)}\leq 1\\ &\forall x\in G, b_x\mbox{ is a binary variable}\end{split}\]

Observe that the integrality of the flow variables is automatic for all available solvers when all capacities are integers.

INPUT:

  • x – source vertex

  • y – sink vertex

  • value_only – boolean (default: True); whether to return only the value of a maximal flow, or to also return a flow graph (a copy of the current graph, such that each edge has the flow using it as a label, the edges without flow being omitted)

  • integer – boolean (default: True); whether to compute an optimal solution under the constraint that the flow going through an edge has to be an integer, or without this constraint

  • use_edge_labels – boolean (default: False); whether to compute a maximum flow where each edge has a capacity defined by its label (if an edge has no label, capacity \(1\) is assumed), or to use default edge capacity of \(1\)

  • vertex_bound – boolean (default: False); when set to True, sets the maximum flow leaving a vertex different from \(x\) to \(1\) (useful for vertex connectivity parameters)

  • algorithm – string (default: None); the algorithm to use among:

    • "FF", a Python implementation of the Ford-Fulkerson algorithm (only available when vertex_bound = False)

    • "LP", the flow problem is solved using Linear Programming

    • "igraph", the igraph implementation of the Goldberg-Tarjan algorithm is used (only available when igraph is installed and vertex_bound = False)

    When algorithm = None (default), we use LP if vertex_bound = True, otherwise, we use igraph if it is available, FF if it is not available.

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

    Only useful when algorithm "LP" is used to solve the flow problem.

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

    Only useful when algorithm "LP" is used to solve the flow problem.

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

    Only useful when algorithm == "LP" and integer == True.

Note

Even though the three different implementations are meant to return the same Flow values, they cannot be expected to return the same Flow graphs.

Besides, the use of Linear Programming may possibly mean a (slight) numerical noise.

EXAMPLES:

Two basic applications of the flow method for the PappusGraph and the ButterflyGraph with parameter \(2\)

sage: g=graphs.PappusGraph()
sage: int(g.flow(1,2))
3
sage: b=digraphs.ButterflyGraph(2)
sage: int(b.flow(('00', 1), ('00', 2)))
1

The flow method can be used to compute a matching in a bipartite graph by linking a source \(s\) to all the vertices of the first set and linking a sink \(t\) to all the vertices of the second set, then computing a maximum \(s-t\) flow

sage: g = DiGraph()
sage: g.add_edges(('s', i) for i in range(4))
sage: g.add_edges((i, 4 + j) for i in range(4) for j in range(4))
sage: g.add_edges((4 + i, 't') for i in range(4))
sage: [cardinal, flow_graph] = g.flow('s', 't', integer=True, value_only=False)
sage: flow_graph.delete_vertices(['s', 't'])
sage: flow_graph.size()
4

The undirected case:

sage: g = Graph()
sage: g.add_edges(('s', i) for i in range(4))
sage: g.add_edges((i, 4 + j) for i in range(4) for j in range(4))
sage: g.add_edges((4 + i, 't') for i in range(4))
sage: [cardinal, flow_graph] = g.flow('s', 't', integer=True, value_only=False)
sage: flow_graph.delete_vertices(['s', 't'])
sage: flow_graph.size()
4
genus(set_embedding=True, on_embedding=None, minimal=True, maximal=False, circular=None, ordered=True)#

Return the minimal genus of the graph.

The genus of a compact surface is the number of handles it has. The genus of a graph is the minimal genus of the surface it can be embedded into. It can be seen as a measure of non-planarity; a planar graph has genus zero.

Note

This function uses Euler’s formula and thus it is necessary to consider only connected graphs.

INPUT:

  • set_embedding – boolean (default: True); whether or not to store an embedding attribute of the computed (minimal) genus of the graph

  • on_embedding – two kinds of input are allowed (default:

    None):

    • a dictionary representing a combinatorial embedding on which the genus should be computed. Note that this must be a valid embedding for the graph. The dictionary structure is given by: vertex1: [neighbor1, neighbor2, neighbor3], vertex2: [neighbor] where there is a key for each vertex in the graph and a (clockwise) ordered list of each vertex’s neighbors as values. The value of on_embedding takes precedence over a stored _embedding attribute if minimal is set to False.

    • The value True, in order to indicate that the embedding stored as _embedding should be used (see examples).

  • minimal – boolean (default: True); whether or not to compute the minimal genus of the graph (i.e., testing all embeddings). If minimal is False, then either maximal must be True or on_embedding must not be None. If on_embedding is not None, it will take priority over minimal. Similarly, if maximal is True, it will take priority over minimal.

  • maximal – boolean (default: False); whether or not to compute the maximal genus of the graph (i.e., testing all embeddings). If maximal is False, then either minimal must be True or on_embedding must not be None. If on_embedding is not None, it will take priority over maximal. However, maximal takes priority over the default minimal.

  • circular – list (default: None); if circular is a list of vertices, the method computes the genus preserving a planar embedding of the this list. If circular is defined, on_embedding is not a valid option.

  • ordered – boolean (default: True); if circular is True, then whether or not the boundary order may be permuted (default is True, which means the boundary order is preserved)

EXAMPLES:

sage: g = graphs.PetersenGraph()
sage: g.genus() # tests for minimal genus by default
1
sage: g.genus(on_embedding=True, maximal=True) # on_embedding overrides minimal and maximal arguments
1
sage: g.genus(maximal=True) # setting maximal to True overrides default minimal=True
3
sage: g.genus(on_embedding=g.get_embedding()) # can also send a valid combinatorial embedding dict
3
sage: (graphs.CubeGraph(3)).genus()
0
sage: K23 = graphs.CompleteBipartiteGraph(2,3)
sage: K23.genus()
0
sage: K33 = graphs.CompleteBipartiteGraph(3,3)
sage: K33.genus()
1

Using the circular argument, we can compute the minimal genus preserving a planar, ordered boundary:

sage: cube = graphs.CubeGraph(2)
sage: cube.genus(circular=['01','10'])
0
sage: cube.is_circular_planar()
True
sage: cube.genus(circular=['01','10'])
0
sage: cube.genus(circular=['01','10'], on_embedding=True)
Traceback (most recent call last):
...
ValueError: on_embedding is not a valid option when circular is defined
sage: cube.genus(circular=['01','10'], maximal=True)
Traceback (most recent call last):
...
NotImplementedError: cannot compute the maximal genus of a genus respecting a boundary

Note: not everything works for multigraphs, looped graphs or digraphs. But the minimal genus is ultimately computable for every connected graph – but the embedding we obtain for the simple graph can’t be easily converted to an embedding of a non-simple graph. Also, the maximal genus of a multigraph does not trivially correspond to that of its simple graph:

sage: G = DiGraph({0: [0, 1, 1, 1], 1: [2, 2, 3, 3], 2: [1, 3, 3], 3: [0, 3]})
sage: G.genus()
Traceback (most recent call last):
...
NotImplementedError: cannot work with embeddings of non-simple graphs
sage: G.to_simple().genus()
0
sage: G.genus(set_embedding=False)
0
sage: G.genus(maximal=True, set_embedding=False)
Traceback (most recent call last):
...
NotImplementedError: cannot compute the maximal genus of a graph with loops or multiple edges

We break graphs with cut vertices into their blocks, which greatly speeds up computation of minimal genus. This is not implemented for maximal genus:

sage: G = graphs.RandomBlockGraph(10, 5)
sage: G.genus()
10
get_embedding()#

Return the stored embedding or None.

If the stored embedding is no longer valid (because of vertex/edge additions) then the stored embedding is discarded and None is returned. In case some vertex/edge has been deleted, the stored embedding is updated accordingly.

EXAMPLES:

sage: G = graphs.PetersenGraph()
sage: G.genus()
1
sage: G.get_embedding()
{0: [1, 4, 5], 1: [0, 2, 6], 2: [1, 3, 7], 3: [2, 4, 8],
 4: [0, 3, 9], 5: [0, 7, 8], 6: [1, 9, 8], 7: [2, 5, 9],
 8: [3, 6, 5], 9: [4, 6, 7]}

Note that the embeddings gets properly modified on vertex or edge deletion:

sage: G.delete_edge(0, 1)
sage: G.delete_vertex(3)
sage: G.get_embedding()
{0: [4, 5],
 1: [2, 6],
 2: [1, 7],
 4: [0, 9],
 5: [0, 7, 8],
 6: [1, 9, 8],
 7: [2, 5, 9],
 8: [6, 5],
 9: [4, 6, 7]}

But not under edge addition:

sage: G.add_edge(0, 7)
sage: G.get_embedding() is None
True
get_pos(dim=2)#

Return the position dictionary.

The position dictionary specifies the coordinates of each vertex.

INPUT:

  • dim – integer (default: 2); whether to return the position dictionary in the plane (dim == 2) or in the 3-dimensional space

EXAMPLES:

By default, the position of a graph is None:

sage: G = Graph()
sage: G.get_pos()
sage: G.get_pos() is None
True
sage: P = G.plot(save_pos=True)                                             # needs sage.plot
sage: G.get_pos()                                                           # needs sage.plot
{}

Some of the named graphs come with a pre-specified positioning:

sage: G = graphs.PetersenGraph()
sage: G.get_pos()
{0: (0.0, 1.0),
 ...
 9: (0.475..., 0.154...)}

Note that the position dictionary is modified on vertex removal:

sage: G.delete_vertex(0)
sage: G.get_pos()
{1: (-0.951..., 0.309...),
...
 9: (0.475..., 0.154...)}

But is deleted on vertex addition:

sage: G.add_vertex(0)
sage: G.get_pos() is None
True
get_vertex(vertex)#

Retrieve the object associated with a given vertex.

If no associated object is found, None is returned.

INPUT:

  • vertex – the given vertex

EXAMPLES:

sage: d = {0: graphs.DodecahedralGraph(), 1: graphs.FlowerSnark(), 2: graphs.MoebiusKantorGraph(), 3: graphs.PetersenGraph()}
sage: d[2]
Moebius-Kantor Graph: Graph on 16 vertices
sage: T = graphs.TetrahedralGraph()
sage: T.vertices(sort=True)
[0, 1, 2, 3]
sage: T.set_vertices(d)
sage: T.get_vertex(1)
Flower Snark: Graph on 20 vertices
get_vertices(verts=None)#

Return a dictionary of the objects associated to each vertex.

INPUT:

  • verts – iterable container of vertices

EXAMPLES:

sage: d = {0: graphs.DodecahedralGraph(), 1: graphs.FlowerSnark(), 2: graphs.MoebiusKantorGraph(), 3: graphs.PetersenGraph()}
sage: T = graphs.TetrahedralGraph()
sage: T.set_vertices(d)
sage: T.get_vertices([1, 2])
{1: Flower Snark: Graph on 20 vertices,
 2: Moebius-Kantor Graph: Graph on 16 vertices}
girth(certificate=False)#

Return the girth of the graph.

The girth is the length of the shortest cycle in the graph (directed cycle if the graph is directed). Graphs without (directed) cycles have infinite girth.

INPUT:

  • certificate – boolean (default: False); whether to return (g, c), where g is the girth and c is a list of vertices of a (directed) cycle of length g in the graph, thus providing a certificate that the girth is at most g, or None if g infinite

EXAMPLES:

sage: graphs.TetrahedralGraph().girth()
3
sage: graphs.CubeGraph(3).girth()
4
sage: graphs.PetersenGraph().girth(certificate=True)  # random
(5, [4, 3, 2, 1, 0])
sage: graphs.HeawoodGraph().girth()
6
sage: next(graphs.trees(9)).girth()
+Infinity

See also

graphics_array_defaults = {'graph_border': True, 'layout': 'circular', 'vertex_labels': False, 'vertex_size': 50}#
graphplot(**options)#

Return a GraphPlot object.

See GraphPlot for more details.

INPUT:

  • **options – parameters for the GraphPlot constructor

EXAMPLES:

Creating a GraphPlot object uses the same options as plot():

sage: g = Graph({}, loops=True, multiedges=True, sparse=True)
sage: g.add_edges([(0,0,'a'),(0,0,'b'),(0,1,'c'),(0,1,'d'),
....:     (0,1,'e'),(0,1,'f'),(0,1,'f'),(2,1,'g'),(2,2,'h')])
sage: GP = g.graphplot(edge_labels=True, color_by_label=True,               # needs sage.plot
....:                  edge_style='dashed')
sage: GP.plot()                                                             # needs sage.plot
Graphics object consisting of 22 graphics primitives

We can modify the GraphPlot object. Notice that the changes are cumulative:

sage: # needs sage.plot
sage: GP.set_edges(edge_style='solid')
sage: GP.plot()
Graphics object consisting of 22 graphics primitives
sage: GP.set_vertices(talk=True)
sage: GP.plot()
Graphics object consisting of 22 graphics primitives
graphviz_string(labels='string', vertex_labels=True, edge_labels=False, edge_color=None, edge_colors=None, edge_options=(), color_by_label=False, rankdir='down', subgraph_clusters=[], **options)#

Return a representation in the dot language.

The dot language is a text based format for graphs. It is used by the software suite graphviz. The specifications of the language are available on the web (see the reference [dotspec]).

INPUT:

  • labels – string (default: "string"); either "string" or "latex". If labels is "string", latex commands are not interpreted. This option stands for both vertex labels and edge labels.

  • vertex_labels – boolean (default: True); whether to add the labels on vertices

  • edge_labels – boolean (default: False); whether to add the labels on edges

  • edge_color – (default: None); specify a default color for the edges. The color could be one of

    • a name given as a string such as "blue" or "orchid"

    • a HSV sequence in a string such as ".52,.386,.22"

    • an hexadecimal code such as "#DA3305"

    • a 3-tuple of floating point (to be interpreted as RGB tuple). In this case the 3-tuple is converted in hexadecimal code.

  • edge_colors – dictionary (default: None); a dictionary whose keys are colors and values are list of edges. The list of edges need not to be complete in which case the default color is used. See the option edge_color for a description of valid color formats.

  • color_by_label – a boolean or dictionary or function (default: False); whether to color each edge with a different color according to its label; the colors are chosen along a rainbow, unless they are specified by a function or dictionary mapping labels to colors; this option is incompatible with edge_color and edge_colors. See the option edge_color for a description of valid color formats.

  • edge_options – a function (or tuple thereof) mapping edges to a dictionary of options for this edge

  • rankdir'left', 'right', 'up', or 'down' (default: 'down', for consistency with graphviz): the preferred ranking direction for acyclic layouts; see the rankdir option of graphviz.

  • subgraph_clusters – a list of lists of vertices (default: []); From [dotspec]: “If supported, the layout engine will do the layout so that the nodes belonging to the cluster are drawn together, with the entire drawing of the cluster contained within a bounding rectangle. Note that, for good and bad, cluster subgraphs are not part of the dot language, but solely a syntactic convention adhered to by certain of the layout engines.”

EXAMPLES:

sage: G = Graph({0: {1: None, 2: None}, 1: {0: None, 2: None},
....:            2: {0: None, 1: None, 3: 'foo'}, 3: {2: 'foo'}},
....:           sparse=True)
sage: print(G.graphviz_string(edge_labels=True))
graph {
  node_0  [label="0"];
  node_1  [label="1"];
  node_2  [label="2"];
  node_3  [label="3"];

  node_0 -- node_1;
  node_0 -- node_2;
  node_1 -- node_2;
  node_2 -- node_3 [label="foo"];
}

A variant, with the labels in latex, for post-processing with dot2tex:

sage: print(G.graphviz_string(edge_labels=True, labels="latex"))
graph {
  node [shape="plaintext"];
  node_0  [label=" ", texlbl="$0$"];
  node_1  [label=" ", texlbl="$1$"];
  node_2  [label=" ", texlbl="$2$"];
  node_3  [label=" ", texlbl="$3$"];

  node_0 -- node_1;
  node_0 -- node_2;
  node_1 -- node_2;
  node_2 -- node_3 [label=" ", texlbl="$\text{\texttt{foo}}$"];
}

Same, with a digraph and a color for edges:

sage: G = DiGraph({0: {1: None, 2: None}, 1: {2: None}, 2: {3: 'foo'}, 3: {}},
....:             sparse=True)
sage: print(G.graphviz_string(edge_color="red"))
digraph {
  node_0  [label="0"];
  node_1  [label="1"];
  node_2  [label="2"];
  node_3  [label="3"];

edge [color="red"];
  node_0 -> node_1;
  node_0 -> node_2;
  node_1 -> node_2;
  node_2 -> node_3;
}

A digraph using latex labels for vertices and edges:

sage: # needs sage.symbolic
sage: f(x) = -1 / x
sage: g(x) = 1 / (x + 1)
sage: G = DiGraph()
sage: G.add_edges((i, f(i), f) for i in (1, 2, 1/2, 1/4))
sage: G.add_edges((i, g(i), g) for i in (1, 2, 1/2, 1/4))
sage: print(G.graphviz_string(labels="latex",               # random
....:                         edge_labels=True))
digraph {
  node [shape="plaintext"];
  node_10  [label=" ", texlbl="$1$"];
  node_11  [label=" ", texlbl="$2$"];
  node_3  [label=" ", texlbl="$-\frac{1}{2}$"];
  node_6  [label=" ", texlbl="$\frac{1}{2}$"];
  node_7  [label=" ", texlbl="$\frac{1}{2}$"];
  node_5  [label=" ", texlbl="$\frac{1}{3}$"];
  node_8  [label=" ", texlbl="$\frac{2}{3}$"];
  node_4  [label=" ", texlbl="$\frac{1}{4}$"];
  node_1  [label=" ", texlbl="$-2$"];
  node_9  [label=" ", texlbl="$\frac{4}{5}$"];
  node_0  [label=" ", texlbl="$-4$"];
  node_2  [label=" ", texlbl="$-1$"];

  node_10 -> node_2 [label=" ", texlbl="$x \ {\mapsto}\ -\frac{1}{x}$"];
  node_10 -> node_6 [label=" ", texlbl="$x \ {\mapsto}\ \frac{1}{x + 1}$"];
  node_11 -> node_3 [label=" ", texlbl="$x \ {\mapsto}\ -\frac{1}{x}$"];
  node_11 -> node_5 [label=" ", texlbl="$x \ {\mapsto}\ \frac{1}{x + 1}$"];
  node_7 -> node_1 [label=" ", texlbl="$x \ {\mapsto}\ -\frac{1}{x}$"];
  node_7 -> node_8 [label=" ", texlbl="$x \ {\mapsto}\ \frac{1}{x + 1}$"];
  node_4 -> node_0 [label=" ", texlbl="$x \ {\mapsto}\ -\frac{1}{x}$"];
  node_4 -> node_9 [label=" ", texlbl="$x \ {\mapsto}\ \frac{1}{x + 1}$"];
}

sage: print(G.graphviz_string(labels="latex",               # random        # needs sage.symbolic
....:                         color_by_label=True))
digraph {
  node [shape="plaintext"];
  node_10  [label=" ", texlbl="$1$"];
  node_11  [label=" ", texlbl="$2$"];
  node_3  [label=" ", texlbl="$-\frac{1}{2}$"];
  node_6  [label=" ", texlbl="$\frac{1}{2}$"];
  node_7  [label=" ", texlbl="$\frac{1}{2}$"];
  node_5  [label=" ", texlbl="$\frac{1}{3}$"];
  node_8  [label=" ", texlbl="$\frac{2}{3}$"];
  node_4  [label=" ", texlbl="$\frac{1}{4}$"];
  node_1  [label=" ", texlbl="$-2$"];
  node_9  [label=" ", texlbl="$\frac{4}{5}$"];
  node_0  [label=" ", texlbl="$-4$"];
  node_2  [label=" ", texlbl="$-1$"];

  node_10 -> node_2 [color = "#ff0000"];
  node_10 -> node_6 [color = "#00ffff"];
  node_11 -> node_3 [color = "#ff0000"];
  node_11 -> node_5 [color = "#00ffff"];
  node_7 -> node_1 [color = "#ff0000"];
  node_7 -> node_8 [color = "#00ffff"];
  node_4 -> node_0 [color = "#ff0000"];
  node_4 -> node_9 [color = "#00ffff"];
}

sage: print(G.graphviz_string(labels="latex",               # random        # needs sage.symbolic
....:                         color_by_label={f: "red", g: "blue"}))
digraph {
  node [shape="plaintext"];
  node_10  [label=" ", texlbl="$1$"];
  node_11  [label=" ", texlbl="$2$"];
  node_3  [label=" ", texlbl="$-\frac{1}{2}$"];
  node_6  [label=" ", texlbl="$\frac{1}{2}$"];
  node_7  [label=" ", texlbl="$\frac{1}{2}$"];
  node_5  [label=" ", texlbl="$\frac{1}{3}$"];
  node_8  [label=" ", texlbl="$\frac{2}{3}$"];
  node_4  [label=" ", texlbl="$\frac{1}{4}$"];
  node_1  [label=" ", texlbl="$-2$"];
  node_9  [label=" ", texlbl="$\frac{4}{5}$"];
  node_0  [label=" ", texlbl="$-4$"];
  node_2  [label=" ", texlbl="$-1$"];

  node_10 -> node_2 [color = "red"];
  node_10 -> node_6 [color = "blue"];
  node_11 -> node_3 [color = "red"];
  node_11 -> node_5 [color = "blue"];
  node_7 -> node_1 [color = "red"];
  node_7 -> node_8 [color = "blue"];
  node_4 -> node_0 [color = "red"];
  node_4 -> node_9 [color = "blue"];
}

By default graphviz renders digraphs using a hierarchical layout, ranking the vertices down from top to bottom. Here we specify alternative ranking directions for this layout:

sage: D = DiGraph([(1, 2)])
sage: print(D.graphviz_string(rankdir="up"))
digraph {
  rankdir=BT
  node_0  [label="1"];
  node_1  [label="2"];

  node_0 -> node_1;
}
sage: print(D.graphviz_string(rankdir="down"))
digraph {
  node_0  [label="1"];
  node_1  [label="2"];

  node_0 -> node_1;
}
sage: print(D.graphviz_string(rankdir="left"))
digraph {
  rankdir=RL
  node_0  [label="1"];
  node_1  [label="2"];

  node_0 -> node_1;
}
sage: print(D.graphviz_string(rankdir="right"))
digraph {
  rankdir=LR
  node_0  [label="1"];
  node_1  [label="2"];

  node_0 -> node_1;
}

Edge-specific options can also be specified by providing a function (or tuple thereof) which maps each edge to a dictionary of options. Valid options are

  • "color"

  • "dot" (a string containing a sequence of options in dot format)

  • "label" (a string)

  • "label_style" ("string" or "latex")

  • "edge_string" ("--" or "->")

  • "dir" ("forward", "back", "both" or "none")

  • "backward" (boolean), instead of defining the edge in the graphviz string as u -> v it draws it as v -> u [dir=back] and instead of u -> v [dir=back] it draws it as v -> u, this changes the way it is drawn by Graphviz’s dot program: vertex v will be above vertex u instead of below.

Here we state that the graph should be laid out so that edges starting from 1 are going backward (e.g. going up instead of down):

sage: def edge_options(data):
....:     u, v, label = data
....:     return {"dir":"back"} if u == 1 else {}
sage: print(G.graphviz_string(edge_options=edge_options))   # random        # needs sage.symbolic
digraph {
  node_0  [label="-1"];
  node_1  [label="-1/2"];
  node_2  [label="1/2"];
  node_3  [label="-2"];
  node_4  [label="1/4"];
  node_5  [label="-4"];
  node_6  [label="1/3"];
  node_7  [label="2/3"];
  node_8  [label="4/5"];
  node_9  [label="1"];
  node_10  [label="2"];

  node_2 -> node_3;
  node_2 -> node_7;
  node_4 -> node_5;
  node_4 -> node_8;
  node_9 -> node_0 [dir=back];
  node_9 -> node_2 [dir=back];
  node_10 -> node_1;
  node_10 -> node_6;
}

We now test all options:

sage: def edge_options(data):
....:     u, v, label = data
....:     options = {"color": {f: "red", g: "blue"}[label]}
....:     if (u,v) == (1/2, -2): options["label"]       = "coucou"; options["label_style"] = "string"
....:     if (u,v) == (1/2,2/3): options["dot"]         = "x=1,y=2"
....:     if (u,v) == (1,   -1): options["label_style"] = "latex"
....:     if (u,v) == (1,  1/2): options["dir"]         = "back"
....:     return options
sage: print(G.graphviz_string(edge_options=edge_options))   # random        # needs sage.symbolic
digraph {
  node_0  [label="-1"];
  node_1  [label="-1/2"];
  node_2  [label="1/2"];
  node_3  [label="-2"];
  node_4  [label="1/4"];
  node_5  [label="-4"];
  node_6  [label="1/3"];
  node_7  [label="2/3"];
  node_8  [label="4/5"];
  node_9  [label="1"];
  node_10  [label="2"];

  node_2 -> node_3 [label="coucou", color = "red"];
  node_2 -> node_7 [x=1,y=2, color = "blue"];
  node_4 -> node_5 [color = "red"];
  node_4 -> node_8 [color = "blue"];
  node_9 -> node_0 [label=" ", texlbl="$x \ {\mapsto}\ -\frac{1}{x}$", color = "red"];
  node_9 -> node_2 [color = "blue", dir=back];
  node_10 -> node_1 [color = "red"];
  node_10 -> node_6 [color = "blue"];
}

We test the possible values of the 'dir' edge option:

sage: edges = [(0,1,'a'), (1,2,'b'), (2,3,'c'), (3,4,'d')]
sage: G = DiGraph(edges)
sage: def edge_options(data):
....:     u,v,label = data
....:     if label == 'a': return {'dir':'forward'}
....:     if label == 'b': return {'dir':'back'}
....:     if label == 'c': return {'dir':'none'}
....:     if label == 'd': return {'dir':'both'}
sage: print(G.graphviz_string(edge_options=edge_options))
digraph {
  node_0  [label="0"];
  node_1  [label="1"];
  node_2  [label="2"];
  node_3  [label="3"];
  node_4  [label="4"];

  node_0 -> node_1;
  node_1 -> node_2 [dir=back];
  node_2 -> node_3 [dir=none];
  node_3 -> node_4 [dir=both];
}

We test the same graph and 'dir' edge options but with backward=True, which reverses the natural direction each edge wants to be pointing for the layout:

sage: def edge_options(data):
....:     u,v,label = data
....:     if label == 'a': return {'dir':'forward', 'backward':True}
....:     if label == 'b': return {'dir':'back', 'backward':True}
....:     if label == 'c': return {'dir':'none', 'backward':True}
....:     if label == 'd': return {'dir':'both', 'backward':True}
sage: print(G.graphviz_string(edge_options=edge_options))
digraph {
  node_0  [label="0"];
  node_1  [label="1"];
  node_2  [label="2"];
  node_3  [label="3"];
  node_4  [label="4"];

  node_1 -> node_0 [dir=back];
  node_2 -> node_1;
  node_3 -> node_2 [dir=none];
  node_4 -> node_3 [dir=both];
}
graphviz_to_file_named(filename, **options)#

Write a representation in the dot language in a file.

The dot language is a plaintext format for graph structures. See the documentation of graphviz_string() for available options.

INPUT:

  • filename – the name of the file to write in

  • **options – options for the graphviz string

EXAMPLES:

sage: G = Graph({0: {1: None, 2: None}, 1: {0: None, 2: None},
....:            2: {0: None, 1: None, 3: 'foo'}, 3: {2: 'foo'}},
....:           sparse=True)
sage: import tempfile
sage: with tempfile.NamedTemporaryFile(mode="a+t") as f:
....:     G.graphviz_to_file_named(f.name, edge_labels=True)
....:     print(f.read())
graph {
  node_0  [label="0"];
  node_1  [label="1"];
  node_2  [label="2"];
  node_3  [label="3"];

  node_0 -- node_1;
  node_0 -- node_2;
  node_1 -- node_2;
  node_2 -- node_3 [label="foo"];
}
greedy_dominating_set(G, k=1, vertices=None, ordering=None, return_sets=False, closest=False)#

Return a greedy distance-\(k\) dominating set of the graph.

A distance-\(k\) dominating set \(S\) of a graph \(G\) is a set of its vertices of minimal cardinality such that any vertex of \(G\) is in \(S\) or is at distance at most \(k\) from a vertex in \(S\). See the Wikipedia article Dominating_set.

When \(G\) is directed, vertex \(u\) can be a dominator of vertex \(v\) if there is a directed path of length at most \(k\) from \(u\) to \(v\).

This method implements a greedy heuristic to find a minimal dominatic set.

INPUT:

  • G – a Graph

  • k – integer (default: 1); the domination distance to consider

  • vertices – iterable container of vertices (default: None); when specified, return a dominating set of the specified vertices only

  • ordering – string (default: None); specify the order in which to consider the vertices

    • None – if vertices is None, then consider the vertices in the order given by list(G). Otherwise, consider the vertices in the order of iteration of vertices.

    • "degree_min" – consider the vertices by increasing degree

    • "degree_max" – consider the vertices by decreasing degree

  • return_sets – boolean (default: False); whether to return the vertices of the dominating set only (default), or a dictionary mapping each vertex of the dominating set to the set of vertices it dominates.

  • closest – boolean (default: False); whether to attach a vertex to its closest dominator or not. This parameter is use only when return_sets is True.

EXAMPLES:

Dominating sets of a path:

sage: from sage.graphs.domination import greedy_dominating_set
sage: G = graphs.PathGraph(5)
sage: sorted(greedy_dominating_set(G, ordering=None))
[0, 2, 4]
sage: sorted(greedy_dominating_set(G, ordering="degree_min"))
[0, 2, 4]
sage: sorted(greedy_dominating_set(G, ordering="degree_max"))
[1, 3]
sage: sorted(greedy_dominating_set(G, k=2, ordering=None))
[0, 3]
sage: sorted(greedy_dominating_set(G, k=2, ordering="degree_min"))
[0, 4]
sage: sorted(greedy_dominating_set(G, k=2, ordering="degree_max"))
[1, 4]
sage: greedy_dominating_set(G, k=3, ordering="degree_min", return_sets=True, closest=False)
{0: {0, 1, 2, 3}, 4: {4}}
sage: greedy_dominating_set(G, k=3, ordering="degree_min", return_sets=True, closest=True)
{0: {0, 2, 3}, 4: {1, 4}}

Asking for a dominating set of a subset of vertices:

sage: from sage.graphs.domination import greedy_dominating_set
sage: from sage.graphs.domination import is_dominating
sage: G = graphs.PetersenGraph()
sage: vertices = {0, 1, 2, 3, 4, 5}
sage: dom = greedy_dominating_set(G, vertices=vertices, return_sets=True)
sage: sorted(dom)
[0, 2]
sage: is_dominating(G, dom, focus=vertices)
True
sage: is_dominating(G, dom)
False
sage: dominated = [u for v in dom for u in dom[v]]
sage: sorted(dominated) == sorted(vertices)
True

Influence of the ordering of the vertices on the result:

sage: from sage.graphs.domination import greedy_dominating_set
sage: G = graphs.StarGraph(4)
sage: greedy_dominating_set(G, vertices=[0, 1, 2, 3, 4])
[0]
sage: sorted(greedy_dominating_set(G, vertices=[1, 2, 3, 4, 0]))
[1, 2, 3, 4]

Dominating set of a directed graph:

sage: from sage.graphs.domination import greedy_dominating_set
sage: D = digraphs.Path(3)
sage: sorted(greedy_dominating_set(D, vertices=[0, 1, 2]))
[0, 2]
hamiltonian_cycle(algorithm, solver='tsp', constraint_generation=None, verbose=None, verbose_constraints=0, integrality_tolerance=False)#

Return a Hamiltonian cycle/circuit of the current graph/digraph.

A graph (resp. digraph) is said to be Hamiltonian if it contains as a subgraph a cycle (resp. a circuit) going through all the vertices.

Computing a Hamiltonian cycle/circuit being NP-Complete, this algorithm could run for some time depending on the instance.

ALGORITHM:

See traveling_salesman_problem() for ‘tsp’ algorithm and find_hamiltonian() from sage.graphs.generic_graph_pyx for ‘backtrack’ algorithm.

INPUT:

  • algorithm – string (default: 'tsp'); one of ‘tsp’ or ‘backtrack’

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

  • constraint_generation – boolean (default: None); whether to use constraint generation when solving the Mixed Integer Linear Program.

    When constraint_generation = None, constraint generation is used whenever the graph has a density larger than 70%.

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

  • verbose_constraints – boolean (default: False); whether to display which constraints are being generated

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

OUTPUT:

If using the 'tsp' algorithm, returns a Hamiltonian cycle/circuit if it exists; otherwise, raises a EmptySetError exception. If using the 'backtrack' algorithm, returns a pair (B, P). If B is True then P is a Hamiltonian cycle and if B is False, P is a longest path found by the algorithm. Observe that if B is False, the graph may still be Hamiltonian. The 'backtrack' algorithm is only implemented for undirected graphs.

Warning

The 'backtrack' algorithm may loop endlessly on graphs with vertices of degree 1.

NOTE:

This function, as is_hamiltonian(), computes a Hamiltonian cycle if it exists: the user should NOT test for Hamiltonicity using is_hamiltonian() before calling this function, as it would result in computing it twice.

The backtrack algorithm is only implemented for undirected graphs.

EXAMPLES:

The Heawood Graph is known to be Hamiltonian

sage: g = graphs.HeawoodGraph()
sage: g.hamiltonian_cycle()                                                 # needs sage.numerical.mip
TSP from Heawood graph: Graph on 14 vertices

The Petersen Graph, though, is not

sage: g = graphs.PetersenGraph()
sage: g.hamiltonian_cycle()                                                 # needs sage.numerical.mip
Traceback (most recent call last):
...
EmptySetError: the given graph is not Hamiltonian

Now, using the backtrack algorithm in the Heawood graph

sage: G=graphs.HeawoodGraph()
sage: G.hamiltonian_cycle(algorithm='backtrack')
(True, [...])

And now in the Petersen graph

sage: G=graphs.PetersenGraph()
sage: B, P = G.hamiltonian_cycle(algorithm='backtrack')
sage: B
False
sage: len(P)
10
sage: G.has_edge(P[0], P[-1])
False

Finally, we test the algorithm in a cube graph, which is Hamiltonian

sage: G=graphs.CubeGraph(3)
sage: G.hamiltonian_cycle(algorithm='backtrack')
(True, [...])
hamiltonian_path(s, t=None, use_edge_labels=None, maximize=False, algorithm=False, solver='MILP', verbose=None, integrality_tolerance=0)#

Return a Hamiltonian path of the current graph/digraph.

A path is Hamiltonian if it goes through all the vertices exactly once. Computing a Hamiltonian path being NP-Complete, this algorithm could run for some time depending on the instance.

When use_edge_labels == True, this method returns either a minimum weight hamiltonian path or a maximum weight Hamiltonian path (if maximize == True).

INPUT:

  • s – vertex (default: None); if specified, then forces the source of the path (the method then returns a Hamiltonian path starting at s)

  • t – vertex (default: None); if specified, then forces the destination of the path (the method then returns a Hamiltonian path ending at t)

  • use_edge_labels – boolean (default: False); whether to compute a weighted hamiltonian path where the weight of an edge is defined by its label (a label set to None or {} being considered as a weight of \(1\)), or a non-weighted hamiltonian path

  • maximize – boolean (default: False); whether to compute a minimum (default) or a maximum (when maximize == True) weight hamiltonian path. This parameter is considered only if use_edge_labels == True.

  • algorithm – string (default: "MILP"); the algorithm the use among "MILP" and "backtrack"; two remarks on this respect:

    • While the MILP formulation returns an exact answer, the backtrack algorithm is a randomized heuristic.

    • The backtrack algorithm does not support edge weighting, so setting use_edge_labels=True will force the use of the MILP algorithm.

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

OUTPUT:

A subgraph of self corresponding to a (directed if self is directed) hamiltonian path. If no hamiltonian path is found, return None. If use_edge_labels == True, a pair weight, path is returned.

EXAMPLES:

The \(3 \times 3\)-grid has an Hamiltonian path, an hamiltonian path starting from vertex \((0, 0)\) and ending at vertex \((2, 2)\), but no Hamiltonian path starting from \((0, 0)\) and ending at \((0, 1)\):

sage: # needs sage.numerical.mip
sage: g = graphs.Grid2dGraph(3, 3)
sage: g.hamiltonian_path()
Hamiltonian path from 2D Grid Graph for [3, 3]: Graph on 9 vertices
sage: g.hamiltonian_path(s=(0, 0), t=(2, 2))
Hamiltonian path from 2D Grid Graph for [3, 3]: Graph on 9 vertices
sage: g.hamiltonian_path(s=(0, 0), t=(2, 2), use_edge_labels=True)
(8, Hamiltonian path from 2D Grid Graph for [3, 3]: Graph on 9 vertices)
sage: g.hamiltonian_path(s=(0, 0), t=(0, 1)) is None
True
sage: g.hamiltonian_path(s=(0, 0), t=(0, 1), use_edge_labels=True)
(0, None)
has_edge(u, v=None, label=None)#

Check whether (u, v) is an edge of the (di)graph.

INPUT: The following forms are accepted:

  • G.has_edge( 1, 2 )

  • G.has_edge( (1, 2) )

  • G.has_edge( 1, 2, ‘label’ )

  • G.has_edge( (1, 2, ‘label’) )

EXAMPLES:

sage: graphs.EmptyGraph().has_edge(9, 2)
False
sage: DiGraph().has_edge(9, 2)
False
sage: G = Graph(sparse=True)
sage: G.add_edge(0, 1, "label")
sage: G.has_edge(0, 1, "different label")
False
sage: G.has_edge(0, 1, "label")
True
has_loops()#

Return whether there are loops in the (di)graph

EXAMPLES:

sage: G = Graph(loops=True); G
Looped graph on 0 vertices
sage: G.has_loops()
False
sage: G.allows_loops()
True
sage: G.add_edge((0, 0))
sage: G.has_loops()
True
sage: G.loops()
[(0, 0, None)]
sage: G.allow_loops(False); G
Graph on 1 vertex
sage: G.has_loops()
False
sage: G.edges(sort=True)
[]

sage: D = DiGraph(loops=True); D
Looped digraph on 0 vertices
sage: D.has_loops()
False
sage: D.allows_loops()
True
sage: D.add_edge((0, 0))
sage: D.has_loops()
True
sage: D.loops()
[(0, 0, None)]
sage: D.allow_loops(False); D
Digraph on 1 vertex
sage: D.has_loops()
False
sage: D.edges(sort=True)
[]
has_multiple_edges(to_undirected=False)#

Return whether there are multiple edges in the (di)graph.

INPUT:

  • to_undirected – (default: False); if True, runs the test on the undirected version of a DiGraph. Otherwise, treats DiGraph edges (u, v) and (v, u) as unique individual edges.

EXAMPLES:

sage: G = Graph(multiedges=True, sparse=True); G
Multi-graph on 0 vertices
sage: G.has_multiple_edges()
False
sage: G.allows_multiple_edges()
True
sage: G.add_edges([(0, 1)] * 3)
sage: G.has_multiple_edges()
True
sage: G.multiple_edges()
[(0, 1, None), (0, 1, None), (0, 1, None)]
sage: G.allow_multiple_edges(False); G
Graph on 2 vertices
sage: G.has_multiple_edges()
False
sage: G.edges(sort=True)
[(0, 1, None)]

sage: D = DiGraph(multiedges=True, sparse=True); D
Multi-digraph on 0 vertices
sage: D.has_multiple_edges()
False
sage: D.allows_multiple_edges()
True
sage: D.add_edges([(0, 1)] * 3)
sage: D.has_multiple_edges()
True
sage: D.multiple_edges()
[(0, 1, None), (0, 1, None), (0, 1, None)]
sage: D.allow_multiple_edges(False); D
Digraph on 2 vertices
sage: D.has_multiple_edges()
False
sage: D.edges(sort=True)
[(0, 1, None)]

sage: G = DiGraph({1: {2: 'h'}, 2: {1: 'g'}}, sparse=True)
sage: G.has_multiple_edges()
False
sage: G.has_multiple_edges(to_undirected=True)
True
sage: G.multiple_edges()
[]
sage: G.multiple_edges(to_undirected=True)
[(1, 2, 'h'), (2, 1, 'g')]

A loop is not a multiedge:

sage: g = Graph(loops=True, multiedges=True)
sage: g.add_edge(0, 0)
sage: g.has_multiple_edges()
False
has_vertex(vertex)#

Check if vertex is one of the vertices of this graph.

INPUT:

EXAMPLES:

sage: g = Graph({0: [1, 2, 3], 2: [4]}); g
Graph on 5 vertices
sage: 2 in g
True
sage: 10 in g
False
sage: graphs.PetersenGraph().has_vertex(99)
False
igraph_graph(vertex_list=None, vertex_attrs={}, edge_attrs={})#

Return an igraph graph from the Sage graph.

Optionally, it is possible to add vertex attributes and edge attributes to the output graph.

Note

This routine needs the optional package igraph to be installed: to do so, it is enough to run sage -i python_igraph. For more information on the Python version of igraph, see http://igraph.org/python/.

INPUT:

  • vertex_list – list (default: None); defines a mapping from the vertices of the graph to consecutive integers in (0, \ldots, n-1)`. Otherwise, the result of :meth:`vertices` will be used instead. Because :meth:`vertices` only works if the vertices can be sorted, using ``vertex_list is useful when working with possibly non-sortable objects in Python 3.

  • vertex_attrs – dictionary (default: {}); a dictionary where the key is a string (the attribute name), and the value is an iterable containing in position \(i\) the label of the \(i\)-th vertex in the list vertex_list if it is given or in vertices() when vertex_list == None (see http://igraph.org/python/doc/igraph.Graph-class.html#__init__ for more information)

  • edge_attrs – dictionary (default: {}); a dictionary where the key is a string (the attribute name), and the value is an iterable containing in position \(i\) the label of the \(i\)-th edge in the list outputted by edge_iterator() (see http://igraph.org/python/doc/igraph.Graph-class.html#__init__ for more information)

Note

In igraph, a graph is weighted if the edge labels have attribute weight. Hence, to create a weighted graph, it is enough to add this attribute.

Note

Often, Sage uses its own defined types for integer/floats. These types may not be igraph-compatible (see example below).

EXAMPLES:

Standard conversion:

sage: G = graphs.TetrahedralGraph()
sage: H = G.igraph_graph()          # optional - python_igraph
sage: H.summary()                   # optional - python_igraph
'IGRAPH U--- 4 6 -- '
sage: G = digraphs.Path(3)
sage: H = G.igraph_graph()          # optional - python_igraph
sage: H.summary()                   # optional - python_igraph
'IGRAPH D--- 3 2 -- '

Adding edge attributes:

sage: G = Graph([(1, 2, 'a'), (2, 3, 'b')])
sage: E = list(G.edge_iterator())
sage: H = G.igraph_graph(edge_attrs={'label': [e[2] for e in E]}) # optional - python_igraph
sage: H.es['label']                                               # optional - python_igraph
['a', 'b']

If edges have an attribute weight, the igraph graph is considered weighted:

sage: G = Graph([(1, 2, {'weight': 1}), (2, 3, {'weight': 2})])
sage: E = list(G.edge_iterator())
sage: H = G.igraph_graph(edge_attrs={'weight': [e[2]['weight'] for e in E]}) # optional - python_igraph
sage: H.is_weighted()                                                        # optional - python_igraph
True
sage: H.es['weight']                                                         # optional - python_igraph
[1, 2]

Adding vertex attributes:

sage: G = graphs.GridGraph([2, 2])
sage: H = G.igraph_graph(vertex_attrs={'name': G.vertices(sort=True)}) # optional - python_igraph
sage: H.vs()['name']                                          # optional - python_igraph
[(0, 0), (0, 1), (1, 0), (1, 1)]

Providing a mapping from vertices to consecutive integers:

sage: G = graphs.GridGraph([2, 2])
sage: V = list(G)
sage: H = G.igraph_graph(vertex_list=V, vertex_attrs={'name': V}) # optional - python_igraph
sage: H.vs()['name'] == V                                         # optional - python_igraph
True

Sometimes, Sage integer/floats are not compatible with igraph:

sage: G = Graph([(0, 1, 2)])
sage: E = list(G.edge_iterator())
sage: H = G.igraph_graph(edge_attrs={'capacity': [e[2] for e in E]}) # optional - python_igraph
sage: H.maxflow_value(0, 1, 'capacity')                              # optional - python_igraph
1.0
sage: H = G.igraph_graph(edge_attrs={'capacity': [float(e[2]) for e in E]}) # optional - python_igraph
sage: H.maxflow_value(0, 1, 'capacity')                                     # optional - python_igraph
2.0
incidence_matrix(oriented, sparse=None, vertices=True, edges=None, base_ring=None, **kwds)#

Return the incidence matrix of the (di)graph.

Each row is a vertex, and each column is an edge. The vertices are ordered as obtained by the method vertices(), except when parameter vertices is given (see below), and the edges as obtained by the method edge_iterator().

If the graph is not directed, then return a matrix with entries in \(\{0,1,2\}\). Each column will either contain two \(1\) (at the position of the endpoint of the edge), or one \(2\) (if the corresponding edge is a loop).

If the graph is directed return a matrix in \(\{-1,0,1\}\) where \(-1\) and \(+1\) correspond respectively to the source and the target of the edge. A loop will correspond to a zero column. In particular, it is not possible to recover the loops of an oriented graph from its incidence matrix.

See the Wikipedia article Incidence_matrix for more information.

INPUT:

  • oriented – boolean (default: None); when set to True, the matrix will be oriented (i.e. with entries in \(-1\), \(0\), \(1\)) and if set to False the matrix will be not oriented (i.e. with entries in \(0\), \(1\), \(2\)). By default, this argument is inferred from the graph type. Note that in the case the graph is not directed and with the option directed=True, a somewhat random direction is chosen for each edge.

  • sparse – boolean (default: True); whether to use a sparse or a dense matrix

  • vertices – list (default: None); when specified, the \(i\)-th row of the matrix corresponds to the \(i\)-th vertex in the ordering of vertices, otherwise, the \(i\)-th row of the matrix corresponds to the \(i\)-th vertex in the ordering given by method vertices().

  • edges – list (default: None); when specified, the \(i\)-th column of the matrix corresponds to the \(i\)-th edge in the ordering of edges, otherwise, the \(i\)-th column of the matrix corresponds to the \(i\)-th edge in the ordering given by method edge_iterator().

  • base_ring – a ring (default: ZZ); the base ring of the matrix space to use.

  • **kwds – other keywords to pass to matrix().

EXAMPLES:

sage: G = graphs.PetersenGraph()
sage: G.incidence_matrix()                                                  # needs sage.modules
[1 1 1 0 0 0 0 0 0 0 0 0 0 0 0]
[1 0 0 1 1 0 0 0 0 0 0 0 0 0 0]
[0 0 0 1 0 1 1 0 0 0 0 0 0 0 0]
[0 0 0 0 0 1 0 1 1 0 0 0 0 0 0]
[0 1 0 0 0 0 0 1 0 1 0 0 0 0 0]
[0 0 1 0 0 0 0 0 0 0 1 1 0 0 0]
[0 0 0 0 1 0 0 0 0 0 0 0 1 1 0]
[0 0 0 0 0 0 1 0 0 0 1 0 0 0 1]
[0 0 0 0 0 0 0 0 1 0 0 1 1 0 0]
[0 0 0 0 0 0 0 0 0 1 0 0 0 1 1]
sage: G.incidence_matrix(oriented=True)                                     # needs sage.modules
[-1 -1 -1  0  0  0  0  0  0  0  0  0  0  0  0]
[ 1  0  0 -1 -1  0  0  0  0  0  0  0  0  0  0]
[ 0  0  0  1  0 -1 -1  0  0  0  0  0  0  0  0]
[ 0  0  0  0  0  1  0 -1 -1  0  0  0  0  0  0]
[ 0  1  0  0  0  0  0  1  0 -1  0  0  0  0  0]
[ 0  0  1  0  0  0  0  0  0  0 -1 -1  0  0  0]
[ 0  0  0  0  1  0  0  0  0  0  0  0 -1 -1  0]
[ 0  0  0  0  0  0  1  0  0  0  1  0  0  0 -1]
[ 0  0  0  0  0  0  0  0  1  0  0  1  1  0  0]
[ 0  0  0  0  0  0  0  0  0  1  0  0  0  1  1]

sage: G = digraphs.Circulant(4, [1, 3])
sage: G.incidence_matrix()                                                  # needs sage.modules
[-1 -1  1  0  0  0  1  0]
[ 1  0 -1 -1  1  0  0  0]
[ 0  0  0  1 -1 -1  0  1]
[ 0  1  0  0  0  1 -1 -1]

sage: graphs.CompleteGraph(3).incidence_matrix()                            # needs sage.modules
[1 1 0]
[1 0 1]
[0 1 1]
sage: G = Graph([(0, 0), (0, 1), (0, 1)], loops=True, multiedges=True)
sage: G.incidence_matrix(oriented=False)                                    # needs sage.modules
[2 1 1]
[0 1 1]

A well known result states that the product of the (oriented) incidence matrix with its transpose of a (non-oriented graph) is in fact the Kirchhoff matrix:

sage: G = graphs.PetersenGraph()
sage: m = G.incidence_matrix(oriented=True)                                 # needs sage.modules
sage: m * m.transpose() == G.kirchhoff_matrix()                             # needs sage.modules
True

sage: K = graphs.CompleteGraph(3)
sage: m = K.incidence_matrix(oriented=True)                                 # needs sage.modules
sage: m * m.transpose() == K.kirchhoff_matrix()                             # needs sage.modules
True

sage: H = Graph([(0, 0), (0, 1), (0, 1)], loops=True, multiedges=True)
sage: m = H.incidence_matrix(oriented=True)                                 # needs sage.modules
sage: m * m.transpose() == H.kirchhoff_matrix()                             # needs sage.modules
True

A different ordering of the vertices:

sage: P5 = graphs.PathGraph(5)
sage: P5.incidence_matrix()                                                 # needs sage.modules
[1 0 0 0]
[1 1 0 0]
[0 1 1 0]
[0 0 1 1]
[0 0 0 1]
sage: P5.incidence_matrix(vertices=[2, 4, 1, 3, 0])                         # needs sage.modules
[0 1 1 0]
[0 0 0 1]
[1 1 0 0]
[0 0 1 1]
[1 0 0 0]

A different ordering of the edges:

sage: E = list(P5.edge_iterator(labels=False))
sage: P5.incidence_matrix(edges=E[::-1])                                    # needs sage.modules
[0 0 0 1]
[0 0 1 1]
[0 1 1 0]
[1 1 0 0]
[1 0 0 0]
sage: P5.incidence_matrix(vertices=[2, 4, 1, 3, 0], edges=E[::-1])          # needs sage.modules
[0 1 1 0]
[1 0 0 0]
[0 0 1 1]
[1 1 0 0]
[0 0 0 1]

A different base ring:

sage: P5.incidence_matrix(base_ring=RDF)                                    # needs sage.modules
[1.0 0.0 0.0 0.0]
[1.0 1.0 0.0 0.0]
[0.0 1.0 1.0 0.0]
[0.0 0.0 1.0 1.0]
[0.0 0.0 0.0 1.0]

Creating an immutable matrix:

sage: m = P5.incidence_matrix(immutable=True); m                            # needs sage.modules
[1 0 0 0]
[1 1 0 0]
[0 1 1 0]
[0 0 1 1]
[0 0 0 1]
sage: m[1,2] = 1                                                            # needs sage.modules
Traceback (most recent call last):
...
ValueError: matrix is immutable; please change a copy instead
(i.e., use copy(M) to change a copy of M).
is_bipartite(certificate=False)#

Check whether the graph is bipartite.

Traverse the graph \(G\) with breadth-first-search and color nodes.

INPUT:

  • certificate – boolean (default: False); whether to return a certificate. If set to True, the certificate returned is a proper 2-coloring when \(G\) is bipartite, and an odd cycle otherwise.

EXAMPLES:

sage: graphs.CycleGraph(4).is_bipartite()
True
sage: graphs.CycleGraph(5).is_bipartite()
False
sage: graphs.RandomBipartite(10, 10, 0.7).is_bipartite()                    # needs numpy
True

A random graph is very rarely bipartite:

sage: g = graphs.PetersenGraph()
sage: g.is_bipartite()
False
sage: false, oddcycle = g.is_bipartite(certificate=True)
sage: len(oddcycle) % 2
1

The method works identically with oriented graphs:

sage: g = DiGraph({0: [1, 2, 3], 2: [1], 3: [4]})
sage: g.is_bipartite()
False
sage: false, oddcycle = g.is_bipartite(certificate=True)
sage: len(oddcycle) % 2
1

sage: graphs.CycleGraph(4).random_orientation().is_bipartite()
True
sage: graphs.CycleGraph(5).random_orientation().is_bipartite()
False
is_cayley(return_group=False, mapping=False, generators=False, allow_disconnected=False)#

Check whether the graph is a Cayley graph.

If none of the parameters are True, return a boolean indicating whether the graph is a Cayley graph. Otherwise, return a tuple containing said boolean and the requested data. If the graph is not a Cayley graph, each of the data will be None.

The empty graph is defined to be not a Cayley graph.

Note

For this routine to work on all graphs, the optional package gap_packages needs to be installed: to do so, it is enough to run sage -i gap_packages.

INPUT:

  • return_group (boolean; False) – If True, return a group for which the graph is a Cayley graph.

  • mapping (boolean; False) – If True, return a mapping from vertices to group elements.

  • generators (boolean; False) – If True, return the generating set of the Cayley graph.

  • allow_disconnected (boolean; False) – If True, disconnected graphs are considered Cayley if they can be obtained from the Cayley construction with a generating set that does not generate the group.

ALGORITHM:

For connected graphs, find a regular subgroup of the automorphism group. For disconnected graphs, check that the graph is vertex-transitive and perform the check on one of its connected components. If a simple graph has density over 1/2, perform the check on its complement as its disconnectedness may increase performance.

EXAMPLES:

A Petersen Graph is not a Cayley graph:

sage: g = graphs.PetersenGraph()
sage: g.is_cayley()                                                         # needs sage.groups
False

A Cayley digraph is a Cayley graph:

sage: C7 = groups.permutation.Cyclic(7)                                     # needs sage.groups
sage: S = [(1,2,3,4,5,6,7), (1,3,5,7,2,4,6), (1,5,2,6,3,7,4)]
sage: d = C7.cayley_graph(generators=S)                                     # needs sage.groups
sage: d.is_cayley()                                                         # needs sage.groups
True

Graphs with loops and multiedges will have identity and repeated elements, respectively, among the generators:

sage: # needs sage.rings.finite_rings
sage: g = Graph(graphs.PaleyGraph(9), loops=True, multiedges=True)
sage: g.add_edges([(u, u) for u in g])
sage: g.add_edges([(u, u+1) for u in g])
sage: _, S = g.is_cayley(generators=True)                                   # needs sage.groups
sage: S  # random                                                           # needs sage.groups
[(),
 (0,2,1)(a,a + 2,a + 1)(2*a,2*a + 2,2*a + 1),
 (0,2,1)(a,a + 2,a + 1)(2*a,2*a + 2,2*a + 1),
 (0,1,2)(a,a + 1,a + 2)(2*a,2*a + 1,2*a + 2),
 (0,1,2)(a,a + 1,a + 2)(2*a,2*a + 1,2*a + 2),
 (0,2*a + 2,a + 1)(1,2*a,a + 2)(2,2*a + 1,a),
 (0,a + 1,2*a + 2)(1,a + 2,2*a)(2,a,2*a + 1)]
is_chordal(certificate=False, algorithm='B')#

Check whether the given graph is chordal.

A Graph \(G\) is said to be chordal if it contains no induced hole (a cycle of length at least 4).

Alternatively, chordality can be defined using a Perfect Elimination Order:

A Perfect Elimination Order of a graph \(G\) is an ordering \(v_1,...,v_n\) of its vertex set such that for all \(i\), the neighbors of \(v_i\) whose index is greater that \(i\) induce a complete subgraph in \(G\). Hence, the graph \(G\) can be totally erased by successively removing vertices whose neighborhood is a clique (also called simplicial vertices) [FG1965].

(It can be seen that if \(G\) contains an induced hole, then it cannot have a perfect elimination order. Indeed, if we write \(h_1,...,h_k\) the \(k\) vertices of such a hole, then the first of those vertices to be removed would have two non-adjacent neighbors in the graph.)

A Graph is then chordal if and only if it has a Perfect Elimination Order.

INPUT:

  • certificate – boolean (default: False); whether to return a certificate.

    • If certificate = False (default), returns True or False accordingly.

    • If certificate = True, returns :

      • (True, peo) when the graph is chordal, where peo is a perfect elimination order of its vertices.

      • (False, Hole) when the graph is not chordal, where Hole (a Graph object) is an induced subgraph of self isomorphic to a hole.

  • algorithm – string (default: "B"); the algorithm to choose among "A" or "B". While they will agree on whether the given graph is chordal, they cannot be expected to return the same certificates.

ALGORITHM:

This method implements the algorithm proposed in [RT1975] for the recognition of chordal graphs with time complexity in \(O(m)\). The algorithm works through computing a Lex BFS on the graph, then checking whether the order is a Perfect Elimination Order by computing for each vertex \(v\) the subgraph induced by its non-deleted neighbors, then testing whether this graph is complete.

EXAMPLES:

The lexicographic product of a Path and a Complete Graph is chordal

sage: g = graphs.PathGraph(5).lexicographic_product(graphs.CompleteGraph(3))
sage: g.is_chordal()
True

The same goes with the product of a random lobster (which is a tree) and a Complete Graph

sage: grl = graphs.RandomLobster(10, .5, .5)                                # needs networkx
sage: g = grl.lexicographic_product(graphs.CompleteGraph(3))                # needs networkx
sage: g.is_chordal()                                                        # needs networkx
True

The disjoint union of chordal graphs is still chordal:

sage: (2 * g).is_chordal()                                                  # needs networkx
True

Let us check the certificate given by Sage is indeed a perfect elimination order:

sage: _, peo = g.is_chordal(certificate=True)                               # needs networkx
sage: for v in peo:                                                         # needs networkx
....:     if not g.subgraph(g.neighbors(v)).is_clique():
....:          raise ValueError("this should never happen")
....:     g.delete_vertex(v)

Of course, the Petersen Graph is not chordal as it has girth 5:

sage: g = graphs.PetersenGraph()
sage: g.girth()
5
sage: g.is_chordal()
False

We can even obtain such a cycle as a certificate:

sage: _, hole = g.is_chordal(certificate=True)
sage: hole
Subgraph of (Petersen graph): Graph on 5 vertices
sage: hole.is_isomorphic(graphs.CycleGraph(5))
True
is_circulant(certificate=False)#

Check whether the graph is circulant.

For more information, see Wikipedia article Circulant_graph.

INPUT:

  • certificate – boolean (default: False); whether to return a certificate for yes-answers (see OUTPUT section)

OUTPUT:

When certificate is set to False (default) this method only returns True or False answers. When certificate is set to True, the method either returns (False, None) or (True, lists_of_parameters) each element of lists_of_parameters can be used to define the graph as a circulant graph.

See the documentation of CirculantGraph() and Circulant() for more information, and the examples below.

See also

CirculantGraph() – a constructor for circulant graphs.

EXAMPLES:

The Petersen graph is not a circulant graph:

sage: g = graphs.PetersenGraph()
sage: g.is_circulant()                                                      # needs sage.groups
False

A cycle is obviously a circulant graph, but several sets of parameters can be used to define it:

sage: g = graphs.CycleGraph(5)
sage: g.is_circulant(certificate=True)                                      # needs sage.groups
(True, [(5, [1, 4]), (5, [2, 3])])

The same goes for directed graphs:

sage: g = digraphs.Circuit(5)
sage: g.is_circulant(certificate=True)                                      # needs sage.groups
(True, [(5, [1]), (5, [3]), (5, [2]), (5, [4])])

With this information, it is very easy to create (and plot) all possible drawings of a circulant graph:

sage: g = graphs.CirculantGraph(13, [2, 3, 10, 11])
sage: for param in g.is_circulant(certificate=True)[1]:                     # needs sage.groups
....:    graphs.CirculantGraph(*param)
Circulant graph ([2, 3, 10, 11]): Graph on 13 vertices
Circulant graph ([1, 5, 8, 12]): Graph on 13 vertices
Circulant graph ([4, 6, 7, 9]): Graph on 13 vertices
is_circular_planar(on_embedding=None, kuratowski=False, set_embedding=True, boundary=None, ordered=False, set_pos=False)#

Check whether the graph is circular planar (outerplanar)

A graph is circular planar if it has a planar embedding in which all vertices can be drawn in order on a circle. This method can also be used to check the existence of a planar embedding in which the vertices of a specific set (the boundary) can be drawn on a circle, all other vertices being drawn inside of the circle. An order can be defined on the vertices of the boundary in order to define how they are to appear on the circle.

INPUT:

  • on_embedding – dictionary (default: None); the embedding dictionary to test planarity on (i.e.: will return True or False only for the given embedding)

  • kuratowski – boolean (default: False); whether to return a tuple with boolean first entry and the Kuratowski subgraph (i.e. an edge subdivision of \(K_5\) or \(K_{3,3}\)) as the second entry (see OUTPUT below)

  • set_embedding – boolean (default: True); whether or not to set the instance field variable that contains a combinatorial embedding (clockwise ordering of neighbors at each vertex). This value will only be set if a circular planar embedding is found. It is stored as a Python dict: v1: [n1,n2,n3] where v1 is a vertex and n1,n2,n3 are its neighbors.

  • boundary – list (default: None); an ordered list of vertices that are required to be drawn on the circle, all others being drawn inside of it. It is set to None by default, meaning that all vertices should be drawn on the boundary.

  • ordered – boolean (default: False); whether or not to consider the order of the boundary. It required boundary to be defined.

  • set_pos – boolean (default: False); whether or not to set the position dictionary (for plotting) to reflect the combinatorial embedding. Note that this value will default to False if set_embedding is set to False. Also, the position dictionary will only be updated if a circular planar embedding is found.

OUTPUT:

The method returns True if the graph is circular planar, and False if it is not.

If kuratowski is set to True, then this function will return a tuple, whose first entry is a boolean and whose second entry is the Kuratowski subgraph (i.e. an edge subdivision of \(K_5\) or \(K_{3,3}\)) isolated by the Boyer-Myrvold algorithm. Note that this graph might contain a vertex or edges that were not in the initial graph. These would be elements referred to below as parts of the wheel and the star, which were added to the graph to require that the boundary can be drawn on the boundary of a disc, with all other vertices drawn inside (and no edge crossings).

ALGORITHM:

This is a linear time algorithm to test for circular planarity. It relies on the edge-addition planarity algorithm due to Boyer-Myrvold. We accomplish linear time for circular planarity by modifying the graph before running the general planarity algorithm.

REFERENCE:

[BM2004]

EXAMPLES:

sage: g439 = Graph({1: [5, 7], 2: [5, 6], 3: [6, 7], 4: [5, 6, 7]})
sage: g439.show()                                                           # needs sage.plot
sage: g439.is_circular_planar(boundary=[1, 2, 3, 4])
False
sage: g439.is_circular_planar(kuratowski=True, boundary=[1, 2, 3, 4])
(False, Kuratowski subgraph of (): Graph on 8 vertices)
sage: g439.is_circular_planar(kuratowski=True, boundary=[1, 2, 3])
(True, None)
sage: g439.get_embedding()
{1: [5, 7],
 2: [6, 5],
 3: [7, 6],
 4: [5, 6, 7],
 5: [2, 4, 1],
 6: [3, 4, 2],
 7: [1, 4, 3]}

Order matters:

sage: K23 = graphs.CompleteBipartiteGraph(2, 3)
sage: K23.is_circular_planar(boundary=[0, 1, 2, 3])
True
sage: K23.is_circular_planar(ordered=True, boundary=[0, 1, 2, 3])
False

With a different order:

sage: K23.is_circular_planar(set_embedding=True, boundary=[0, 2, 1, 3])
True
is_clique(vertices=None, directed_clique=False, induced=True, loops=False)#

Check whether a set of vertices is a clique

A clique is a set of vertices such that there is exactly one edge between any two vertices.

INPUT:

  • vertices – a single vertex or an iterable container of vertices (default: None); when set, check whether the set of vertices is a clique, otherwise check whether ``self is a clique

  • directed_clique – boolean (default: False); if set to False, only consider the underlying undirected graph. If set to True and the graph is directed, only return True if all possible edges in _both_ directions exist.

  • induced – boolean (default: True); if set to True, check that the graph has exactly one edge between any two vertices. If set to False, check that the graph has at least one edge between any two vertices.

  • loops – boolean (default: False); if set to True, check that each vertex of the graph has a loop, and exactly one if furthermore induced == True. If set to False, check that the graph has no loop when induced == True, and ignore loops otherwise.

EXAMPLES:

sage: g = graphs.CompleteGraph(4)
sage: g.is_clique([1, 2, 3])
True
sage: g.is_clique()
True
sage: h = graphs.CycleGraph(4)
sage: h.is_clique([1, 2])
True
sage: h.is_clique([1, 2, 3])
False
sage: h.is_clique()
False
sage: i = digraphs.Complete(4)
sage: i.delete_edge([0, 1])
sage: i.is_clique(directed_clique=False, induced=True)
False
sage: i.is_clique(directed_clique=False, induced=False)
True
sage: i.is_clique(directed_clique=True)
False
is_connected(G)#

Check whether the (di)graph is connected.

Note that in a graph, path connected is equivalent to connected.

INPUT:

  • G – the input graph

See also

EXAMPLES:

sage: from sage.graphs.connectivity import is_connected
sage: G = Graph({0: [1, 2], 1: [2], 3: [4, 5], 4: [5]})
sage: is_connected(G)
False
sage: G.is_connected()
False
sage: G.add_edge(0,3)
sage: is_connected(G)
True
sage: D = DiGraph({0: [1, 2], 1: [2], 3: [4, 5], 4: [5]})
sage: is_connected(D)
False
sage: D.add_edge(0, 3)
sage: is_connected(D)
True
sage: D = DiGraph({1: [0], 2: [0]})
sage: is_connected(D)
True
is_cut_edge(G, u, v=None, label=None)#

Returns True if the input edge is a cut-edge or a bridge.

A cut edge (or bridge) is an edge that when removed increases the number of connected components. This function works with simple graphs as well as graphs with loops and multiedges. In a digraph, a cut edge is an edge that when removed increases the number of (weakly) connected components.

INPUT: The following forms are accepted

  • is_cut_edge(G, 1, 2 )

  • is_cut_edge(G, (1, 2) )

  • is_cut_edge(G, 1, 2, ‘label’ )

  • is_cut_edge(G, (1, 2, ‘label’) )

OUTPUT:

  • Returns True if (u,v) is a cut edge, False otherwise

EXAMPLES:

sage: from sage.graphs.connectivity import is_cut_edge
sage: G = graphs.CompleteGraph(4)
sage: is_cut_edge(G,0,2)
False
sage: G.is_cut_edge(0,2)
False

sage: G = graphs.CompleteGraph(4)
sage: G.add_edge((0,5,'silly'))
sage: is_cut_edge(G,(0,5,'silly'))
True

sage: G = Graph([[0,1],[0,2],[3,4],[4,5],[3,5]])
sage: is_cut_edge(G,(0,1))
True

sage: G = Graph([[0,1],[0,2],[1,1]], loops = True)
sage: is_cut_edge(G,(1,1))
False

sage: G = digraphs.Circuit(5)
sage: is_cut_edge(G,(0,1))
False

sage: G = graphs.CompleteGraph(6)
sage: is_cut_edge(G,(0,7))
Traceback (most recent call last):
...
ValueError: edge not in graph
is_cut_vertex(G, u, weak=False)#

Check whether the input vertex is a cut-vertex.

A vertex is a cut-vertex if its removal from the (di)graph increases the number of (strongly) connected components. Isolated vertices or leafs are not cut-vertices. This function works with simple graphs as well as graphs with loops and multiple edges.

INPUT:

  • G – a Sage (Di)Graph

  • u – a vertex

  • weak – boolean (default: False); whether the connectivity of directed graphs is to be taken in the weak sense, that is ignoring edges orientations

OUTPUT:

Return True if u is a cut-vertex, and False otherwise.

EXAMPLES:

Giving a LollipopGraph(4,2), that is a complete graph with 4 vertices with a pending edge:

sage: from sage.graphs.connectivity import is_cut_vertex
sage: G = graphs.LollipopGraph(4, 2)
sage: is_cut_vertex(G, 0)
False
sage: is_cut_vertex(G, 3)
True
sage: G.is_cut_vertex(3)
True

Comparing the weak and strong connectivity of a digraph:

sage: from sage.graphs.connectivity import is_strongly_connected
sage: D = digraphs.Circuit(6)
sage: is_strongly_connected(D)
True
sage: is_cut_vertex(D, 2)
True
sage: is_cut_vertex(D, 2, weak=True)
False

Giving a vertex that is not in the graph:

sage: G = graphs.CompleteGraph(4)
sage: is_cut_vertex(G, 7)
Traceback (most recent call last):
...
ValueError: vertex (7) is not a vertex of the graph
is_cycle(directed_cycle=True)#

Check whether self is a (directed) cycle graph.

We follow the definition provided in [BM2008] for undirected graphs. A cycle on three or more vertices is a simple graph whose vertices can be arranged in a cyclic order so that two vertices are adjacent if they are consecutive in the order, and not adjacent otherwise. A cycle on a vertex consists of a single vertex provided with a loop and a cycle with two vertices consists of two vertices connected by a pair of parallel edges. In other words, an undirected graph is a cycle if it is 2-regular and connected. The empty graph is not a cycle.

For directed graphs, a directed cycle, or circuit, on two or more vertices is a strongly connected directed graph without loops nor multiple edges with has many arcs as vertices. A circuit on a vertex consists of a single vertex provided with a loop.

INPUT:

  • directed_cycle – boolean (default: True); if set to True and the graph is directed, only return True if self is a directed cycle graph (i.e., a circuit). If set to False, we ignore the direction of edges and so opposite arcs become multiple (parallel) edges. This parameter is ignored for undirected graphs.

EXAMPLES:

sage: G = graphs.PetersenGraph()
sage: G.is_cycle()
False
sage: graphs.CycleGraph(5).is_cycle()
True
sage: Graph([(0,1 )]).is_cycle()
False
sage: Graph([(0, 1), (0, 1)], multiedges=True).is_cycle()
True
sage: Graph([(0, 1), (0, 1), (0, 1)], multiedges=True).is_cycle()
False
sage: Graph().is_cycle()
False
sage: G = Graph([(0, 0)], loops=True)
sage: G.is_cycle()
True
sage: digraphs.Circuit(3).is_cycle()
True
sage: digraphs.Circuit(2).is_cycle()
True
sage: digraphs.Circuit(2).is_cycle(directed_cycle=False)
True
sage: D = DiGraph(graphs.CycleGraph(3))
sage: D.is_cycle()
False
sage: D.is_cycle(directed_cycle=False)
False
sage: D.edges(sort=True, labels=False)
[(0, 1), (0, 2), (1, 0), (1, 2), (2, 0), (2, 1)]
is_drawn_free_of_edge_crossings()#

Check whether the position dictionary for this graph is set and that position dictionary gives a planar embedding.

This simply checks all pairs of edges that don’t share a vertex to make sure that they don’t intersect.

Note

This function require that _pos attribute is set (Returns False otherwise)

EXAMPLES:

sage: D = graphs.DodecahedralGraph()
sage: pos = D.layout(layout='planar', save_pos=True)
sage: D.is_drawn_free_of_edge_crossings()
True
is_equitable(partition, quotient_matrix=False)#

Checks whether the given partition is equitable with respect to self.

A partition is equitable with respect to a graph if for every pair of cells C1, C2 of the partition, the number of edges from a vertex of C1 to C2 is the same, over all vertices in C1.

INPUT:

  • partition - a list of lists

  • quotient_matrix - (default False) if True, and the partition is equitable, returns a matrix over the integers whose rows and columns represent cells of the partition, and whose i,j entry is the number of vertices in cell j adjacent to each vertex in cell i (since the partition is equitable, this is well defined)

EXAMPLES:

sage: G = graphs.PetersenGraph()
sage: G.is_equitable([[0,4],[1,3,5,9],[2,6,8],[7]])
False
sage: G.is_equitable([[0,4],[1,3,5,9],[2,6,8,7]])
True
sage: G.is_equitable([[0,4],[1,3,5,9],[2,6,8,7]], quotient_matrix=True)     # needs sage.modules
[1 2 0]
[1 0 2]
[0 2 1]
sage: ss = (graphs.WheelGraph(6)).line_graph(labels=False)
sage: prt = [[(0, 1)], [(0, 2), (0, 3), (0, 4), (1, 2), (1, 4)], [(2, 3), (3, 4)]]
sage: ss.is_equitable(prt)
Traceback (most recent call last):
...
TypeError: Partition ([[(0, 1)], [(0, 2), (0, 3), (0, 4), (1, 2), (1, 4)],
[(2, 3), (3, 4)]]) is not valid for this graph: vertices are incorrect.
sage: ss = (graphs.WheelGraph(5)).line_graph(labels=False)
sage: ss.is_equitable(prt)
False
is_eulerian(path=False)#

Check whether the graph is Eulerian.

A graph is Eulerian if it has a (closed) tour that visits each edge exactly once.

INPUT:

  • path – boolean (default: False); by default this function finds if the graph contains a closed tour visiting each edge once, i.e. an Eulerian cycle. If you want to test the existence of an Eulerian path, set this argument to True. Graphs with this property are sometimes called semi-Eulerian.

OUTPUT:

True or False for the closed tour case. For an open tour search (path``=``True) the function returns False if the graph is not semi-Eulerian, or a tuple (u, v) in the other case. This tuple defines the edge that would make the graph Eulerian, i.e. close an existing open tour. This edge may or may not be already present in the graph.

EXAMPLES:

sage: graphs.CompleteGraph(4).is_eulerian()
False
sage: graphs.CycleGraph(4).is_eulerian()
True
sage: g = DiGraph({0:[1,2], 1:[2]}); g.is_eulerian()
False
sage: g = DiGraph({0:[2], 1:[3], 2:[0,1], 3:[2]}); g.is_eulerian()
True
sage: g = DiGraph({0:[1], 1:[2], 2:[0], 3:[]}); g.is_eulerian()
True
sage: g = Graph([(1,2), (2,3), (3,1), (4,5), (5,6), (6,4)]); g.is_eulerian()
False
sage: g = DiGraph({0: [1]}); g.is_eulerian(path=True)
(1, 0)
sage: graphs.CycleGraph(4).is_eulerian(path=True)
False
sage: g = DiGraph({0: [1], 1: [2,3], 2: [4]}); g.is_eulerian(path=True)
False
sage: g = Graph({0:[1,2,3], 1:[2,3], 2:[3,4], 3:[4]}, multiedges=True)
sage: g.is_eulerian()
False
sage: e = g.is_eulerian(path=True); e
(0, 1)
sage: g.add_edge(e)
sage: g.is_eulerian(path=False)
True
sage: g.is_eulerian(path=True)
False
is_gallai_tree()#

Return whether the current graph is a Gallai tree.

A graph is a Gallai tree if and only if it is connected and its \(2\)-connected components are all isomorphic to complete graphs or odd cycles.

A connected graph is not degree-choosable if and only if it is a Gallai tree [ERT1979].

EXAMPLES:

A complete graph is, or course, a Gallai Tree:

sage: g = graphs.CompleteGraph(15)
sage: g.is_gallai_tree()
True

The Petersen Graph is not:

sage: g = graphs.PetersenGraph()
sage: g.is_gallai_tree()
False

A Graph built from vertex-disjoint complete graphs linked by one edge to a special vertex \(-1\) is a ‘’star-shaped’’ Gallai tree:

sage: g = 8 * graphs.CompleteGraph(6)
sage: g.add_edges([(-1, c[0]) for c in g.connected_components(sort=False)])
sage: g.is_gallai_tree()
True
is_geodetic(G)#

Check whether the input (di)graph is geodetic.

A graph \(G\) is geodetic if there exists only one shortest path between every pair of its vertices. This can be checked in time \(O(nm)\) in unweighted (di)graphs with \(n\) nodes and \(m\) edges. Examples of geodetic graphs are trees, cliques and odd cycles. See the Wikipedia article Geodetic_graph for more details.

(Di)graphs with multiple edges are not considered geodetic.

INPUT:

  • G – a graph or a digraph

EXAMPLES:

Trees, cliques and odd cycles are geodetic:

sage: T = graphs.RandomTree(20)
sage: T.is_geodetic()
True
sage: all(graphs.CompleteGraph(n).is_geodetic() for n in range(8))
True
sage: all(graphs.CycleGraph(n).is_geodetic() for n in range(3, 16, 2))
True

Even cycles of order at least 4 are not geodetic:

sage: all(graphs.CycleGraph(n).is_geodetic() for n in range(4, 17, 2))
False

The Petersen graph is geodetic:

sage: P = graphs.PetersenGraph()
sage: P.is_geodetic()
True

Grid graphs are not geodetic:

sage: G = graphs.Grid2dGraph(2, 3)
sage: G.is_geodetic()
False

This method is also valid for digraphs:

sage: G = DiGraph(graphs.PetersenGraph())
sage: G.is_geodetic()
True
sage: G = digraphs.Path(5)
sage: G.add_path([0, 'a', 'b', 'c', 4])
sage: G.is_geodetic()
False
is_hamiltonian(solver, constraint_generation=None, verbose=None, verbose_constraints=0, integrality_tolerance=False)#

Test whether the current graph is Hamiltonian.

A graph (resp. digraph) is said to be Hamiltonian if it contains as a subgraph a cycle (resp. a circuit) going through all the vertices.

Testing for Hamiltonicity being NP-Complete, this algorithm could run for some time depending on the instance.

ALGORITHM:

See traveling_salesman_problem().

INPUT:

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

  • constraint_generation (boolean) – whether to use constraint generation when solving the Mixed Integer Linear Program. When constraint_generation = None, constraint generation is used whenever the graph has a density larger than 70%.

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

  • verbose_constraints – boolean (default: False); whether to display which constraints are being generated

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

OUTPUT:

Returns True if a Hamiltonian cycle/circuit exists, and False otherwise.

NOTE:

This function, as hamiltonian_cycle and traveling_salesman_problem, computes a Hamiltonian cycle if it exists: the user should NOT test for Hamiltonicity using is_hamiltonian before calling hamiltonian_cycle or traveling_salesman_problem as it would result in computing it twice.

EXAMPLES:

The Heawood Graph is known to be Hamiltonian

sage: g = graphs.HeawoodGraph()
sage: g.is_hamiltonian()                                                    # needs sage.numerical.mip
True

The Petergraph, though, is not

sage: g = graphs.PetersenGraph()
sage: g.is_hamiltonian()                                                    # needs sage.numerical.mip
False
is_immutable()#

Check whether the graph is immutable.

EXAMPLES:

sage: G = graphs.PetersenGraph()
sage: G.is_immutable()
False
sage: Graph(G, immutable=True).is_immutable()
True
is_independent_set(vertices=None)#

Check whether vertices is an independent set of self.

An independent set is a set of vertices such that there is no edge between any two vertices.

INPUT:

  • vertices – a single vertex or an iterable container of vertices (default: None); when set, check whether the given set of vertices is an independent set, otherwise, check whether the set of vertices of ``self is an independent set

EXAMPLES:

sage: graphs.CycleGraph(4).is_independent_set([1,3])
True
sage: graphs.CycleGraph(4).is_independent_set([1,2,3])
False
is_interval(certificate=False)#

Check whether the graph is an interval graph.

An interval graph is one where every vertex can be seen as an interval on the real line so that there is an edge in the graph iff the corresponding intervals intersects.

See the Wikipedia article Interval_graph for more information.

INPUT:

  • certificate – boolean (default: False);

    • When certificate=False, returns True is the graph is an interval graph and False otherwise

    • When certificate=True, returns either (False, None) or (True, d) where d is a dictionary whose keys are the vertices and values are pairs of integers. They correspond to an embedding of the interval graph, each vertex being represented by an interval going from the first of the two values to the second.

ALGORITHM:

Through the use of PQ-Trees.

AUTHOR:

Nathann Cohen (implementation)

EXAMPLES:

sage: g = Graph({1: [2, 3, 4], 4: [2, 3]})
sage: g.is_interval()
True
sage: g.is_interval(certificate=True)
(True, {1: (0, 5), 2: (4, 6), 3: (1, 3), 4: (2, 7)})

The Petersen Graph is not chordal, so it cannot be an interval graph:

sage: g = graphs.PetersenGraph()
sage: g.is_interval()
False

A chordal but still not an interval graph:

sage: g = Graph({1: [4, 2, 3], 2: [3, 5], 3: [6]})
sage: g.is_interval()
False
is_isomorphic(other, certificate=False, verbosity=0, edge_labels=False)#

Tests for isomorphism between self and other.

INPUT:

  • certificate – if True, then output is \((a, b)\), where \(a\) is a boolean and \(b\) is either a map or None

  • edge_labels – boolean (default: False); if True allows only permutations respecting edge labels

OUTPUT:

  • either a boolean or, if certificate is True, a tuple consisting of a boolean and a map or None

EXAMPLES:

Graphs:

sage: from sage.groups.perm_gps.permgroup_named import SymmetricGroup       # needs sage.groups
sage: D = graphs.DodecahedralGraph()
sage: E = copy(D)
sage: gamma = SymmetricGroup(20).random_element()                           # needs sage.groups
sage: E.relabel(gamma)                                                      # needs sage.groups
sage: D.is_isomorphic(E)
True
sage: D = graphs.DodecahedralGraph()
sage: S = SymmetricGroup(20)                                                # needs sage.groups
sage: gamma = S.random_element()                                            # needs sage.groups
sage: E = copy(D)                                                           # needs sage.groups
sage: E.relabel(gamma)                                                      # needs sage.groups
sage: a,b = D.is_isomorphic(E, certificate=True); a
True
sage: from sage.graphs.generic_graph_pyx import spring_layout_fast
sage: position_D = spring_layout_fast(D)
sage: position_E = {}
sage: for vert in position_D:
....:  position_E[b[vert]] = position_D[vert]
sage: graphics_array([D.plot(pos=position_D), E.plot(pos=position_E)]).show()  # long time, needs sage.plot
sage: g=graphs.HeawoodGraph()
sage: g.is_isomorphic(g)
True

Multigraphs:

sage: G = Graph(multiedges=True,sparse=True)
sage: G.add_edge((0,1,1))
sage: G.add_edge((0,1,2))
sage: G.add_edge((0,1,3))
sage: G.add_edge((0,1,4))
sage: H = Graph(multiedges=True,sparse=True)
sage: H.add_edge((3,4))
sage: H.add_edge((3,4))
sage: H.add_edge((3,4))
sage: H.add_edge((3,4))
sage: G.is_isomorphic(H)
True

Digraphs:

sage: A = DiGraph( { 0 : [1,2] } )
sage: B = DiGraph( { 1 : [0,2] } )
sage: A.is_isomorphic(B, certificate=True)
(True, {0: 1, 1: 0, 2: 2})

Edge labeled graphs:

sage: G = Graph(sparse=True)
sage: G.add_edges( [(0,1,'a'),(1,2,'b'),(2,3,'c'),(3,4,'b'),(4,0,'a')] )
sage: H = G.relabel([1,2,3,4,0], inplace=False)
sage: G.is_isomorphic(H, edge_labels=True)
True

Edge labeled digraphs:

sage: G = DiGraph()
sage: G.add_edges( [(0,1,'a'),(1,2,'b'),(2,3,'c'),(3,4,'b'),(4,0,'a')] )
sage: H = G.relabel([1,2,3,4,0], inplace=False)
sage: G.is_isomorphic(H, edge_labels=True)
True
sage: G.is_isomorphic(H, edge_labels=True, certificate=True)
(True, {0: 1, 1: 2, 2: 3, 3: 4, 4: 0})
is_planar(on_embedding=None, kuratowski=False, set_embedding=False, set_pos=False)#

Check whether the graph is planar.

This wraps the reference implementation provided by John Boyer of the linear time planarity algorithm by edge addition due to Boyer Myrvold. (See reference code in planarity).

Note

The argument on_embedding takes precedence over set_embedding. This means that only the on_embedding combinatorial embedding will be tested for planarity and no _embedding attribute will be set as a result of this function call, unless on_embedding is None.

REFERENCE:

[BM2004]

See also

INPUT:

  • on_embedding – dictionary (default: None); the embedding dictionary to test planarity on (i.e.: will return True or False only for the given embedding)

  • kuratowski – boolean (default: False); whether to return a tuple with boolean as first entry. If the graph is nonplanar, will return the Kuratowski subgraph (i.e. an edge subdivision of \(K_5\) or \(K_{3,3}\)) as the second tuple entry. If the graph is planar, returns None as the second entry. When set to False, only a boolean answer is returned.

  • set_embedding – boolean (default: False); whether to set the instance field variable that contains a combinatorial embedding (clockwise ordering of neighbors at each vertex). This value will only be set if a planar embedding is found. It is stored as a Python dict: v1: [n1,n2,n3] where v1 is a vertex and n1,n2,n3 are its neighbors.

  • set_pos – boolean (default: False); whether to set the position dictionary (for plotting) to reflect the combinatorial embedding. Note that this value will default to False if set_emb is set to False. Also, the position dictionary will only be updated if a planar embedding is found.

EXAMPLES:

sage: g = graphs.CubeGraph(4)
sage: g.is_planar()
False
sage: g = graphs.CircularLadderGraph(4)
sage: g.is_planar(set_embedding=True)
True
sage: g.get_embedding()
{0: [1, 4, 3],
 1: [2, 5, 0],
 2: [3, 6, 1],
 3: [0, 7, 2],
 4: [0, 5, 7],
 5: [1, 6, 4],
 6: [2, 7, 5],
 7: [4, 6, 3]}
sage: g = graphs.PetersenGraph()
sage: (g.is_planar(kuratowski=True))[1].adjacency_matrix()                  # needs sage.modules
[0 1 0 0 0 1 0 0 0]
[1 0 1 0 0 0 1 0 0]
[0 1 0 1 0 0 0 1 0]
[0 0 1 0 0 0 0 0 1]
[0 0 0 0 0 0 1 1 0]
[1 0 0 0 0 0 0 1 1]
[0 1 0 0 1 0 0 0 1]
[0 0 1 0 1 1 0 0 0]
[0 0 0 1 0 1 1 0 0]
sage: k43 = graphs.CompleteBipartiteGraph(4, 3)
sage: result = k43.is_planar(kuratowski=True); result
(False,
 Kuratowski subgraph of (Complete bipartite graph of order 4+3): Graph on 6 vertices)
sage: result[1].is_isomorphic(graphs.CompleteBipartiteGraph(3, 3))
True

Multi-edged and looped graphs are partially supported:

sage: G = Graph({0: [1, 1]}, multiedges=True)
sage: G.is_planar()
True
sage: G.is_planar(on_embedding={})
Traceback (most recent call last):
...
NotImplementedError: cannot compute with embeddings of
 multiple-edged or looped graphs
sage: G.is_planar(set_embedding=True)
Traceback (most recent call last):
...
NotImplementedError: cannot compute with embeddings of
 multiple-edged or looped graphs
sage: G.is_planar(kuratowski=True)
(True, None)
sage: G.is_planar(set_pos=True)
True
sage: sorted(G.get_pos().items())
[(0, [0, 0]), (1, [0, 1])]

Digraphs with multiple edges or loops or pairs of opposite arcs are partially supported (github issue #35152):

sage: D = digraphs.Complete(3)
sage: D.is_planar()
True
sage: D.is_planar(set_pos=True)
True
sage: sorted(D.get_pos().items())
[(0, [0, 1]), (1, [1, 1]), (2, [1, 0])]
sage: D.is_planar(on_embedding={})
Traceback (most recent call last):
...
NotImplementedError: cannot compute with embeddings of
 digraphs with pairs of opposite arcs
sage: D.is_planar(set_embedding=True)
Traceback (most recent call last):
...
NotImplementedError: cannot compute with embeddings of
 digraphs with pairs of opposite arcs
sage: D.is_planar(kuratowski=True)
(True, None)
sage: D.allow_multiple_edges(True)
sage: D.add_edges(D.edges(sort=False))
sage: D.allow_loops(True)
sage: D.add_edges((u, u) for u in D)
sage: D.is_planar()
True
sage: D.is_planar(kuratowski=True)
(True, None)
sage: D.is_planar(set_pos=True)
True
sage: D.is_planar(set_embedding=True)
Traceback (most recent call last):
...
NotImplementedError: cannot compute with embeddings of
 multiple-edged or looped graphs
sage: D.is_planar(on_embedding={})
Traceback (most recent call last):
...
NotImplementedError: cannot compute with embeddings of
 multiple-edged or looped graphs
sage: G = graphs.CompleteGraph(5)
sage: G = Graph(G, multiedges=True)
sage: G.add_edge(0, 1)
sage: G.is_planar()
False
sage: b,k = G.is_planar(kuratowski=True)
sage: b
False
sage: k.vertices(sort=True)
[0, 1, 2, 3, 4]
is_regular(k=None)#

Check whether this graph is (\(k\)-)regular.

INPUT:

  • k – integer (default: None); the degree of regularity to check for

EXAMPLES:

sage: G = graphs.HoffmanSingletonGraph()
sage: G.is_regular()
True
sage: G.is_regular(9)
False

So the Hoffman-Singleton graph is regular, but not 9-regular. In fact, we can now find the degree easily as follows:

sage: next(G.degree_iterator())
7

The house graph is not regular:

sage: graphs.HouseGraph().is_regular()
False

A graph without vertices is \(k\)-regular for every \(k\):

sage: Graph().is_regular()
True
is_self_complementary()#

Check whether the graph is self-complementary.

A (di)graph is self-complementary if it is isomorphic to its (di)graph complement. For instance, the path graph \(P_4\) and the cycle graph \(C_5\) are self-complementary.

See also

EXAMPLES:

The only self-complementary path graph is \(P_4\):

sage: graphs.PathGraph(4).is_self_complementary()
True
sage: graphs.PathGraph(5).is_self_complementary()
False

The only self-complementary directed path is \(P_2\):

sage: digraphs.Path(2).is_self_complementary()
True
sage: digraphs.Path(3).is_self_complementary()
False

Every Paley graph is self-complementary:

sage: G = graphs.PaleyGraph(9)                                              # needs sage.libs.pari
sage: G.is_self_complementary()                                             # needs sage.libs.pari
True
is_subgraph(other, induced=True, up_to_isomorphism=False)#

Check whether self is a subgraph of other.

Warning

The arguments induced and up_to_isomorphism are set respectively to True and False by default.

INPUT:

  • other – a Sage (Di)Graph

  • induced – boolean (default: True); if set to True check whether the graph is an induced subgraph of other that is if the vertices of the graph are also vertices of other, and the edges of the graph are equal to the edges of other between the vertices contained in the graph.

    If set to False tests whether the graph is a subgraph of other that is if all vertices of the graph are also in other and all edges of the graph are also in other.

  • up_to_isomorphism – boolean (default: False); if set to True check whether other is a subgraph ignoring the labeling of vertices and edges. Otherwise, vertex and edge labellings must coincide in the copy or induced copy.

OUTPUT:

boolean – True iff the graph is a (possibly induced) subgraph of other.

See also

For more advanced search of subgraphs isomorphic to a given graph, you could consider the following methods:

EXAMPLES:

sage: P = graphs.PetersenGraph()
sage: G = P.subgraph(range(6))
sage: G.is_subgraph(P)
True

sage: H = graphs.CycleGraph(5)
sage: G = graphs.PathGraph(5)
sage: G.is_subgraph(H)
False
sage: G.is_subgraph(H, induced=False)
True
sage: H.is_subgraph(G, induced=False)
False

The 4x4 grid contains a path of length 15 and an induced path of length 11:

sage: p11 = graphs.PathGraph(11)
sage: p15 = graphs.PathGraph(15)
sage: g = graphs.Grid2dGraph(4, 4)
sage: p15.is_subgraph(g, induced=False, up_to_isomorphism=True)            # needs sage.modules
True
sage: p15.is_subgraph(g, induced=True, up_to_isomorphism=True)             # needs sage.modules
False
sage: p11.is_subgraph(g, induced=True, up_to_isomorphism=True)             # needs sage.modules
True
is_transitively_reduced()#

Check whether the digraph is transitively reduced.

A digraph is transitively reduced if it is equal to its transitive reduction. A graph is transitively reduced if it is a forest.

EXAMPLES:

sage: d = DiGraph({0: [1], 1: [2], 2: [3]})
sage: d.is_transitively_reduced()
True

sage: d = DiGraph({0: [1, 2], 1: [2]})
sage: d.is_transitively_reduced()
False

sage: d = DiGraph({0: [1, 2], 1: [2], 2: []})
sage: d.is_transitively_reduced()
False
is_vertex_transitive(partition=None, verbosity=0, edge_labels=False, order=False, return_group=True, orbits=False)#

Returns whether the automorphism group of self is transitive within the partition provided, by default the unit partition of the vertices of self (thus by default tests for vertex transitivity in the usual sense).

EXAMPLES:

sage: G = Graph({0:[1],1:[2]})
sage: G.is_vertex_transitive()
False
sage: P = graphs.PetersenGraph()
sage: P.is_vertex_transitive()                                              # needs sage.groups
True
sage: D = graphs.DodecahedralGraph()
sage: D.is_vertex_transitive()                                              # needs sage.groups
True
sage: R = graphs.RandomGNP(2000, .01)                                       # needs networkx
sage: R.is_vertex_transitive()                                              # needs networkx
False
katz_centrality(alpha, u=None)#

Return the Katz centrality of vertex \(u\).

Katz centrality of a node is a measure of centrality in a graph network. Katz centrality computes the relative influence of a node within a network. Connections made with distant neighbors are, however penalized by an attenuation factor \(\alpha\).

See the Wikipedia article Katz_centrality for more information.

INPUT:

  • alpha – a nonnegative real number, must be less than the reciprocal of the spectral radius of the graph (the maximum absolute eigenvalue of the adjacency matrix).

  • u – the vertex whose Katz centrality needs to be measured (default: None)

OUTPUT: a list containing the Katz centrality of each vertex if u=None otherwise Katz centrality of the vertex u.

EXAMPLES:

We compute the Katz centrality of a 4-cycle (note that by symmetry, all 4 vertices have the same centrality)

sage: G = graphs.CycleGraph(4)
sage: G.katz_centrality(1/20)                                               # needs sage.modules sage.rings.number_field
{0: 1/9, 1: 1/9, 2: 1/9, 3: 1/9}

Note that in the below example the nodes having indegree \(0\) also have the Katz centrality value as \(0\), as these nodes are not influenced by other nodes.

sage: G = DiGraph({1: [10], 2:[10,11], 3:[10,11], 4:[], 5:[11, 4], 6:[11],
....:              7:[10,11], 8:[10,11], 9:[10], 10:[11, 5, 8], 11:[6]})
sage: G.katz_centrality(.85)  # rel tol 1e-14                               # needs sage.modules sage.rings.number_field
{1: 0.000000000000000,
 2: 0.000000000000000,
 3: 0.000000000000000,
 4: 16.7319819819820,
 5: 18.6846846846847,
 6: 173.212076941807,
 7: 0.000000000000000,
 8: 18.6846846846847,
 9: 0.000000000000000,
 10: 20.9819819819820,
 11: 202.778914049184}
katz_matrix(alpha, nonedgesonly=False, vertices=None)#

Return the Katz matrix of the graph.

Katz centrality of a node is a measure of centrality in a graph network. Katz centrality computes the relative influence of a node within a network. Connections made with distant neighbors are, however penalized by an attenuation factor \(\alpha\).

Adding the values in the Katz matrix of all columns in a particular row gives the Katz centrality measure of the vertex represented by that particular row. Katz centrality measures influence by taking into account the total number of walks between a pair of nodes.

See the Wikipedia article Katz_centrality for more information.

INPUT:

  • alpha – a nonnegative real number, must be less than the reciprocal of the spectral radius of the graph (the maximum absolute eigenvalue of the adjacency matrix)

  • nonedgesonly – boolean (default: True); if True, value for each edge present in the graph is set to zero.

  • vertices – list (default: None); the ordering of the vertices defining how they should appear in the matrix. By default, the ordering given by GenericGraph.vertices() is used.

OUTPUT: the Katz matrix of the graph with parameter alpha

EXAMPLES:

We find the Katz matrix of an undirected 4-cycle.

sage: G = graphs.CycleGraph(4)
sage: G.katz_matrix(1/20)                                                   # needs sage.modules sage.rings.number_field
[1/198  5/99 1/198  5/99]
[ 5/99 1/198  5/99 1/198]
[1/198  5/99 1/198  5/99]
[ 5/99 1/198  5/99 1/198]

We find the Katz matrix of an undirected 4-cycle with all entries other than those which correspond to non-edges zeroed out.

sage: G.katz_matrix(1/20, True)                                             # needs sage.modules sage.rings.number_field
[    0     0 1/198     0]
[    0     0     0 1/198]
[1/198     0     0     0]
[    0     1/198 0     0]

This will give an error if alpha<=0 or alpha>=1/spectral_radius = 1/max (A.eigenvalues()).

We find the Katz matrix in a fan on 6 vertices.

sage: H = Graph([(0,1),(0,2),(0,3),(0,4),(0,5),(0,6),(1,2),(2,3),(3,4),(4,5)])
sage: H.katz_matrix(1/10)                                                   # needs sage.modules sage.rings.number_field
[   169/2256    545/4512      25/188    605/4512      25/188    545/4512    485/4512]
[   545/4512 7081/297792  4355/37224    229/9024   595/37224 4073/297792    109/9024]
[     25/188  4355/37224    172/4653      45/376    125/4653   595/37224       5/376]
[   605/4512    229/9024      45/376    337/9024      45/376    229/9024    121/9024]
[     25/188   595/37224    125/4653      45/376    172/4653  4355/37224       5/376]
[   545/4512 4073/297792   595/37224    229/9024  4355/37224 7081/297792    109/9024]
[   485/4512    109/9024       5/376    121/9024       5/376    109/9024     97/9024]
kirchhoff_matrix(weighted=None, indegree=True, normalized=False, signless=False, **kwds)#

Return the Kirchhoff matrix (a.k.a. the Laplacian) of the graph.

The Kirchhoff matrix is defined to be \(D + M\) if signless and \(D - M\) otherwise, where \(D\) is the diagonal degree matrix (each diagonal entry is the degree of the corresponding vertex), and \(M\) is the adjacency matrix. If normalized is True, then the returned matrix is \(D^{-1/2}(D+M)D^{-1/2}\) if signless and \(D^{-1/2}(D-M)D^{-1/2}\) otherwise.

(In the special case of DiGraphs, \(D\) is defined as the diagonal in-degree matrix or diagonal out-degree matrix according to the value of indegree)

INPUT:

  • weighted – boolean (default: None);

    • If True, the weighted adjacency matrix is used for \(M\), and the diagonal matrix \(D\) takes into account the weight of edges (replace in the definition “degree” by “sum of the incident edges”)

    • Else, each edge is assumed to have weight 1

    Default is to take weights into consideration if and only if the graph is weighted.

  • indegree – boolean (default: True); this parameter is considered only for digraphs.

    • If True, each diagonal entry of \(D\) is equal to the in-degree of the corresponding vertex

    • Else, each diagonal entry of \(D\) is equal to the out-degree of the corresponding vertex

    By default, indegree is set to True

  • normalized – boolean (default: False);

    • If True, the returned matrix is \(D^{-1/2}(D+M)D^{-1/2}\) for signless and \(D^{-1/2}(D-M)D^{-1/2}\) otherwise, a normalized version of the Laplacian matrix. More accurately, the normalizing matrix used is equal to \(D^{-1/2}\) only for non-isolated vertices. If vertex \(i\) is isolated, then diagonal entry \(i\) in the matrix is 1, rather than a division by zero

    • Else, the matrix \(D+M\) for signless and \(D-M\) otherwise is returned

  • signless – boolean (default: False);

    • If True, \(D+M\) is used in calculation of Kirchhoff matrix

    • Else, \(D-M\) is used in calculation of Kirchhoff matrix

Note that any additional keywords will be passed on to either the adjacency_matrix() or weighted_adjacency_matrix() method.

AUTHORS:

  • Tom Boothby

  • Jason Grout

EXAMPLES:

sage: G = Graph(sparse=True)
sage: G.add_edges([(0, 1, 1), (1, 2, 2), (0, 2, 3), (0, 3, 4)])
sage: M = G.kirchhoff_matrix(weighted=True); M                              # needs sage.modules
[ 8 -1 -3 -4]
[-1  3 -2  0]
[-3 -2  5  0]
[-4  0  0  4]
sage: M = G.kirchhoff_matrix(); M                                           # needs sage.modules
[ 3 -1 -1 -1]
[-1  2 -1  0]
[-1 -1  2  0]
[-1  0  0  1]
sage: M = G.laplacian_matrix(normalized=True); M                            # needs sage.modules sage.symbolic
[                   1 -1/6*sqrt(3)*sqrt(2) -1/6*sqrt(3)*sqrt(2)         -1/3*sqrt(3)]
[-1/6*sqrt(3)*sqrt(2)                    1                 -1/2                    0]
[-1/6*sqrt(3)*sqrt(2)                 -1/2                    1                    0]
[        -1/3*sqrt(3)                    0                    0                    1]
sage: M = G.kirchhoff_matrix(weighted=True, signless=True); M               # needs sage.modules
[8 1 3 4]
[1 3 2 0]
[3 2 5 0]
[4 0 0 4]

sage: G = Graph({0: [], 1: [2]})
sage: G.laplacian_matrix(normalized=True)                                   # needs sage.modules
[ 0  0  0]
[ 0  1 -1]
[ 0 -1  1]
sage: G.laplacian_matrix(normalized=True, signless=True)                    # needs sage.modules
[0 0 0]
[0 1 1]
[0 1 1]

A weighted directed graph with loops, changing the variable indegree

sage: G = DiGraph({1: {1: 2, 2: 3}, 2: {1: 4}}, weighted=True, sparse=True)
sage: G.laplacian_matrix()                                                  # needs sage.modules
[ 4 -3]
[-4  3]
sage: G = DiGraph({1: {1: 2, 2: 3}, 2: {1: 4}}, weighted=True, sparse=True)
sage: G.laplacian_matrix(indegree=False)                                    # needs sage.modules
[ 3 -3]
[-4  4]

A different ordering of the vertices (see adjacency_matrix() and weighted_adjacency_matrix()):

sage: G = Graph(sparse=True)
sage: G.add_edges([(0, 1, 1), (1, 2, 2), (0, 2, 3), (0, 3, 4)])
sage: M = G.kirchhoff_matrix(vertices=[3, 2, 1, 0]); M                      # needs sage.modules
[ 1  0  0 -1]
[ 0  2 -1 -1]
[ 0 -1  2 -1]
[-1 -1 -1  3]
sage: M = G.kirchhoff_matrix(weighted=True, vertices=[3, 2, 1, 0]); M       # needs sage.modules
[ 4  0  0 -4]
[ 0  5 -2 -3]
[ 0 -2  3 -1]
[-4 -3 -1  8]

When parameter immutable=True is passed, the output matrix is immutable:

sage: G = Graph([(0, 1)])
sage: M = G.kirchhoff_matrix(vertices=[0, 1], immutable=True)               # needs sage.modules
sage: M.is_immutable()                                                      # needs sage.modules
True
kronecker_product(other)#

Return the tensor product of self and other.

The tensor product of \(G\) and \(H\) is the graph \(L\) with vertex set \(V(L)\) equal to the Cartesian product of the vertices \(V(G)\) and \(V(H)\), and \(((u,v), (w,x))\) is an edge iff - \((u, w)\) is an edge of self, and - \((v, x)\) is an edge of other.

The tensor product is also known as the categorical product and the Kronecker product (referring to the Kronecker matrix product). See the Wikipedia article Kronecker_product.

EXAMPLES:

sage: Z = graphs.CompleteGraph(2)
sage: C = graphs.CycleGraph(5)
sage: T = C.tensor_product(Z); T
Graph on 10 vertices
sage: T.size()
10
sage: T.plot()                      # long time                             # needs sage.plot
Graphics object consisting of 21 graphics primitives
sage: D = graphs.DodecahedralGraph()
sage: P = graphs.PetersenGraph()
sage: T = D.tensor_product(P); T
Graph on 200 vertices
sage: T.size()
900
sage: T.plot()                      # long time                             # needs sage.plot
Graphics object consisting of 1101 graphics primitives
laplacian_matrix(weighted=None, indegree=True, normalized=False, signless=False, **kwds)#

Return the Kirchhoff matrix (a.k.a. the Laplacian) of the graph.

The Kirchhoff matrix is defined to be \(D + M\) if signless and \(D - M\) otherwise, where \(D\) is the diagonal degree matrix (each diagonal entry is the degree of the corresponding vertex), and \(M\) is the adjacency matrix. If normalized is True, then the returned matrix is \(D^{-1/2}(D+M)D^{-1/2}\) if signless and \(D^{-1/2}(D-M)D^{-1/2}\) otherwise.

(In the special case of DiGraphs, \(D\) is defined as the diagonal in-degree matrix or diagonal out-degree matrix according to the value of indegree)

INPUT:

  • weighted – boolean (default: None);

    • If True, the weighted adjacency matrix is used for \(M\), and the diagonal matrix \(D\) takes into account the weight of edges (replace in the definition “degree” by “sum of the incident edges”)

    • Else, each edge is assumed to have weight 1

    Default is to take weights into consideration if and only if the graph is weighted.

  • indegree – boolean (default: True); this parameter is considered only for digraphs.

    • If True, each diagonal entry of \(D\) is equal to the in-degree of the corresponding vertex

    • Else, each diagonal entry of \(D\) is equal to the out-degree of the corresponding vertex

    By default, indegree is set to True

  • normalized – boolean (default: False);

    • If True, the returned matrix is \(D^{-1/2}(D+M)D^{-1/2}\) for signless and \(D^{-1/2}(D-M)D^{-1/2}\) otherwise, a normalized version of the Laplacian matrix. More accurately, the normalizing matrix used is equal to \(D^{-1/2}\) only for non-isolated vertices. If vertex \(i\) is isolated, then diagonal entry \(i\) in the matrix is 1, rather than a division by zero

    • Else, the matrix \(D+M\) for signless and \(D-M\) otherwise is returned

  • signless – boolean (default: False);

    • If True, \(D+M\) is used in calculation of Kirchhoff matrix

    • Else, \(D-M\) is used in calculation of Kirchhoff matrix

Note that any additional keywords will be passed on to either the adjacency_matrix() or weighted_adjacency_matrix() method.

AUTHORS:

  • Tom Boothby

  • Jason Grout

EXAMPLES:

sage: G = Graph(sparse=True)
sage: G.add_edges([(0, 1, 1), (1, 2, 2), (0, 2, 3), (0, 3, 4)])
sage: M = G.kirchhoff_matrix(weighted=True); M                              # needs sage.modules
[ 8 -1 -3 -4]
[-1  3 -2  0]
[-3 -2  5  0]
[-4  0  0  4]
sage: M = G.kirchhoff_matrix(); M                                           # needs sage.modules
[ 3 -1 -1 -1]
[-1  2 -1  0]
[-1 -1  2  0]
[-1  0  0  1]
sage: M = G.laplacian_matrix(normalized=True); M                            # needs sage.modules sage.symbolic
[                   1 -1/6*sqrt(3)*sqrt(2) -1/6*sqrt(3)*sqrt(2)         -1/3*sqrt(3)]
[-1/6*sqrt(3)*sqrt(2)                    1                 -1/2                    0]
[-1/6*sqrt(3)*sqrt(2)                 -1/2                    1                    0]
[        -1/3*sqrt(3)                    0                    0                    1]
sage: M = G.kirchhoff_matrix(weighted=True, signless=True); M               # needs sage.modules
[8 1 3 4]
[1 3 2 0]
[3 2 5 0]
[4 0 0 4]

sage: G = Graph({0: [], 1: [2]})
sage: G.laplacian_matrix(normalized=True)                                   # needs sage.modules
[ 0  0  0]
[ 0  1 -1]
[ 0 -1  1]
sage: G.laplacian_matrix(normalized=True, signless=True)                    # needs sage.modules
[0 0 0]
[0 1 1]
[0 1 1]

A weighted directed graph with loops, changing the variable indegree

sage: G = DiGraph({1: {1: 2, 2: 3}, 2: {1: 4}}, weighted=True, sparse=True)
sage: G.laplacian_matrix()                                                  # needs sage.modules
[ 4 -3]
[-4  3]
sage: G = DiGraph({1: {1: 2, 2: 3}, 2: {1: 4}}, weighted=True, sparse=True)
sage: G.laplacian_matrix(indegree=False)                                    # needs sage.modules
[ 3 -3]
[-4  4]

A different ordering of the vertices (see adjacency_matrix() and weighted_adjacency_matrix()):

sage: G = Graph(sparse=True)
sage: G.add_edges([(0, 1, 1), (1, 2, 2), (0, 2, 3), (0, 3, 4)])
sage: M = G.kirchhoff_matrix(vertices=[3, 2, 1, 0]); M                      # needs sage.modules
[ 1  0  0 -1]
[ 0  2 -1 -1]
[ 0 -1  2 -1]
[-1 -1 -1  3]
sage: M = G.kirchhoff_matrix(weighted=True, vertices=[3, 2, 1, 0]); M       # needs sage.modules
[ 4  0  0 -4]
[ 0  5 -2 -3]
[ 0 -2  3 -1]
[-4 -3 -1  8]

When parameter immutable=True is passed, the output matrix is immutable:

sage: G = Graph([(0, 1)])
sage: M = G.kirchhoff_matrix(vertices=[0, 1], immutable=True)               # needs sage.modules
sage: M.is_immutable()                                                      # needs sage.modules
True
latex_options()#

Return an instance of GraphLatex for the graph.

Changes to this object will affect the LaTeX version of the graph. For a full explanation of how to use LaTeX to render graphs, see the introduction to the graph_latex module.

EXAMPLES:

sage: g = graphs.PetersenGraph()
sage: opts = g.latex_options()
sage: opts
LaTeX options for Petersen graph: {}
sage: opts.set_option('tkz_style', 'Classic')                               # needs sage.plot
sage: opts                                                                  # needs sage.plot
LaTeX options for Petersen graph: {'tkz_style': 'Classic'}
layout(layout=None, pos=None, dim=2, save_pos=False, **options)#

Return a layout for the vertices of this graph.

INPUT:

  • layout – string (default: None); specifies a layout algorithm among "acyclic", "acyclic_dummy", "circular", "ranked", "graphviz", "planar", "spring", "forest" or "tree"

  • pos – dictionary (default: None); a dictionary of positions

  • dim – integer (default: 2); the number of dimensions of the layout, 2 or 3

  • save_pos – boolean (default: False); whether to save the positions

  • **options – layout options (see below)

If layout is set, the specified algorithm is used to compute the positions.

Otherwise, if pos is specified, use the given positions.

Otherwise, try to fetch previously computed and saved positions.

Otherwise use the default layout (usually the spring layout).

If save_pos = True, the layout is saved for later use.

EXAMPLES:

sage: g = digraphs.ButterflyGraph(1)
sage: D2 = g.layout(); D2  # random
{('0', 0): [2.69..., 0.43...],
 ('0', 1): [1.35..., 0.86...],
 ('1', 0): [0.89..., -0.42...],
 ('1', 1): [2.26..., -0.87...]}

sage: g.layout(layout="acyclic_dummy", save_pos=True)
{('0', 0): [0.3..., 0],
 ('0', 1): [0.3..., 1],
 ('1', 0): [0.6..., 0],
 ('1', 1): [0.6..., 1]}
sage: g.get_pos()
{('0', 0): [0.3..., 0],
 ('0', 1): [0.3..., 1],
 ('1', 0): [0.6..., 0],
 ('1', 1): [0.6..., 1]}

sage: D3 = g.layout(dim=3); D3  # random
{('0', 0): [0.68..., 0.50..., -0.24...],
 ('0', 1): [1.02..., -0.02..., 0.93...],
 ('1', 0): [2.06..., -0.49..., 0.23...],
 ('1', 1): [1.74..., 0.01..., -0.92...]}

Some safety tests:

sage: sorted(D2.keys()) == sorted(D3.keys()) == sorted(g)
True
sage: isinstance(D2, dict) and isinstance(D3, dict)
True
sage: [c in RDF for c in D2[('0', 0)]]
[True, True]
sage: [c in RDF for c in D3[('0', 0)]]
[True, True, True]

Here is the list of all the available layout options (**options):

sage: from sage.graphs.graph_plot import layout_options
sage: for key, value in sorted(layout_options.items()):
....:     print("option {} : {}".format(key, value))
option by_component : Whether to do the spring layout by connected component -- a boolean.
option dim : The dimension of the layout -- 2 or 3.
option forest_roots : An iterable specifying which vertices to use as roots for the ``layout='forest'`` option. If no root is specified for a tree, then one is chosen close to the center of the tree. Ignored unless ``layout='forest'``.
option heights : A dictionary mapping heights to the list of vertices at this height.
option iterations : The number of times to execute the spring layout algorithm.
option layout : A layout algorithm -- one of : "acyclic", "circular" (plots the graph with vertices evenly distributed on a circle), "ranked", "graphviz", "planar", "spring" (traditional spring layout, using the graph's current positions as initial positions), or "tree" (the tree will be plotted in levels, depending on minimum distance for the root).
option prog : Which graphviz layout program to use -- one of "circo", "dot", "fdp", "neato", or "twopi".
option save_pos : Whether or not to save the computed position for the graph.
option spring : Use spring layout to finalize the current layout.
option tree_orientation : The direction of tree branches -- 'up', 'down', 'left' or 'right'.
option tree_root : A vertex designation for drawing trees. A vertex of the tree to be used as the root for the ``layout='tree'`` option. If no root is specified, then one is chosen close to the center of the tree. Ignored unless ``layout='tree'``.

Some of them only apply to certain layout algorithms. For details, see layout_acyclic(), layout_planar(), layout_circular(), layout_spring(), …

Warning

unknown optional arguments are silently ignored

Warning

graphviz and dot2tex are currently required to obtain a nice 'acyclic' layout. See layout_graphviz() for installation instructions.

A subclass may implement another layout algorithm "blah", by implementing a method .layout_blah. It may override the default layout by overriding layout_default(), and similarly override the predefined layouts.

Todo

use this feature for all the predefined graphs classes (like for the Petersen graph, …), rather than systematically building the layout at construction time.

layout_circular(dim=2, center=(0, 0), radius=1, shift=0, angle=0, **options)#

Return a circular layout for this graph

INPUT:

  • dim – integer (default: 2); the number of dimensions of the layout, 2 or 3

  • center – tuple (default: (0, 0)); position of the center of the circle

  • radius – (default: 1); the radius of the circle

  • shift – (default: 0); rotation of the circle. A value of shift=1 will replace in the drawing the \(i\)-th element of the list by the \((i-1)\)-th. Non-integer values are admissible, and a value of \(\alpha\) corresponds to a rotation of the circle by an angle of \(\alpha 2\pi/n\) (where \(n\) is the number of vertices set on the circle).

  • angle – (default: 0); rotate the embedding of all vertices. For instance, when angle == 0, the first vertex get position (center[0] + radius, center[1]). With a value of \(\pi/2\), the first vertex get position (center[0], center[1] + radius).

  • **options – other parameters not used here

OUTPUT: a dictionary mapping vertices to positions

EXAMPLES:

sage: G = graphs.CirculantGraph(7, [1, 3])
sage: G.layout_circular()
{0: (0.0, 1.0),
 1: (-0.78...,  0.62...),
 2: (-0.97..., -0.22...),
 3: (-0.43..., -0.90...),
 4: (0.43...,  -0.90...),
 5: (0.97...,  -0.22...),
 6: (0.78...,   0.62...)}
sage: G.plot(layout="circular")                                             # needs sage.plot
Graphics object consisting of 22 graphics primitives
layout_default(by_component=True, **options)#

Return a spring layout for this graph.

INPUT:

OUTPUT: a dictionary mapping vertices to positions

EXAMPLES:

sage: g = graphs.LadderGraph(3) #TODO!!!!
sage: g.layout_spring()  # random
{0: [1.0, -0.29...],
1: [1.64..., 0.30...],
2: [2.34..., 0.89...],
3: [1.49..., -0.83...],
4: [2.14..., -0.30...],
5: [2.80..., 0.22...]}
sage: g = graphs.LadderGraph(7)
sage: g.plot(layout="spring")                                               # needs sage.plot
Graphics object consisting of 34 graphics primitives
layout_extend_randomly(pos, dim=2)#

Extend randomly a partial layout

INPUT:

  • pos – a dictionary mapping vertices to positions

  • dim – integer (default: 2); the number of dimensions of the layout, 2 or 3

OUTPUT: a dictionary mapping vertices to positions

The vertices not referenced in pos are assigned random positions within the box delimited by the other vertices.

EXAMPLES:

sage: H = digraphs.ButterflyGraph(1)
sage: pos = {('0', 0): (0, 0), ('1', 1): (1, 1)}
sage: H.layout_extend_randomly(pos)  # random
{('0', 0): (0, 0),
 ('0', 1): [0.0446..., 0.332...],
 ('1', 0): [0.1114..., 0.514...],
 ('1', 1): (1, 1)}
sage: xmin, xmax, ymin, ymax = H._layout_bounding_box(pos)
sage: (xmin, ymin) == (0, 0) and (xmax, ymax) == (1, 1)
True
layout_forest(tree_orientation='down', forest_roots=None, **options)#

Return an ordered forest layout for this graph.

The function relies on layout_tree() to deal with each connected component.

INPUT:

  • forest_roots – an iterable of vertices (default: None); the root vertices of the trees in the forest; a vertex is chosen close to the center of each component for which no root is specified in forest_roots or if forest_roots is None

  • tree_orientation – string (default: 'down'); the direction in which the tree is growing, can be 'up', 'down', 'left' or 'right'

  • **options – other parameters ignored here

EXAMPLES:

sage: G = graphs.RandomTree(4) + graphs.RandomTree(5) + graphs.RandomTree(6)
sage: p = G.layout_forest()
sage: G.plot(pos=p)  # random                                               # needs sage.plot
Graphics object consisting of 28 graphics primitives

sage: P5 = graphs.PathGraph(5)
sage: H = P5 + P5 + graphs.BalancedTree(2,2)                                # needs networkx
sage: p = H.layout_forest(forest_roots=[14,3])                              # needs networkx
sage: H.plot(pos=p)                                                         # needs networkx sage.plot
Graphics object consisting of 32 graphics primitives
layout_graphviz(dim=2, prog='dot', **options)#

Call graphviz to compute a layout of the vertices of this graph.

INPUT:

  • dim – integer (default: 2); the number of dimensions of the layout, 2 or 3

  • prog – one of “dot”, “neato”, “twopi”, “circo”, or “fdp”

  • **options – other parameters used by method graphviz_string()

EXAMPLES:

sage: g = digraphs.ButterflyGraph(2)
sage: g.layout_graphviz()  # optional - dot2tex graphviz
{('...', ...): [...,...],
 ('...', ...): [...,...],
 ('...', ...): [...,...],
 ('...', ...): [...,...],
 ('...', ...): [...,...],
 ('...', ...): [...,...],
 ('...', ...): [...,...],
 ('...', ...): [...,...],
 ('...', ...): [...,...],
 ('...', ...): [...,...],
 ('...', ...): [...,...],
 ('...', ...): [...,...]}
sage: g.plot(layout="graphviz")  # optional - dot2tex graphviz
Graphics object consisting of 29 graphics primitives

Note: the actual coordinates are not deterministic

By default, an acyclic layout is computed using graphviz’s dot layout program. One may specify an alternative layout program:

sage: # optional - dot2tex graphviz
sage: g.plot(layout="graphviz", prog="dot")
Graphics object consisting of 29 graphics primitives
sage: g.plot(layout="graphviz", prog="neato")
Graphics object consisting of 29 graphics primitives
sage: g.plot(layout="graphviz", prog="twopi")
Graphics object consisting of 29 graphics primitives
sage: g.plot(layout="graphviz", prog="fdp")
Graphics object consisting of 29 graphics primitives
sage: g = graphs.BalancedTree(5,2)                                          # needs networkx
sage: g.plot(layout="graphviz", prog="circo")                               # needs networkx
Graphics object consisting of 62 graphics primitives

Todo

Put here some cool examples showcasing graphviz features.

This requires graphviz and the dot2tex spkg. Here are some installation tips:

  • Install graphviz >= 2.14 so that the programs dot, neato, etc. are in your path. The graphviz suite can be download from http://graphviz.org.

  • Install dot2tex with sage -i dot2tex

Todo

Use the graphviz functionality of Networkx 1.0 once it will be merged into Sage.

layout_planar(set_embedding=False, on_embedding=None, external_face=None, test=False, circular=False, **options)#

Compute a planar layout of the graph using Schnyder’s algorithm.

If set_embedding is set, a new combinatorial embedding is computed for the layout. Otherwise: if on_embedding is provided, then that combinatorial embedding is used for the layout. Otherwise: if a combinatorial embedding is set to the instance field variable of the graph (e.g. using set_embedding()), then that one is used, and if no combinatorial embedding is set, then one is computed.

If the graph is not planar, an error is raised.

INPUT:

  • set_embedding – boolean (default: False); whether to set the instance field variable that contains a combinatorial embedding to the combinatorial embedding used for the planar layout (see get_embedding())

  • on_embedding – dictionary (default: None); provide a combinatorial embedding

  • external_face – a pair \((u,v)\) of vertices (default: None); the external face of the drawing is chosen in such a way that \(u\) and \(v\) are consecutive vertices in the clockwise traversal of the external face, in particular \(uv\) has to be an edge of the graph. If external_face == None, an arbitrary external face is chosen.

  • test – boolean (default: False); whether to perform sanity tests along the way

  • circular – ignored

EXAMPLES:

sage: g = graphs.PathGraph(10)
sage: g.layout(layout='planar', save_pos=True, test=True)
{0: [0, 8],
 1: [8, 1],
 2: [1, 0],
 3: [7, 1],
 4: [1, 1],
 5: [5, 3],
 6: [2, 3],
 7: [2, 4],
 8: [1, 6],
 9: [2, 5]}
sage: g = graphs.BalancedTree(3, 4)                                         # needs networkx
sage: pos = g.layout(layout='planar', save_pos=True, test=True)             # needs networkx
sage: pos[0]                                                                # needs networkx
[0, 119]
sage: pos[120]                                                              # needs networkx
[21, 37]
sage: g = graphs.CycleGraph(7)
sage: g.layout(layout='planar', save_pos=True, test=True)
{0: [0, 5], 1: [5, 1], 2: [1, 0], 3: [4, 1], 4: [1, 1], 5: [2, 2], 6: [1, 2]}
sage: g = graphs.CompleteGraph(5)
sage: g.layout(layout='planar', save_pos=True, test=True, set_embedding=True)
Traceback (most recent call last):
...
ValueError: Complete graph is not a planar graph

Choose the external face of the drawing:

sage: g = graphs.CompleteGraph(4)
sage: g.layout(layout='planar', external_face=(0,1))
{0: [0, 2], 1: [2, 1], 2: [1, 0], 3: [1, 1]}
sage: g.layout(layout='planar', external_face=(3,1))
{0: [2, 1], 1: [0, 2], 2: [1, 1], 3: [1, 0]}

Choose the embedding:

sage: H = graphs.LadderGraph(4)
sage: em = {0:[1,4], 4:[0,5], 1:[5,2,0], 5:[4,6,1],
....:       2:[1,3,6], 6:[7,5,2], 3:[7,2], 7:[3,6]}
sage: p = H.layout_planar(on_embedding=em)
sage: p  # random
{2: [8.121320343559642, 1],
 3: [2.1213203435596424, 6],
 7: [3.1213203435596424, 0],
 0: [5.121320343559642, 3],
 1: [3.1213203435596424, 5],
 4: [4.121320343559642, 3],
 5: [4.121320343559642, 2],
 6: [3.1213203435596424, 1],
 9: [9.698670612749268, 1],
 8: [8.698670612749268, 1],
 10: [9.698670612749268, 0]}
layout_ranked(heights=None, dim=2, spring=False, **options)#

Return a ranked layout for this graph

INPUT:

  • heights – dictionary (default: None); a dictionary mapping heights to the list of vertices at this height

  • dim – integer (default: 2); the number of dimensions of the layout, 2 or 3

  • spring – boolean (default: False);

  • **options – options for method spring_layout_fast()

OUTPUT: a dictionary mapping vertices to positions

Returns a layout computed by randomly arranging the vertices along the given heights

EXAMPLES:

sage: g = graphs.LadderGraph(3)
sage: g.layout_ranked(heights={i: (i, i+3) for i in range(3)})  # random
{0: [0.668..., 0],
 1: [0.667..., 1],
 2: [0.677..., 2],
 3: [1.34..., 0],
 4: [1.33..., 1],
 5: [1.33..., 2]}
sage: g = graphs.LadderGraph(7)
sage: g.plot(layout="ranked", heights={i: (i, i+7) for i in range(7)})      # needs sage.plot
Graphics object consisting of 34 graphics primitives
layout_spring(by_component=True, **options)#

Return a spring layout for this graph.

INPUT:

OUTPUT: a dictionary mapping vertices to positions

EXAMPLES:

sage: g = graphs.LadderGraph(3) #TODO!!!!
sage: g.layout_spring()  # random
{0: [1.0, -0.29...],
1: [1.64..., 0.30...],
2: [2.34..., 0.89...],
3: [1.49..., -0.83...],
4: [2.14..., -0.30...],
5: [2.80..., 0.22...]}
sage: g = graphs.LadderGraph(7)
sage: g.plot(layout="spring")                                               # needs sage.plot
Graphics object consisting of 34 graphics primitives
layout_tree(tree_orientation='down', tree_root=None, dim=2, **options)#

Return an ordered tree layout for this graph.

The graph must be a tree (no non-oriented cycles). In case of doubt whether the graph is connected or not, prefer layout_forest().

INPUT:

  • tree_root – a vertex (default: None); the root vertex of the tree. By default (None) a vertex is chosen close to the center of the tree.

  • tree_orientation – string (default: 'down'); the direction in which the tree is growing, can be 'up', 'down', 'left' or 'right'

  • dim – integer (default: 2); the number of dimensions of the layout, 2 or 3

  • **options – other parameters not used here

If the tree has been given a planar embedding (fixed circular order on the set of neighbors of every vertex) using set_embedding, the algorithm will create a layout that respects this embedding.

OUTPUT: a dictionary mapping vertices to positions

EXAMPLES:

sage: G = graphs.RandomTree(80)
sage: G.plot(layout="tree", tree_orientation="right")                       # needs sage.plot
Graphics object consisting of 160 graphics primitives

sage: T = graphs.RandomLobster(25, 0.3, 0.3)                                # needs networkx
sage: T.show(layout='tree', tree_orientation='up')                          # needs networkx sage.plot

sage: G = graphs.HoffmanSingletonGraph()
sage: T = Graph()
sage: T.add_edges(G.min_spanning_tree(starting_vertex=0))
sage: T.show(layout='tree', tree_root=0)                                    # needs sage.plot

sage: G = graphs.BalancedTree(2, 2)                                         # needs networkx
sage: G.layout_tree(tree_root=0)                                            # needs networkx
{0: [1.5, 0],
 1: [2.5, -1],
 2: [0.5, -1],
 3: [3.0, -2],
 4: [2.0, -2],
 5: [1.0, -2],
 6: [0.0, -2]}

sage: G = graphs.BalancedTree(2, 4)                                         # needs networkx
sage: G.plot(layout="tree", tree_root=0, tree_orientation="up")             # needs networkx sage.plot
Graphics object consisting of 62 graphics primitives

Using the embedding when it exists:

sage: T = Graph([(0, 1), (0, 6), (0, 3), (1, 2), (1, 5), (3, 4), (3, 7), (3, 8)])
sage: T.set_embedding({0: [1, 6, 3], 1: [2, 5, 0], 2: [1], 3: [4, 7, 8, 0],
....:     4: [3], 5: [1], 6: [0], 7: [3], 8: [3]})
sage: T.layout_tree()
{0: [2.166..., 0],
 1: [3.5, -1],
 2: [4.0, -2],
 3: [1.0, -1],
 4: [2.0, -2],
 5: [3.0, -2],
 6: [2.0, -1],
 7: [1.0, -2],
 8: [0.0, -2]}
sage: T.plot(layout="tree", tree_root=3)                                    # needs sage.plot
Graphics object consisting of 18 graphics primitives
lex_BFS(G, reverse=False, tree=False, initial_vertex=None, algorithm='fast')#

Perform a lexicographic breadth first search (LexBFS) on the graph.

INPUT:

  • G – a sage graph

  • reverse – boolean (default: False); whether to return the vertices in discovery order, or the reverse

  • tree – boolean (default: False); whether to return the discovery directed tree (each vertex being linked to the one that saw it for the first time)

  • initial_vertex – (default: None); the first vertex to consider

  • algorithm – string (default: "fast"); algorithm to use among:

    • "slow" – This algorithm maintains for each vertex left in the graph a code corresponding to the vertices already removed. The vertex of maximal code (according to the lexicographic order) is then removed, and the codes are updated. See for instance [CK2008] for more details. The time complexity of this algorithm as described in [CK2008] is in \(O(n + m)\), where \(n\) is the number of vertices and \(m\) is the number of edges, but our implementation is in \(O(n^2)\).

    • "fast" – This algorithm uses the notion of slices to refine the position of the vertices in the ordering. The time complexity of this algorithm is in \(O(n + m)\), and our implementation follows that complexity. See [HMPV2000] and next section for more details.

ALGORITHM:

The "fast" algorithm is the \(O(n + m)\) time algorithm proposed in [HMPV2000], where \(n\) is the number of vertices and \(m\) is the number of edges. It uses the notion of slices, i.e., subsets of consecutive vertices in the ordering, and iteratively refines the slices by subdividing them into sub-slices to determine the exact position of the vertices in the ordering.

Consider an ordering \(\sigma\) of the vertices. For a vertex \(v\), we define \(N_i(v) = \{u | u \in N(v) \text{ and } \sigma(u) < i\}\), that is the subset of neighbors of \(v\) appearing before the \(i\)-th vertex in the ordering \(\sigma\). Now, a slice of an ordering \(\sigma\) is a set of consecutive vertices, \(S = \{u | i \leq \sigma(u) \leq j\}\), such that for any \(u \in S\), we have \(N_i(u) = N_i(\sigma^{-1}(i))\) and for any \(v\) such that \(j < \sigma(v)\), \(N_i(v) \neq N_i(\sigma^{-1}(i))\). The head of a slice is the first position of its vertices.

The algorithm starts with a single slice containing all vertices. Then, when the position of the \(i\)-th vertex \(v\) is fixed, it explores the neighbors of \(v\) that have not yet been ordered. Consider a slice \(S\) such that \(N(x)\cap S \neq \emptyset\). The algorithm will rearrange the ordering of the vertices in \(S\) so that the first vertices are the neighbors of \(v\). The sub-slice containing the neighbors of \(v\) is assigned a new slice name, and the head of slice \(S\) is set to the position of the first vertex of \(S \setminus N(v)\) in the ordering \(\sigma\).

Observe that each arc of the graph can induce the subdivision of a slice. Hence, the algorithm can use up to \(m + 1\) different slices.

See also

EXAMPLES:

A Lex BFS is obviously an ordering of the vertices:

sage: g = graphs.CompleteGraph(6)
sage: len(g.lex_BFS()) == g.order()
True

Lex BFS ordering of the 3-sun graph:

sage: g = Graph([(1, 2), (1, 3), (2, 3), (2, 4), (2, 5), (3, 5), (3, 6), (4, 5), (5, 6)])
sage: g.lex_BFS()
[1, 2, 3, 5, 4, 6]

The method also works for directed graphs:

sage: G = DiGraph([(1, 2), (2, 3), (1, 3)])
sage: G.lex_BFS(initial_vertex=2, algorithm="slow")
[2, 3, 1]
sage: G.lex_BFS(initial_vertex=2, algorithm="fast")
[2, 3, 1]

For a Chordal Graph, a reversed Lex BFS is a Perfect Elimination Order:

sage: g = graphs.PathGraph(3).lexicographic_product(graphs.CompleteGraph(2))
sage: g.lex_BFS(reverse=True)
[(2, 1), (2, 0), (1, 1), (1, 0), (0, 1), (0, 0)]

And the vertices at the end of the tree of discovery are, for chordal graphs, simplicial vertices (their neighborhood is a complete graph):

sage: g = graphs.ClawGraph().lexicographic_product(graphs.CompleteGraph(2))
sage: v = g.lex_BFS()[-1]
sage: peo, tree = g.lex_BFS(initial_vertex = v,  tree=True)
sage: leaves = [v for v in tree if tree.in_degree(v) ==0]
sage: all(g.subgraph(g.neighbors(v)).is_clique() for v in leaves)
True

Different orderings for different traversals:

sage: # needs sage.combinat
sage: G = digraphs.DeBruijn(2,3)
sage: G.lex_BFS(initial_vertex='000', algorithm="fast")
['000', '001', '100', '010', '011', '110', '101', '111']
sage: G.lex_BFS(initial_vertex='000', algorithm="slow")
['000', '001', '100', '010', '011', '110', '101', '111']
sage: G.lex_DFS(initial_vertex='000')
['000', '001', '100', '010', '101', '110', '011', '111']
sage: G.lex_UP(initial_vertex='000')
['000', '001', '010', '101', '110', '111', '011', '100']
sage: G.lex_DOWN(initial_vertex='000')
['000', '001', '100', '011', '010', '110', '111', '101']
lex_DFS(G, reverse=False, tree=False, initial_vertex=None)#

Perform a lexicographic depth first search (LexDFS) on the graph.

INPUT:

  • G – a sage graph

  • reverse – boolean (default: False); whether to return the vertices in discovery order, or the reverse

  • tree – boolean (default: False); whether to return the discovery directed tree (each vertex being linked to the one that saw it for the first time)

  • initial_vertex – (default: None); the first vertex to consider

ALGORITHM:

This algorithm maintains for each vertex left in the graph a code corresponding to the vertices already removed. The vertex of maximal code (according to the lexicographic order) is then removed, and the codes are updated. Lex DFS differs from Lex BFS only in the way codes are updated after each iteration.

Time complexity is \(O(n+m)\) where \(n\) is the number of vertices and \(m\) is the number of edges.

See [CK2008] for more details on the algorithm.

See also

  • lex_BFS() – perform a lexicographic breadth first search (LexBFS) on the graph

  • lex_UP() – perform a lexicographic UP search (LexUP) on the graph

  • lex_DOWN() – perform a lexicographic DOWN search (LexDOWN) on the graph

EXAMPLES:

A Lex DFS is obviously an ordering of the vertices:

sage: g = graphs.CompleteGraph(6)
sage: len(g.lex_DFS()) == g.order()
True

Lex DFS ordering of the 3-sun graph:

sage: g = Graph([(1, 2), (1, 3), (2, 3), (2, 4), (2, 5), (3, 5), (3, 6), (4, 5), (5, 6)])
sage: g.lex_DFS()
[1, 2, 3, 5, 6, 4]

The method also works for directed graphs:

sage: G = DiGraph([(1, 2), (2, 3), (1, 3)])
sage: G.lex_DFS(initial_vertex=2)
[2, 3, 1]

Different orderings for different traversals:

sage: # needs sage.combinat
sage: G = digraphs.DeBruijn(2,3)
sage: G.lex_BFS(initial_vertex='000')
['000', '001', '100', '010', '011', '110', '101', '111']
sage: G.lex_DFS(initial_vertex='000')
['000', '001', '100', '010', '101', '110', '011', '111']
sage: G.lex_UP(initial_vertex='000')
['000', '001', '010', '101', '110', '111', '011', '100']
sage: G.lex_DOWN(initial_vertex='000')
['000', '001', '100', '011', '010', '110', '111', '101']
lex_DOWN(G, reverse=False, tree=False, initial_vertex=None)#

Perform a lexicographic DOWN search (LexDOWN) on the graph.

INPUT:

  • G – a sage graph

  • reverse – boolean (default: False); whether to return the vertices in discovery order, or the reverse

  • tree – boolean (default: False); whether to return the discovery directed tree (each vertex being linked to the one that saw it for the first time)

  • initial_vertex – (default: None); the first vertex to consider

ALGORITHM:

This algorithm maintains for each vertex left in the graph a code corresponding to the vertices already removed. The vertex of maximal code (according to the lexicographic order) is then removed, and the codes are updated. During the \(i\)-th iteration of the algorithm \(n-i\) is prepended to the codes of all neighbors of the selected vertex that are left in the graph.

Time complexity is \(O(n+m)\) where \(n\) is the number of vertices and \(m\) is the number of edges.

See [Mil2017] for more details on the algorithm.

See also

  • lex_BFS() – perform a lexicographic breadth first search (LexBFS) on the graph

  • lex_DFS() – perform a lexicographic depth first search (LexDFS) on the graph

  • lex_UP() – perform a lexicographic UP search (LexUP) on the graph

EXAMPLES:

A Lex DOWN is obviously an ordering of the vertices:

sage: g = graphs.CompleteGraph(6)
sage: len(g.lex_DOWN()) == g.order()
True

Lex DOWN ordering of the 3-sun graph:

sage: g = Graph([(1, 2), (1, 3), (2, 3), (2, 4), (2, 5), (3, 5), (3, 6), (4, 5), (5, 6)])
sage: g.lex_DOWN()
[1, 2, 3, 4, 6, 5]

The method also works for directed graphs:

sage: G = DiGraph([(1, 2), (2, 3), (1, 3)])
sage: G.lex_DOWN(initial_vertex=2)
[2, 3, 1]

Different orderings for different traversals:

sage: # needs sage.combinat
sage: G = digraphs.DeBruijn(2,3)
sage: G.lex_BFS(initial_vertex='000')
['000', '001', '100', '010', '011', '110', '101', '111']
sage: G.lex_DFS(initial_vertex='000')
['000', '001', '100', '010', '101', '110', '011', '111']
sage: G.lex_UP(initial_vertex='000')
['000', '001', '010', '101', '110', '111', '011', '100']
sage: G.lex_DOWN(initial_vertex='000')
['000', '001', '100', '011', '010', '110', '111', '101']
lex_UP(G, reverse=False, tree=False, initial_vertex=None)#

Perform a lexicographic UP search (LexUP) on the graph.

INPUT:

  • G – a sage graph

  • reverse – boolean (default: False); whether to return the vertices in discovery order, or the reverse

  • tree – boolean (default: False); whether to return the discovery directed tree (each vertex being linked to the one that saw it for the first time)

  • initial_vertex – (default: None); the first vertex to consider

ALGORITHM:

This algorithm maintains for each vertex left in the graph a code corresponding to the vertices already removed. The vertex of maximal code (according to the lexicographic order) is then removed, and the codes are updated. During the \(i\)-th iteration of the algorithm \(i\) is appended to the codes of all neighbors of the selected vertex that are left in the graph.

Time complexity is \(O(n+m)\) where \(n\) is the number of vertices and \(m\) is the number of edges.

See [Mil2017] for more details on the algorithm.

See also

  • lex_BFS() – perform a lexicographic breadth first search (LexBFS) on the graph

  • lex_DFS() – perform a lexicographic depth first search (LexDFS) on the graph

  • lex_DOWN() – perform a lexicographic DOWN search (LexDOWN) on the graph

EXAMPLES:

A Lex UP is obviously an ordering of the vertices:

sage: g = graphs.CompleteGraph(6)
sage: len(g.lex_UP()) == g.order()
True

Lex UP ordering of the 3-sun graph:

sage: g = Graph([(1, 2), (1, 3), (2, 3), (2, 4), (2, 5), (3, 5), (3, 6), (4, 5), (5, 6)])
sage: g.lex_UP()
[1, 2, 4, 5, 6, 3]

The method also works for directed graphs:

sage: G = DiGraph([(1, 2), (2, 3), (1, 3)])
sage: G.lex_UP(initial_vertex=2)
[2, 3, 1]

Different orderings for different traversals:

sage: # needs sage.combinat
sage: G = digraphs.DeBruijn(2,3)
sage: G.lex_BFS(initial_vertex='000')
['000', '001', '100', '010', '011', '110', '101', '111']
sage: G.lex_DFS(initial_vertex='000')
['000', '001', '100', '010', '101', '110', '011', '111']
sage: G.lex_UP(initial_vertex='000')
['000', '001', '010', '101', '110', '111', '011', '100']
sage: G.lex_DOWN(initial_vertex='000')
['000', '001', '100', '011', '010', '110', '111', '101']
lexicographic_product(other)#

Return the lexicographic product of self and other.

The lexicographic product of \(G\) and \(H\) is the graph \(L\) with vertex set \(V(L)=V(G)\times V(H)\), and \(((u,v), (w,x))\) is an edge iff :

  • \((u, w)\) is an edge of \(G\), or

  • \(u = w\) and \((v, x)\) is an edge of \(H\).

EXAMPLES:

sage: Z = graphs.CompleteGraph(2)
sage: C = graphs.CycleGraph(5)
sage: L = C.lexicographic_product(Z); L
Graph on 10 vertices
sage: L.plot()                      # long time                             # needs sage.plot
Graphics object consisting of 36 graphics primitives
sage: D = graphs.DodecahedralGraph()
sage: P = graphs.PetersenGraph()
sage: L = D.lexicographic_product(P); L
Graph on 200 vertices
sage: L.plot()                      # long time                             # needs sage.plot
Graphics object consisting of 3501 graphics primitives
line_graph(g, labels=True)#

Return the line graph of the (di)graph g.

INPUT:

  • labels – boolean (default: True); whether edge labels should be taken in consideration. If labels=True, the vertices of the line graph will be triples (u,v,label), and pairs of vertices otherwise.

The line graph of an undirected graph G is an undirected graph H such that the vertices of H are the edges of G and two vertices e and f of H are adjacent if e and f share a common vertex in G. In other words, an edge in H represents a path of length 2 in G.

The line graph of a directed graph G is a directed graph H such that the vertices of H are the edges of G and two vertices e and f of H are adjacent if e and f share a common vertex in G and the terminal vertex of e is the initial vertex of f. In other words, an edge in H represents a (directed) path of length 2 in G.

Note

As a Graph object only accepts hashable objects as vertices (and as the vertices of the line graph are the edges of the graph), this code will fail if edge labels are not hashable. You can also set the argument labels=False to ignore labels.

See also

EXAMPLES:

sage: g = graphs.CompleteGraph(4)
sage: h = g.line_graph()
sage: h.vertices(sort=True)
[(0, 1, None),
 (0, 2, None),
 (0, 3, None),
 (1, 2, None),
 (1, 3, None),
 (2, 3, None)]
sage: h.am()                                                                    # needs sage.modules
[0 1 1 1 1 0]
[1 0 1 1 0 1]
[1 1 0 0 1 1]
[1 1 0 0 1 1]
[1 0 1 1 0 1]
[0 1 1 1 1 0]
sage: h2 = g.line_graph(labels=False)
sage: h2.vertices(sort=True)
[(0, 1), (0, 2), (0, 3), (1, 2), (1, 3), (2, 3)]
sage: h2.am() == h.am()                                                         # needs sage.modules
True
sage: g = DiGraph([[1..4], lambda i,j: i < j])
sage: h = g.line_graph()
sage: h.vertices(sort=True)
[(1, 2, None),
 (1, 3, None),
 (1, 4, None),
 (2, 3, None),
 (2, 4, None),
 (3, 4, None)]
sage: h.edges(sort=True)
[((1, 2, None), (2, 3, None), None),
 ((1, 2, None), (2, 4, None), None),
 ((1, 3, None), (3, 4, None), None),
 ((2, 3, None), (3, 4, None), None)]
longest_path(s, t=None, use_edge_labels=None, algorithm=False, solver='MILP', verbose=None, integrality_tolerance=0)#

Return a longest path of self.

INPUT:

  • s – a vertex (default: None); forces the source of the path (the method then returns the longest path starting at s). The argument is set to None by default, which means that no constraint is set upon the first vertex in the path.

  • t – a vertex (default: None); forces the destination of the path (the method then returns the longest path ending at t). The argument is set to None by default, which means that no constraint is set upon the last vertex in the path.

  • use_edge_labels – boolean (default: False); whether to compute a path with maximum weight where the weight of an edge is defined by its label (a label set to None or {} being considered as a weight of \(1\)), or to compute a path with the longest possible number of edges (i.e., edge weights are set to 1)

  • algorithm – string (default: "MILP"); the algorithm to use among "MILP" and "backtrack". Two remarks on this respect:

    • While the MILP formulation returns an exact answer, the backtrack algorithm is a randomized heuristic.

    • As the backtrack algorithm does not support edge weighting, setting use_edge_labels=True will force the use of the MILP algorithm.

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

Note

The length of a path is assumed to be the number of its edges, or the sum of their labels (when use_edge_labels == True).

OUTPUT:

A subgraph of self corresponding to a (directed if self is directed) longest path. If use_edge_labels == True, a pair weight, path is returned.

ALGORITHM:

Mixed Integer Linear Programming (this problem is known to be NP-Hard).

EXAMPLES:

Petersen’s graph being hypohamiltonian, it has a longest path of length \(n - 2\):

sage: g = graphs.PetersenGraph()
sage: lp = g.longest_path()                                                 # needs sage.numerical.mip
sage: lp.order() >= g.order() - 2                                           # needs sage.numerical.mip
True

The heuristic totally agrees:

sage: g = graphs.PetersenGraph()
sage: p = g.longest_path(algorithm="backtrack").edges(sort=True, labels=False)
sage: len(p)
9
../../_images/generic_graph-1.svg

Let us compute the longest path on a random graph with random weights, and ensure the resulting graph is indeed a path:

sage: g = graphs.RandomGNP(15, 0.3)
sage: for u, v in g.edge_iterator(labels=False):
....:     g.set_edge_label(u, v, random())
sage: lp = g.longest_path()                                                 # needs sage.numerical.mip
sage: (not lp.is_forest() or not max(lp.degree()) <= 2                      # needs sage.numerical.mip
....:  or not lp.is_connected())
False
loop_edges(labels=True)#

Return a list of all loops in the (di)graph

INPUT:

  • labels – boolean (default: True); whether returned edges have labels ((u,v,l)) or not ((u,v))

EXAMPLES:

sage: G = Graph(loops=True); G
Looped graph on 0 vertices
sage: G.has_loops()
False
sage: G.allows_loops()
True
sage: G.add_edges([(0, 0), (1, 1), (2, 2), (3, 3), (2, 3)])
sage: G.loop_edges()
[(0, 0, None), (1, 1, None), (2, 2, None), (3, 3, None)]
sage: G.loop_edges(labels=False)
[(0, 0), (1, 1), (2, 2), (3, 3)]
sage: G.allows_loops()
True
sage: G.has_loops()
True
sage: G.allow_loops(False)
sage: G.has_loops()
False
sage: G.loop_edges()
[]
sage: G.edges(sort=True)
[(2, 3, None)]

sage: D = DiGraph(loops=True); D
Looped digraph on 0 vertices
sage: D.has_loops()
False
sage: D.allows_loops()
True
sage: D.add_edge((0, 0))
sage: D.has_loops()
True
sage: D.loops()
[(0, 0, None)]
sage: D.allow_loops(False); D
Digraph on 1 vertex
sage: D.has_loops()
False
sage: D.edges(sort=True)
[]

sage: G = graphs.PetersenGraph()
sage: G.loops()
[]
sage: D = DiGraph(4, loops=True)
sage: D.add_edges([(0, 0), (1, 1), (2, 2), (3, 3), (2, 3)])
sage: D.loop_edges()
[(0, 0, None), (1, 1, None), (2, 2, None), (3, 3, None)]
sage: G = Graph(4, loops=True, multiedges=True, sparse=True)
sage: G.add_edges((i, i) for i in range(4))
sage: G.loop_edges()
[(0, 0, None), (1, 1, None), (2, 2, None), (3, 3, None)]
sage: G.add_edges([(0, 0), (1, 1)])
sage: G.loop_edges(labels=False)
[(0, 0), (0, 0), (1, 1), (1, 1), (2, 2), (3, 3)]
loop_vertices()#

Return a list of vertices with loops

EXAMPLES:

sage: G = Graph({0: [0], 1: [1, 2, 3], 2: [3]}, loops=True)
sage: G.loop_vertices()
[0, 1]
loops(labels=True)#

Return a list of all loops in the (di)graph

INPUT:

  • labels – boolean (default: True); whether returned edges have labels ((u,v,l)) or not ((u,v))

EXAMPLES:

sage: G = Graph(loops=True); G
Looped graph on 0 vertices
sage: G.has_loops()
False
sage: G.allows_loops()
True
sage: G.add_edges([(0, 0), (1, 1), (2, 2), (3, 3), (2, 3)])
sage: G.loop_edges()
[(0, 0, None), (1, 1, None), (2, 2, None), (3, 3, None)]
sage: G.loop_edges(labels=False)
[(0, 0), (1, 1), (2, 2), (3, 3)]
sage: G.allows_loops()
True
sage: G.has_loops()
True
sage: G.allow_loops(False)
sage: G.has_loops()
False
sage: G.loop_edges()
[]
sage: G.edges(sort=True)
[(2, 3, None)]

sage: D = DiGraph(loops=True); D
Looped digraph on 0 vertices
sage: D.has_loops()
False
sage: D.allows_loops()
True
sage: D.add_edge((0, 0))
sage: D.has_loops()
True
sage: D.loops()
[(0, 0, None)]
sage: D.allow_loops(False); D
Digraph on 1 vertex
sage: D.has_loops()
False
sage: D.edges(sort=True)
[]

sage: G = graphs.PetersenGraph()
sage: G.loops()
[]
sage: D = DiGraph(4, loops=True)
sage: D.add_edges([(0, 0), (1, 1), (2, 2), (3, 3), (2, 3)])
sage: D.loop_edges()
[(0, 0, None), (1, 1, None), (2, 2, None), (3, 3, None)]
sage: G = Graph(4, loops=True, multiedges=True, sparse=True)
sage: G.add_edges((i, i) for i in range(4))
sage: G.loop_edges()
[(0, 0, None), (1, 1, None), (2, 2, None), (3, 3, None)]
sage: G.add_edges([(0, 0), (1, 1)])
sage: G.loop_edges(labels=False)
[(0, 0), (0, 0), (1, 1), (1, 1), (2, 2), (3, 3)]
max_cut(value_only, use_edge_labels=True, vertices=False, solver=False, verbose=None, integrality_tolerance=0)#

Return a maximum edge cut of the graph.

For more information, see the Wikipedia article Maximum_cut.

INPUT:

  • value_only – boolean (default: False); whether to return only the size of the maximum edge cut, or to also return the list of edges of the maximum edge cut

  • use_edge_labels – boolean (default: False); whether to compute a weighted maximum cut where the weight of an edge is defined by its label (if an edge has no label, \(1\) is assumed), or to compute a cut of maximum cardinality (i.e., edge weights are set to 1)

  • vertices – boolean (default: False); whether to return the two sets of vertices that are disconnected by the cut. This implies value_only=False.

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

EXAMPLES:

Quite obviously, the max cut of a bipartite graph is the number of edges, and the two sets of vertices are the two sides:

sage: # needs sage.numerical.mip
sage: g = graphs.CompleteBipartiteGraph(5,6)
sage: [ value, edges, [ setA, setB ]] = g.max_cut(vertices=True)
sage: value == 5*6
True
sage: bsetA, bsetB = map(list, g.bipartite_sets())
sage: ((bsetA == setA and bsetB == setB)
....:  or (bsetA == setB and bsetB == setA))
True

The max cut of a Petersen graph:

sage: g = graphs.PetersenGraph()
sage: g.max_cut()                                                            # needs sage.numerical.mip
12
maximum_leaf_number(G, solver=None, verbose=0, integrality_tolerance=0.001)#

Return the maximum leaf number of the graph.

The maximum leaf number is the maximum possible number of leaves of a spanning tree of \(G\). This is also the cardinality of the complement of a minimum connected dominating set. See the Wikipedia article Connected_dominating_set.

The MLN of a graph with less than 2 vertices is 0, while the MLN of a connected graph with 2 or 3 vertices is 1 or 2 respectively.

INPUT:

  • G – a Graph

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

EXAMPLES:

Empty graph:

sage: G = Graph()
sage: G.maximum_leaf_number()
0

Petersen graph:

sage: G = graphs.PetersenGraph()
sage: G.maximum_leaf_number()
6
merge_vertices(vertices)#

Merge vertices.

This function replaces a set \(S\) of vertices by a single vertex \(v_{new}\), such that the edge \(uv_{new}\) exists if and only if \(\exists v'\in S: (u,v')\in G\).

The new vertex is named after the first vertex in the list given in argument. If this first name is \(None\), a new vertex is created.

In the case of multigraphs, the multiplicity is preserved.

INPUT:

  • vertices – the list of vertices to be merged

Note

If u and v are distinct vertices in vertices, any edges between u and v will be lost.

EXAMPLES:

sage: g = graphs.CycleGraph(3)
sage: g.merge_vertices([0, 1])
sage: g.edges(sort=True)
[(0, 2, None)]

sage: P = graphs.PetersenGraph()
sage: P.merge_vertices([5, 7])
sage: P.vertices(sort=True)
[0, 1, 2, 3, 4, 5, 6, 8, 9]

When the first vertex in vertices is None, a new vertex is created:

sage: g = graphs.CycleGraph(5)
sage: g.vertices(sort=True)
[0, 1, 2, 3, 4]
sage: g.merge_vertices([None, 1, 3])
sage: g.edges(sort=True, labels=False)
[(0, 4), (0, 5), (2, 5), (4, 5)]

With a Multigraph

sage: g = graphs.CycleGraph(3)
sage: g.allow_multiple_edges(True)
sage: g.merge_vertices([0, 1])
sage: g.edges(sort=True, labels=False)
[(0, 2), (0, 2)]
min_spanning_tree(weight_function=None, algorithm='Prim_Boost', starting_vertex=None, check=False, by_weight=False, check_weight=True)#

Return the edges of a minimum spanning tree.

At the moment, no algorithm for directed graph is implemented: if the graph is directed, a minimum spanning tree of the corresponding undirected graph is returned.

We expect all weights of the graph to be convertible to float. Otherwise, an exception is raised.

INPUT:

  • algorithm – string (default: "Prim_Boost"); the algorithm to use in computing a minimum spanning tree of G. The following algorithms are supported:

    • "Prim_Boost" – Prim’s algorithm (Boost implementation)

    • "Prim_fringe" – a variant of Prim’s algorithm that ignores the labels on the edges

    • "Prim_edge" – a variant of Prim’s algorithm

    • "Kruskal" – Kruskal’s algorithm

    • "Filter_Kruskal" – a variant of Kruskal’s algorithm [OSS2009]

    • "Kruskal_Boost" – Kruskal’s algorithm (Boost implementation)

    • "Boruvka" – Boruvka’s algorithm

    • NetworkX – uses NetworkX’s minimum spanning tree implementation

  • starting_vertex – a vertex (default: None); the vertex from which to begin the search for a minimum spanning tree (available only for Prim_fringe and Prim_edge).

  • check – boolean (default: False); whether to first perform sanity checks on the input graph G. If appropriate, check is passed on to any minimum spanning tree functions that are invoked from the current method. See the documentation of the corresponding functions for details on what sort of sanity checks will be performed.

  • weight_function – function (default: None); a function that takes as input an edge (u, v, l) and outputs its weight. If not None, by_weight is automatically set to True. If None and by_weight is True, we use the edge label l , if l is not None, else 1 as a weight. The weight_function can be used to transform the label into a weight (note that, if the weight returned is not convertible to a float, an error is raised)

  • by_weight – boolean (default: False); if True, the edges in the graph are weighted, otherwise all edges have weight 1

  • check_weight – boolean (default: True); whether to check that the weight_function outputs a number for each edge.

OUTPUT:

The edges of a minimum spanning tree of G, if one exists, otherwise returns the empty list.

EXAMPLES:

Kruskal’s algorithm:

sage: g = graphs.CompleteGraph(5)
sage: len(g.min_spanning_tree())
4
sage: weight = lambda e: 1 / ((e[0] + 1) * (e[1] + 1))
sage: E = g.min_spanning_tree(weight_function=weight)
sage: T = Graph(E)
sage: set(g) == set(T) and T.order() == T.size() + 1 and T.is_tree()
True
sage: sum(map(weight, E))
5/12
sage: E = g.min_spanning_tree(weight_function=weight,
....:                         algorithm='Kruskal_Boost')
sage: Graph(E).is_tree(); sum(map(weight, E))
True
5/12
sage: g = graphs.PetersenGraph()
sage: g.allow_multiple_edges(True)
sage: g.add_edges(g.edge_iterator())
sage: T = Graph(g.min_spanning_tree())
sage: set(g) == set(T) and T.order() == T.size() + 1 and T.is_tree()
True

Boruvka’s algorithm:

sage: sorted(g.min_spanning_tree(algorithm='Boruvka'))
[(0, 1, None), (0, 4, None), (0, 5, None), (1, 2, None), (1, 6, None),
 (2, 3, None), (2, 7, None),  (3, 8, None), (4, 9, None)]

Prim’s algorithm:

sage: g = graphs.CompleteGraph(5)
sage: for algo in ['Prim_edge', 'Prim_fringe', 'Prim_Boost']:
....:     E = g.min_spanning_tree(algorithm=algo, weight_function=weight)
....:     T = Graph(E)
....:     print(set(g) == set(T) and T.order() == T.size() + 1 and T.is_tree())
True
True
True

NetworkX algorithm:

sage: sorted(g.min_spanning_tree(algorithm='NetworkX'))                     # needs networkx
[(0, 1, None), (0, 2, None), (0, 3, None), (0, 4, None)]

More complicated weights:

sage: G = Graph([(0, 1, {'name': 'a', 'weight': 1}),
....:            (0, 2, {'name': 'b', 'weight': 3}),
....:            (1, 2, {'name': 'b', 'weight': 1})])
sage: sorted(G.min_spanning_tree(algorithm='Boruvka',
....:                            weight_function=lambda e: e[2]['weight']))
[(0, 1, {'name': 'a', 'weight': 1}), (1, 2, {'name': 'b', 'weight': 1})]

If the graph is not weighted, edge labels are not considered, even if they are numbers:

sage: g = Graph([(1, 2, 1), (1, 3, 2), (2, 3, 1)])
sage: sorted(g.min_spanning_tree(algorithm='Boruvka'))
[(1, 2, 1), (1, 3, 2)]

In order to use weights, we need either to set variable weighted to True, or to specify a weight function or set by_weight to True:

sage: g.weighted(True)
sage: Graph(g.min_spanning_tree()).edges(sort=True)
[(1, 2, 1), (2, 3, 1)]
sage: g.weighted(False)
sage: Graph(g.min_spanning_tree()).edges(sort=True)
[(1, 2, 1), (1, 3, 2)]
sage: Graph(g.min_spanning_tree(by_weight=True)).edges(sort=True)
[(1, 2, 1), (2, 3, 1)]
sage: Graph(g.min_spanning_tree(weight_function=lambda e: e[2])).edges(sort=True)
[(1, 2, 1), (2, 3, 1)]

Note that the order of the vertices on each edge is not guaranteed and may differ from an algorithm to the other:

sage: g.weighted(True)
sage: sorted(g.min_spanning_tree())
[(2, 1, 1), (3, 2, 1)]
sage: sorted(g.min_spanning_tree(algorithm='Boruvka'))
[(1, 2, 1), (2, 3, 1)]
sage: Graph(g.min_spanning_tree()).edges(sort=True)
[(1, 2, 1), (2, 3, 1)]
minimum_cycle_basis(algorithm=None, weight_function=None, by_weight=False, check_weight=True)#

Return a minimum weight cycle basis of the graph.

A cycle basis is a list of cycles (list of vertices forming a cycle) of self. Note that the vertices are not necessarily returned in the order in which they appear in the cycle.

A minimum weight cycle basis is a cycle basis that minimizes the sum of the weights (length for unweighted graphs) of its cycles.

Not implemented for directed graphs and multigraphs.

INPUT:

  • algorithm – string (default: None); algorithm to use:

    • If algorithm = "NetworkX", use networkx implementation

    • If algorithm = None, use Sage Cython implementation

  • weight_function – function (default: None); a function that takes as input an edge (u, v, l) and outputs its weight. If not None, by_weight is automatically set to True. If None and by_weight is True, we use the edge label l, if l is not None, else 1 as a weight.

  • by_weight – boolean (default: False); if True, the edges in the graph are weighted, otherwise all edges have weight 1

  • check_weight – boolean (default: True); whether to check that the weight_function outputs a number for each edge.

EXAMPLES:

sage: g = Graph([(1, 2, 3), (2, 3, 5), (3, 4, 8), (4, 1, 13),
....:            (1, 3, 250), (5, 6, 9), (6, 7, 17), (7, 5, 20)])
sage: sorted(g.minimum_cycle_basis(by_weight=True))
[[1, 2, 3], [1, 2, 3, 4], [5, 6, 7]]
sage: sorted(g.minimum_cycle_basis(by_weight=False))
[[1, 2, 3], [1, 3, 4], [5, 6, 7]]
sage: sorted(g.minimum_cycle_basis(by_weight=True, algorithm='NetworkX'))   # needs networkx, random (changes in networkx 3.2)
[[1, 2, 3], [1, 2, 3, 4], [5, 6, 7]]
sage: g.minimum_cycle_basis(by_weight=False, algorithm='NetworkX')          # needs networkx, random (changes in networkx 3.2)
[[1, 2, 3], [1, 3, 4], [5, 6, 7]]
sage: g = Graph([(1, 2), (2, 3), (3, 4), (4, 5), (5, 1), (5, 3)])
sage: sorted(g.minimum_cycle_basis(by_weight=False))
[[1, 2, 3, 5], [3, 4, 5]]
sage: sorted(g.minimum_cycle_basis(by_weight=False, algorithm='NetworkX'))  # needs networkx, random (changes in networkx 3.2)
[[1, 2, 3, 5], [3, 4, 5]]
multicommodity_flow(terminals, integer, use_edge_labels=True, vertex_bound=False, solver=False, verbose=None, integrality_tolerance=0)#

Solve a multicommodity flow problem.

In the multicommodity flow problem, we are given a set of pairs \((s_i, t_i)\), called terminals meaning that \(s_i\) is willing some flow to \(t_i\).

Even though it is a natural generalisation of the flow problem this version of it is NP-Complete to solve when the flows are required to be integer.

For more information, see the Wikipedia article Multi-commodity_flow_problem.

INPUT:

  • terminals – a list of pairs \((s_i, t_i)\) or triples \((s_i, t_i, w_i)\) representing a flow from \(s_i\) to \(t_i\) of intensity \(w_i\). When the pairs are of size \(2\), an intensity of \(1\) is assumed.

  • integer boolean (default: True); whether to require an integer multicommodity flow

  • use_edge_labels – boolean (default: False); whether to compute a multicommodity flow where each edge has a capacity defined by its label (if an edge has no label, capacity \(1\) is assumed), or to use default edge capacity of \(1\)

  • vertex_bound – boolean (default: False); whether to require that a vertex can stand at most \(1\) commodity of flow through it of intensity \(1\). Terminals can obviously still send or receive several units of flow even though vertex_bound is set to True, as this parameter is meant to represent topological properties.

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

    Only useful when parameter ìnteger is True.

ALGORITHM:

(Mixed Integer) Linear Program, depending on the value of integer.

EXAMPLES:

An easy way to obtain a satisfiable multicommodity flow is to compute a matching in a graph, and to consider the paired vertices as terminals

sage: g = graphs.PetersenGraph()
sage: matching = [(u,v) for u,v,_ in g.matching()]                          # needs networkx
sage: h = g.multicommodity_flow(matching)                                   # needs networkx
sage: len(h)                                                                # needs networkx
5

We could also have considered g as symmetric and computed the multicommodity flow in this version instead. In this case, however edges can be used in both directions at the same time:

sage: h = DiGraph(g).multicommodity_flow(matching)                          # needs networkx
sage: len(h)                                                                # needs networkx
5

An exception is raised when the problem has no solution

sage: h = g.multicommodity_flow([(u,v,3) for u,v in matching])              # needs networkx
Traceback (most recent call last):
...
EmptySetError: the multicommodity flow problem has no solution
multiple_edges(to_undirected=False, labels=True, sort=False, key=None)#

Return any multiple edges in the (di)graph.

INPUT:

  • to_undirected – boolean (default: False)

  • labels – boolean (default: True); whether to include labels

  • sort – boolean (default: False); whether to sort the result

  • key – a function (default: None); a function that takes an edge as its one argument and returns a value that can be used for comparisons in the sorting algorithm (we must have sort=True)

EXAMPLES:

sage: G = Graph(multiedges=True, sparse=True); G
Multi-graph on 0 vertices
sage: G.has_multiple_edges()
False
sage: G.allows_multiple_edges()
True
sage: G.add_edges([(0, 1)] * 3)
sage: G.has_multiple_edges()
True
sage: G.multiple_edges(sort=True)
[(0, 1, None), (0, 1, None), (0, 1, None)]
sage: G.allow_multiple_edges(False); G
Graph on 2 vertices
sage: G.has_multiple_edges()
False
sage: G.edges(sort=True)
[(0, 1, None)]

sage: D = DiGraph(multiedges=True, sparse=True); D
Multi-digraph on 0 vertices
sage: D.has_multiple_edges()
False
sage: D.allows_multiple_edges()
True
sage: D.add_edges([(0, 1)] * 3)
sage: D.has_multiple_edges()
True
sage: D.multiple_edges(sort=True)
[(0, 1, None), (0, 1, None), (0, 1, None)]
sage: D.allow_multiple_edges(False); D
Digraph on 2 vertices
sage: D.has_multiple_edges()
False
sage: D.edges(sort=True)
[(0, 1, None)]

sage: G = DiGraph({1: {2: 'h'}, 2: {1: 'g'}}, sparse=True)
sage: G.has_multiple_edges()
False
sage: G.has_multiple_edges(to_undirected=True)
True
sage: G.multiple_edges()
[]
sage: G.multiple_edges(to_undirected=True, sort=True)
[(1, 2, 'h'), (2, 1, 'g')]

Using the key argument to order multiple edges of incomparable types (see github issue #35903):

sage: G = Graph([('A', 'B', 3), (1, 2, 1), ('A', 'B', 4), (1, 2, 2)], multiedges=True)
sage: G.multiple_edges(sort=True)
Traceback (most recent call last):
...
TypeError: unsupported operand parent(s) for <: 'Integer Ring' and '<class 'str'>'
sage: G.multiple_edges(labels=False, sort=True, key=str)
[('A', 'B'), ('A', 'B'), (1, 2), (1, 2)]
sage: G.multiple_edges(sort=True, key=str)
[('A', 'B', 3), ('A', 'B', 4), (1, 2, 1), (1, 2, 2)]
sage: G.multiple_edges(labels=True, sort=True, key=lambda e:e[2])
[(1, 2, 1), (1, 2, 2), ('A', 'B', 3), ('A', 'B', 4)]
sage: G.multiple_edges(labels=False, sort=True, key=lambda e:e[2])
Traceback (most recent call last):
...
IndexError: tuple index out of range
multiway_cut(vertices, value_only, use_edge_labels=False, solver=False, verbose=None, integrality_tolerance=0)#

Return a minimum edge multiway cut.

A multiway cut for a vertex set \(S\) in a graph or a digraph \(G\) is a set \(C\) of edges such that any two vertices \(u,v\) in \(S\) are disconnected when removing the edges of \(C\) from \(G\). ( cf. http://www.d.kth.se/~viggo/wwwcompendium/node92.html )

Such a cut is said to be minimum when its cardinality (or weight) is minimum.

INPUT:

  • vertices – iterable; the set of vertices

  • value_only – boolean (default: False); whether to return only the size of the minimum multiway cut, or to return the list of edges of the multiway cut

  • use_edge_labels – boolean (default: False); whether to compute a weighted minimum multiway cut where the weight of an edge is defined by its label (if an edge has no label, \(1\) is assumed), or to compute a cut of minimum cardinality (i.e., edge weights are set to 1)

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

EXAMPLES:

Of course, a multiway cut between two vertices correspond to a minimum edge cut:

sage: g = graphs.PetersenGraph()
sage: g.edge_cut(0,3) == g.multiway_cut([0,3], value_only=True)             # needs sage.numerical.mip
True

As Petersen’s graph is \(3\)-regular, a minimum multiway cut between three vertices contains at most \(2\times 3\) edges (which could correspond to the neighborhood of 2 vertices):

sage: g.multiway_cut([0,3,9], value_only=True) == 2*3                       # needs sage.numerical.mip
True

In this case, though, the vertices are an independent set. If we pick instead vertices \(0,9,\) and \(7\), we can save \(4\) edges in the multiway cut:

sage: g.multiway_cut([0,7,9], value_only=True) == 2*3 - 1                   # needs sage.numerical.mip
True

This example, though, does not work in the directed case anymore, as it is not possible in Petersen’s graph to mutualise edges:

sage: g = DiGraph(g)
sage: g.multiway_cut([0,7,9], value_only=True) == 3*3                       # needs sage.numerical.mip
True

Of course, a multiway cut between the whole vertex set contains all the edges of the graph:

sage: C = g.multiway_cut(g.vertices(sort=False))                            # needs sage.numerical.mip
sage: set(C) == set(g.edges(sort=False))                                    # needs sage.numerical.mip
True
name(new=None)#

Return or set the graph’s name.

INPUT:

  • new – string (default: None); by default (new == None), the method returns the name of the graph. When name is set, the string representation of that object becomes the new name of the (di)graph (new == '' removes any name).

EXAMPLES:

sage: d = {0: [1,4,5], 1: [2,6], 2: [3,7], 3: [4,8], 4: [9], 5: [7, 8], 6: [8,9], 7: [9]}
sage: G = Graph(d); G
Graph on 10 vertices
sage: G.name("Petersen Graph"); G
Petersen Graph: Graph on 10 vertices
sage: G.name(new=""); G
Graph on 10 vertices
sage: G.name()
''
sage: G.name(42); G
42: Graph on 10 vertices
sage: G.name()
'42'
neighbor_iterator(vertex, closed=False)#

Return an iterator over neighbors of vertex.

When closed is set to True, the returned iterator also contains vertex.

INPUT:

  • vertex – a vertex of self

  • closed – a boolean (default: False); whether to return the closed neighborhood of vertex, i.e., including vertex, or the open neighborhood in which vertex is included only if there is a loop on that vertex.

EXAMPLES:

sage: G = graphs.PetersenGraph()
sage: for i in G.neighbor_iterator(0):
....:     print(i)
1
4
5
sage: D = G.to_directed()
sage: for i in D.neighbor_iterator(0):
....:     print(i)
1
4
5
sage: D = DiGraph({0: [1, 2], 3: [0]})
sage: sorted(D.neighbor_iterator(0))
[1, 2, 3]
sage: g = graphs.CubeGraph(3)
sage: sorted(g.neighbor_iterator('010', closed=True))
['000', '010', '011', '110']
sage: g = Graph(3, loops = True)
sage: g.add_edge(0,1)
sage: g.add_edge(0,0)
sage: list(g.neighbor_iterator(0, closed=True))
[0, 1]
sage: list(g.neighbor_iterator(2, closed=True))
[2]
neighbors(vertex, closed=False)#

Return a list of neighbors (in and out if directed) of vertex.

G[vertex] also works. When closed is set to True, the returned iterator also contains vertex.

INPUT:

  • vertex – a vertex of self

  • closed – a boolean (default: False); whether to return the closed neighborhood of vertex, i.e., including vertex, or the open neighborhood in which vertex is included only if there is a loop on that vertex.

EXAMPLES:

sage: P = graphs.PetersenGraph()
sage: sorted(P.neighbors(3))
[2, 4, 8]
sage: sorted(P[4])
[0, 3, 9]
sage: sorted(P.neighbors(3, closed=True))
[2, 3, 4, 8]
networkx_graph(weight_function=None)#

Return a new NetworkX graph from the Sage graph.

INPUT:

  • weight_function – function (default: None); a function that takes as input an edge (u, v, l) and outputs its weight.

EXAMPLES:

sage: G = graphs.TetrahedralGraph()
sage: N = G.networkx_graph()                                                # needs networkx
sage: type(N)                                                               # needs networkx
<class 'networkx.classes.graph.Graph'>

sage: def weight_fn(e):
....:     return e[2]
sage: G1 = Graph([(1,2,1), (1,3,4), (2,3,3), (3,4,4)])
sage: H = G1.networkx_graph(weight_function=weight_fn)                      # needs networkx
sage: H.edges(data=True)                                                    # needs networkx
EdgeDataView([(1, 2, {'weight': 1}), (1, 3, {'weight': 4}),
              (2, 3, {'weight': 3}), (3, 4, {'weight': 4})])
sage: G2 = DiGraph([(1,2,1), (1,3,4), (2,3,3), (3,4,4), (3,4,5)],
....:              multiedges=True)
sage: H = G2.networkx_graph(weight_function=weight_fn)                      # needs networkx
sage: H.edges(data=True)                                                    # needs networkx
OutMultiEdgeDataView([(1, 2, {'weight': 1}), (1, 3, {'weight': 4}),
                      (2, 3, {'weight': 3}), (3, 4, {'weight': 5}),
                      (3, 4, {'weight': 4})])
nowhere_zero_flow(k, solver=None, verbose=None, integrality_tolerance=0)#

Return a k-nowhere zero flow of the (di)graph.

A flow on a graph \(G = (V, E)\) is a pair \((D, f)\) such that \(D\) is an orientation of \(G\) and \(f\) is a function on \(E\) satisfying

\[\sum_{u \in N^-_D(v)} f(uv) = \sum_{w \in N^+_D(v)} f(vw), \ \forall v \in V.\]

A nowhere zero flow on a graph \(G = (V, E)\) is a flow \((D, f)\) such that \(f(e) \neq 0\) for every \(e \in E\). For a positive integer \(k\), a \(k\)-flow on a graph \(G = (V, E)\) is a flow \((D, f)\) such that \(f: E \to Z\) and \(-(k - 1) \leq f(e) \leq k - 1\) for every \(e \in E\). A \(k\)-flow is positive if \(f(e) > 0\) for every \(e \in E\). A \(k\)-flow which is nowhere zero is called a \(k\)-nowhere zero flow (or \(k\)-NZF).

The following are equivalent.

  • \(G\) admits a positive \(k\)-flow.

  • \(G\) admits a \(k\)-NZF.

  • Every orientation of \(G\) admits a \(k\)-NZF.

Furthermore, a (di)graph admits a \(k\)-NZF if and only if it is bridgeless and every bridgeless graph admits a \(6\)-NZF [Sey1981]. See the Wikipedia article Nowhere-zero_flow for more details.

ALGORITHM:

If self is not directed, we search for a \(k\)-NZF on any orientation of self and then build a positive \(k\)-NZF by reverting edges with negative flow.

INPUT:

  • k – integer (default: 6); when set to a positive integer \(\geq 2\), search for a \(k\)-nowhere zero flow

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

OUTPUT:

A digraph with flow values stored as edge labels if a \(k\)-nowhere zero flow is found. If self is undirected, the edges of this digraph indicate the selected orientation. If no feasible solution is found, an error is raised.

EXAMPLES:

The Petersen graph admits a (positive) 5-nowhere zero flow, but no 4-nowhere zero flow:

sage: g = graphs.PetersenGraph()
sage: h = g.nowhere_zero_flow(k=5)                                          # needs sage.numerical.mip
sage: sorted(set(h.edge_labels()))                                          # needs sage.numerical.mip
[1, 2, 3, 4]
sage: h = g.nowhere_zero_flow(k=3)                                          # needs sage.numerical.mip
Traceback (most recent call last):
...
EmptySetError: the problem has no feasible solution

The de Bruijn digraph admits a 2-nowhere zero flow:

sage: g = digraphs.DeBruijn(2, 3)                                           # needs sage.combinat
sage: h = g.nowhere_zero_flow(k=2)                                          # needs sage.combinat sage.numerical.mip
sage: sorted(set(h.edge_labels()))                                          # needs sage.combinat sage.numerical.mip
[-1, 1]
num_edges()#

Return the number of edges.

Note that num_edges() also returns the number of edges in \(G\).

EXAMPLES:

sage: G = graphs.PetersenGraph()
sage: G.size()
15
num_faces(embedding=None)#

Return the number of faces of an embedded graph.

If no embedding is provided or stored as self._embedding, this method uses Euler’s formula (see the Wikipedia article Euler_characteristic) to determine the number of faces if the graph is planar. If the graph is not planar, an error is raised.

If an embedding is provided or stored as self._embedding, this method calls method faces() to get the list of faces induced by the embedding in each connected component of the graph. Then it returns the sum of size of these lists minus the number of connected components plus one to ensure that the external face is counted only once.

INPUT:

  • embedding – dictionary (default: None); a combinatorial embedding dictionary. Format: {v1: [v2,v3], v2: [v1], v3: [v1]} (clockwise ordering of neighbors at each vertex). If set to None (default) the method will use the embedding stored as self._embedding. If none is stored, the method will compute the set of faces from the embedding returned by is_planar() (if the graph is, of course, planar).

EXAMPLES:

sage: T = graphs.TetrahedralGraph()
sage: T.num_faces()
4

The external face of a disconnected graph is counted only once:

sage: (T + T).num_faces()
7
sage: (T + T + T).num_faces()
10

Trees and forests have a single face:

sage: T = graphs.RandomTree(10)
sage: T.num_faces()
1
sage: (T + T).num_faces()
1
num_verts()#

Return the number of vertices.

Note that len(G) and num_verts() also return the number of vertices in \(G\).

EXAMPLES:

sage: G = graphs.PetersenGraph()
sage: G.order()
10
sage: G = graphs.TetrahedralGraph()
sage: len(G)
4
number_of_loops()#

Return the number of edges that are loops

EXAMPLES:

sage: G = Graph(4, loops=True)
sage: G.add_edges([(0, 0), (1, 1), (2, 2), (3, 3), (2, 3)])
sage: G.edges(sort=True, labels=False)
[(0, 0), (1, 1), (2, 2), (2, 3), (3, 3)]
sage: G.number_of_loops()
4
sage: D = DiGraph(4, loops=True)
sage: D.add_edges([(0, 0), (1, 1), (2, 2), (3, 3), (2, 3)])
sage: D.edges(sort=True, labels=False)
[(0, 0), (1, 1), (2, 2), (2, 3), (3, 3)]
sage: D.number_of_loops()
4
odd_girth(algorithm='bfs', certificate=False)#

Return the odd girth of the graph.

The odd girth is the length of the shortest cycle of odd length in the graph (directed cycle if the graph is directed). Bipartite graphs have infinite odd girth.

INPUT:

  • algorithm – string (default: "bfs"); the algorithm to use:

    • "bfs" – BFS-based algorithm

    • any algorithm accepted by charpoly() for computation from the characteristic polynomial (see [Har1962] and [Big1993], p. 45)

  • certificate – boolean (default: False); whether to return (g, c), where g is the odd girth and c is a list of vertices of a (directed) cycle of length g in the graph, thus providing a certificate that the odd girth is at most g, or None if g is infinite. So far, this parameter is accepted only when algorithm = "bfs".

EXAMPLES:

The McGee graph has girth 7 and therefore its odd girth is 7 as well:

sage: G = graphs.McGeeGraph()                                               # needs networkx
sage: G.girth()                                                             # needs networkx
7
sage: G.odd_girth()                                                         # needs networkx
7

Any complete (directed) graph on more than 2 vertices contains a (directed) triangle and has thus odd girth 3:

sage: G = graphs.CompleteGraph(5)
sage: G.odd_girth(certificate=True)  # random
(3, [2, 1, 0])
sage: G = digraphs.Complete(5)
sage: G.odd_girth(certificate=True)  # random
(3, [1, 2, 0])

Bipartite graphs have no odd cycle and consequently have infinite odd girth:

sage: G = graphs.RandomBipartite(6, 6, .5)                                  # needs numpy
sage: G.odd_girth()                                                         # needs numpy
+Infinity
sage: G = graphs.Grid2dGraph(3, 4)
sage: G.odd_girth()
+Infinity

The odd girth of a (directed) graph with loops is 1:

sage: # needs networkx
sage: G = graphs.RandomGNP(10, .5)
sage: G.allow_loops(True)
sage: G.add_edge(0, 0)
sage: G.odd_girth()
1
sage: G = digraphs.RandomDirectedGNP(10, .5)
sage: G.allow_loops(True)
sage: G.add_edge(0, 0)
sage: G.odd_girth()
1

See also

  • girth() – return the girth of the graph.

order()#

Return the number of vertices.

Note that len(G) and num_verts() also return the number of vertices in \(G\).

EXAMPLES:

sage: G = graphs.PetersenGraph()
sage: G.order()
10
sage: G = graphs.TetrahedralGraph()
sage: len(G)
4
pagerank(alpha=0.85, personalization=None, by_weight=False, weight_function=None, check_weight=True, dangling=None, algorithm='scipy')#

Return the PageRank of the vertices of self.

PageRank is a centrality measure earlier used to rank web pages. The PageRank algorithm outputs the probability distribution that a random walker in the graph visits a vertex.

See the Wikipedia article PageRank for more information.

INPUT:

  • alpha – float (default: 0.85); damping parameter for PageRank. alpha is the click-through probability useful for preventing sinks. The probability at any step, that an imaginary surfer who is randomly clicking on links will continue is a damping factor d.

  • personalization – dict (default: None); a dictionary keyed by vertices associating to each vertex a value. The personalization can be specified for a subset of the vertices, if not specified a nodes personalization value will be taken as zero. The sum of the values must be nonzero. By default (None), a uniform distribution is used.

  • by_weight – boolean (default: False); if True, the edges in the graph are weighted, otherwise all edges have weight 1

  • weight_function – function (default: None); a function that takes as input an edge (u, v, l) and outputs its weight. If not None, by_weight is automatically set to True. If None and by_weight is True, we use the edge label l, if l is not None, else 1 as a weight.

  • check_weight – boolean (default: True); whether to check that the weight_function outputs a number for each edge.

  • dangling – dict (default: None); a dictionary keyed by a vertex the outedge of “dangling” vertices, (i.e., vertices without any outedges) points to and the dict value is the weight of that outedge. By default, dangling vertices are given outedges according to the personalization vector (uniform if not specified). It may be common to have the dangling dict to be the same as the personalization dict.

  • algorithm – string (default: None); the algorithm to use in

    computing PageRank of G. The following algorithms are supported:

    • NetworkX – uses NetworkX’s default implementation (Scipy as of 2.6)

    • "Scipy" – uses Scipy’s PageRank algorithm implementation

    • "igraph" – uses igraph’s PageRank algorithm implementation

    • "None" – uses best implementation available

OUTPUT: a dictionary containing the PageRank value of each node

Note

Parameters alpha, by_weight and weight_function are common to all algorithms. Parameters personalization and dangling are used only by algorithms NetworkX, Numpy and Scipy.

EXAMPLES:

sage: G = graphs.CycleGraph(4)
sage: G.pagerank(algorithm="Networkx")                                      # needs networkx
{0: 0.25, 1: 0.25, 2: 0.25, 3: 0.25}
sage: G.pagerank(alpha=0.50, algorithm="igraph")        # abs tol 1e-9, optional - python_igraph
{0: 0.25, 1: 0.25, 2: 0.25, 3: 0.25}
sage: G = Graph([(1, 2, 40), (2, 3, 50), (3, 4, 60),
....:            (1, 4, 70), (4, 5, 80), (5, 6, 20)])
sage: G.pagerank(algorithm="NetworkX")                  # abs tol 1e-9      # needs networkx
{1: 0.16112205885619563,
 2: 0.1619531043247219,
 3: 0.16112205885619563,
 4: 0.2374999999999999,
 5: 0.17775588228760858,
 6: 0.100546895675278}
sage: G.pagerank(algorithm="NetworkX", by_weight=True)  # abs tol 1e-9      # needs networkx
{1: 0.16459583718588994,
 2: 0.13977928595154515,
 3: 0.16539840184339605,
 4: 0.3063198690713853,
 5: 0.1700057609707141,
 6: 0.05390084497706962}
sage: G.pagerank(algorithm="Scipy")                     # abs tol 1e-9      # needs networkx scipy
{1: 0.16112205885619563,
 2: 0.1619531043247219,
 3: 0.16112205885619563,
 4: 0.2374999999999999,
 5: 0.17775588228760858,
 6: 0.100546895675278}
sage: G.pagerank(algorithm="Scipy", by_weight=True)     # abs tol 1e-9      # needs networkx scipy
{1: 0.16459583718588994,
 2: 0.13977928595154515,
 3: 0.16539840184339605,
 4: 0.3063198690713853,
 5: 0.1700057609707141,
 6: 0.05390084497706962}
sage: G.pagerank(algorithm="igraph")                    # abs tol 1e-9, optional - python_igraph
{1: 0.16112198303979128,
 2: 0.16195368558382262,
 3: 0.16112198303979125,
 4: 0.23749999999999993,
 5: 0.17775603392041744,
 6: 0.10054631441617742}
sage: G.pagerank()                                      # abs tol 1e-9      # needs networkx
{1: 0.16112205885619563,
 2: 0.1619531043247219,
 3: 0.16112205885619563,
 4: 0.2374999999999999,
 5: 0.17775588228760858,
 6: 0.100546895675278}
sage: G.pagerank(by_weight=True)                        # abs tol 1e-9      # needs networkx
{1: 0.16459583718588994,
 2: 0.13977928595154515,
 3: 0.16539840184339605,
 4: 0.3063198690713853,
 5: 0.1700057609707141,
 6: 0.05390084497706962}
planar_dual(embedding=None)#

Return the planar dual of an embedded graph.

A combinatorial embedding of a graph is a clockwise ordering of the neighbors of each vertex. From this information one can obtain the dual of a plane graph, which is what the method returns. The vertices of the dual graph correspond to faces of the primal graph.

INPUT:

  • embedding – dictionary (default: None); a combinatorial embedding dictionary. Format: {v1: [v2,v3], v2: [v1], v3: [v1]} (clockwise ordering of neighbors at each vertex). If set to None (default) the method will use the embedding stored as self._embedding. If none is stored, the method will compute the set of faces from the embedding returned by is_planar() (if the graph is, of course, planar).

EXAMPLES:

sage: C = graphs.CubeGraph(3)
sage: C.planar_dual()
Graph on 6 vertices
sage: graphs.IcosahedralGraph().planar_dual().is_isomorphic(graphs.DodecahedralGraph())
True

The planar dual of the planar dual is isomorphic to the graph itself:

sage: g = graphs.BuckyBall()
sage: g.planar_dual().planar_dual().is_isomorphic(g)
True

Todo

Implement the method for graphs that are not 3-vertex-connected, or at least have a faster 3-vertex-connectivity test (github issue #24635).

plot(**options)#

Return a Graphics object representing the (di)graph.

INPUT:

  • pos – an optional positioning dictionary

  • layout – string (default: None); specifies a kind of layout to use, takes precedence over pos

    • 'circular' – plots the graph with vertices evenly distributed on a circle

    • 'spring' – uses the traditional spring layout, using the graph’s current positions as initial positions

    • 'tree' – the (di)graph must be a tree. One can specify the root of the tree using the keyword tree_root, otherwise a root will be selected at random. Then the tree will be plotted in levels, depending on minimum distance for the root.

  • vertex_labels – boolean (default: True); whether to print vertex labels

  • edge_labels – boolean (default: False); whether to print edge labels. If True, the result of str(l) is printed on the edge for each label \(l\). Labels equal to None are not printed (to set edge labels, see set_edge_label()).

  • edge_labels_background – the color of the edge labels background. The default is “white”. To achieve a transparent background use “transparent”.

  • vertex_size – size of vertices displayed

  • vertex_shape – the shape to draw the vertices, for example "o" for circle or "s" for square. Whole list is available at https://matplotlib.org/api/markers_api.html. (Not available for multiedge digraphs.)

  • graph_border – boolean (default: False); whether to include a box around the graph

  • vertex_colors – dictionary (default: None); optional dictionary to specify vertex colors: each key is a color recognizable by matplotlib, and each corresponding entry is a list of vertices. If a vertex is not listed, it looks invisible on the resulting plot (it doesn’t get drawn).

  • edge_colors – dictionary (default: None); a dictionary specifying edge colors: each key is a color recognized by matplotlib, and each entry is a list of edges.

  • partition – a partition of the vertex set (default: None); if specified, plot will show each cell in a different color. vertex_colors takes precedence.

  • talk – boolean (default: False); if True, prints large vertices with white backgrounds so that labels are legible on slides

  • iterations – integer; how many iterations of the spring layout algorithm to go through, if applicable

  • color_by_label – a boolean or dictionary or function (default: False); whether to color each edge with a different color according to its label; the colors are chosen along a rainbow, unless they are specified by a function or dictionary mapping labels to colors; this option is incompatible with edge_color and edge_colors.

  • heights – dictionary (default: None); if specified, this is a dictionary from a set of floating point heights to a set of vertices

  • edge_style – keyword arguments passed into the edge-drawing routine. This currently only works for directed graphs, since we pass off the undirected graph to networkx

  • tree_root – a vertex (default: None); if specified, this vertex is used as the root for the layout="tree" option. Otherwise, then one is chosen at random. Ignored unless layout='tree'.

  • tree_orientation – string (default: "down"); one of “up” or “down”. If “up” (resp., “down”), then the root of the tree will appear on the bottom (resp., top) and the tree will grow upwards (resp. downwards). Ignored unless layout='tree'.

  • save_pos – boolean (default: False); save position computed during plotting

Note

  • This method supports any parameter accepted by sage.plot.graphics.Graphics.show().

  • See the documentation of the sage.graphs.graph_plot module for information and examples of how to define parameters that will be applied to all graph plots.

  • Default parameters for this method and a specific graph can also be set through the options mechanism. For more information on this different way to set default parameters, see the help of the options decorator.

  • See also the sage.graphs.graph_latex module for ways to use LaTeX to produce an image of a graph.

EXAMPLES:

sage: from sage.graphs.graph_plot import graphplot_options
sage: sorted(graphplot_options.items())
[...]

sage: from math import sin, cos, pi
sage: P = graphs.PetersenGraph()
sage: d = {'#FF0000': [0, 5], '#FF9900': [1, 6], '#FFFF00': [2, 7],
....:      '#00FF00': [3, 8], '#0000FF': [4, 9]}
sage: pos_dict = {}
sage: for i in range(5):
....:  x = float(cos(pi/2 + ((2*pi)/5)*i))
....:  y = float(sin(pi/2 + ((2*pi)/5)*i))
....:  pos_dict[i] = [x,y]
sage: for i in range(5, 10):
....:  x = float(0.5*cos(pi/2 + ((2*pi)/5)*i))
....:  y = float(0.5*sin(pi/2 + ((2*pi)/5)*i))
....:  pos_dict[i] = [x,y]
sage: pl = P.plot(pos=pos_dict, vertex_colors=d)                            # needs sage.plot
sage: pl.show()                                                             # needs sage.plot
sage: C = graphs.CubeGraph(8)
sage: P = C.plot(vertex_labels=False, vertex_size=0, graph_border=True)     # needs sage.plot
sage: P.show()                                                              # needs sage.plot
sage: G = graphs.HeawoodGraph()
sage: for u, v, l in G.edges(sort=False):
....:     G.set_edge_label(u, v, '(' + str(u) + ',' + str(v) + ')')
sage: G.plot(edge_labels=True).show()                                       # needs sage.plot
sage: D = DiGraph({0: [1, 10, 19], 1: [8, 2], 2: [3, 6], 3: [19, 4],
....:              4: [17, 5], 5: [6, 15], 6: [7], 7: [8, 14],
....:              8: [9], 9: [10, 13], 10: [11], 11: [12, 18],
....:              12: [16, 13], 13: [14], 14: [15], 15: [16],
....:              16: [17], 17: [18], 18: [19]}, sparse=True)
sage: for u,v,l in D.edges(sort=False):
....:     D.set_edge_label(u, v, '(' + str(u) + ',' + str(v) + ')')
sage: D.plot(edge_labels=True, layout='circular').show()                    # needs sage.plot
sage: # needs sage.plot
sage: from sage.plot.colors import rainbow
sage: C = graphs.CubeGraph(5)
sage: R = rainbow(5)
sage: edge_colors = {R[i]: [] for i in range(5)}
sage: for u, v, l in C.edges(sort=False):
....:  for i in range(5):
....:      if u[i] != v[i]:
....:          edge_colors[R[i]].append((u, v, l))
sage: C.plot(vertex_labels=False, vertex_size=0,
....:        edge_colors=edge_colors).show()
sage: D = graphs.DodecahedralGraph()
sage: Pi = [[6,5,15,14,7], [16,13,8,2,4], [12,17,9,3,1], [0,19,18,10,11]]
sage: D.show(partition=Pi)                                                  # needs sage.plot
sage: G = graphs.PetersenGraph()
sage: G.allow_loops(True)
sage: G.add_edge(0, 0)
sage: G.show()                                                              # needs sage.plot
sage: D = DiGraph({0: [0, 1], 1: [2], 2: [3]}, loops=True)
sage: D.show()                                                              # needs sage.plot
sage: D.show(edge_colors={(0, 1, 0): [(0, 1, None), (1, 2, None)],          # needs sage.plot
....:                     (0, 0, 0): [(2, 3, None)]})
sage: pos = {0: [0.0, 1.5], 1: [-0.8, 0.3], 2: [-0.6, -0.8], 3: [0.6, -0.8], 4: [0.8, 0.3]}
sage: g = Graph({0: [1], 1: [2], 2: [3], 3: [4], 4: [0]})
sage: g.plot(pos=pos, layout='spring', iterations=0)                        # needs sage.plot
Graphics object consisting of 11 graphics primitives
sage: G = Graph()
sage: P = G.plot()                                                          # needs sage.plot
sage: P.axes()                                                              # needs sage.plot
False
sage: G = DiGraph()
sage: P = G.plot()                                                          # needs sage.plot
sage: P.axes()                                                              # needs sage.plot
False
sage: G = graphs.PetersenGraph()
sage: G.get_pos()
{0: (0.0..., 1.0...),
 1: (-0.95..., 0.30...),
 2: (-0.58..., -0.80...),
 3: (0.58..., -0.80...),
 4: (0.95..., 0.30...),
 5: (0.0..., 0.5...),
 6: (-0.47..., 0.15...),
 7: (-0.29..., -0.40...),
 8: (0.29..., -0.40...),
 9: (0.47..., 0.15...)}
sage: P = G.plot(save_pos=True, layout='spring')                            # needs sage.plot

The following illustrates the format of a position dictionary:

sage: G.get_pos()  # currently random across platforms, see #9593           # needs sage.plot
{0: [1.17..., -0.855...],
 1: [1.81..., -0.0990...],
 2: [1.35..., 0.184...],
 3: [1.51..., 0.644...],
 4: [2.00..., -0.507...],
 5: [0.597..., -0.236...],
 6: [2.04..., 0.687...],
 7: [1.46..., -0.473...],
 8: [0.902..., 0.773...],
 9: [2.48..., -0.119...]}
sage: T = list(graphs.trees(7))
sage: t = T[3]
sage: t.plot(heights={0: [0], 1: [4, 5, 1], 2: [2], 3: [3, 6]})             # needs sage.plot
Graphics object consisting of 14 graphics primitives
sage: T = list(graphs.trees(7))
sage: t = T[3]
sage: t.plot(heights={0: [0], 1: [4, 5, 1], 2: [2], 3: [3, 6]})             # needs sage.plot
Graphics object consisting of 14 graphics primitives
sage: t.set_edge_label(0, 1, -7)
sage: t.set_edge_label(0, 5, 3)
sage: t.set_edge_label(0, 5, 99)
sage: t.set_edge_label(1, 2, 1000)
sage: t.set_edge_label(3, 2, 'spam')
sage: t.set_edge_label(2, 6, 3/2)
sage: t.set_edge_label(0, 4, 66)
sage: t.plot(heights={0: [0], 1: [4, 5, 1], 2: [2], 3: [3, 6]},             # needs sage.plot
....:        edge_labels=True)
Graphics object consisting of 20 graphics primitives
sage: T = list(graphs.trees(7))
sage: t = T[3]
sage: t.plot(layout='tree')                                                 # needs sage.plot
Graphics object consisting of 14 graphics primitives
sage: t = DiGraph('JCC???@A??GO??CO??GO??')
sage: t.plot(layout='tree', tree_root=0, tree_orientation="up")             # needs sage.plot
Graphics object consisting of 22 graphics primitives
sage: D = DiGraph({0: [1, 2, 3], 2: [1, 4], 3: [0]})
sage: D.plot()                                                              # needs sage.plot
Graphics object consisting of 16 graphics primitives

sage: D = DiGraph(multiedges=True,sparse=True)
sage: for i in range(5):
....:   D.add_edge((i, i + 1, 'a'))
....:   D.add_edge((i, i - 1, 'b'))
sage: D.plot(edge_labels=True, edge_colors=D._color_by_label())             # needs sage.plot
Graphics object consisting of 34 graphics primitives
sage: D.plot(edge_labels=True, color_by_label={'a': 'blue', 'b': 'red'},    # needs sage.plot
....:        edge_style='dashed')
Graphics object consisting of 34 graphics primitives

sage: g = Graph({}, loops=True, multiedges=True, sparse=True)
sage: g.add_edges([(0, 0, 'a'), (0, 0, 'b'), (0, 1, 'c'), (0, 1, 'd'),
....:   (0, 1, 'e'), (0, 1, 'f'), (0, 1, 'f'), (2, 1, 'g'), (2, 2, 'h')])
sage: g.plot(edge_labels=True, color_by_label=True, edge_style='dashed')    # needs sage.plot
Graphics object consisting of 22 graphics primitives
sage: # needs sage.modular
sage: S = SupersingularModule(389)
sage: H = S.hecke_matrix(2)
sage: D = DiGraph(H, sparse=True)
sage: P = D.plot()                                                          # needs sage.plot
sage: G = Graph({'a': ['a','b','b','b','e'], 'b': ['c','d','e'],
....:            'c':['c','d','d','d'],'d':['e']}, sparse=True)
sage: G.show(pos={'a':[0,1],'b':[1,1],'c':[2,0],'d':[1,0],'e':[0,0]})       # needs sage.plot
plot3d(bgcolor=(1, 1, 1), vertex_colors=None, vertex_size=0.06, vertex_labels=False, edge_colors=None, edge_size=0.02, edge_size2=0.0325, pos3d=None, color_by_label=False, engine='threejs', **kwds)#

Plot a graph in three dimensions.

See also the sage.graphs.graph_latex module for ways to use LaTeX to produce an image of a graph.

INPUT:

  • bgcolor – rgb tuple (default: (1,1,1))

  • vertex_size – float (default: 0.06)

  • vertex_labels – a boolean (default: False); whether to display vertices using text labels instead of spheres

  • vertex_colors – dictionary (default: None); optional dictionary to specify vertex colors: each key is a color recognizable by tachyon (rgb tuple (default: (1,0,0))), and each corresponding entry is a list of vertices. If a vertex is not listed, it looks invisible on the resulting plot (it does not get drawn).

  • edge_colors – dictionary (default: None); a dictionary specifying edge colors: each key is a color recognized by tachyon (default: (0,0,0)), and each entry is a list of edges.

  • color_by_label – a boolean or dictionary or function (default: False) whether to color each edge with a different color according to its label; the colors are chosen along a rainbow, unless they are specified by a function or dictionary mapping labels to colors; this option is incompatible with edge_color and edge_colors.

  • edge_size – float (default: 0.02)

  • edge_size2 – float (default: 0.0325); used for Tachyon sleeves

  • pos3d – a position dictionary for the vertices

  • layout, iterations, … – layout options; see layout()

  • engine – string (default: 'threejs'); the renderer to use among:

    • 'threejs': interactive web-based 3D viewer using JavaScript and a WebGL renderer

    • 'jmol': interactive 3D viewer using Java

    • 'tachyon': ray tracer generating a static PNG image

  • xres – resolution

  • yres – resolution

  • **kwds – passed on to the rendering engine

EXAMPLES:

sage: G = graphs.CubeGraph(5)
sage: G.plot3d(iterations=500, edge_size=None, vertex_size=0.04)    # long time, needs sage.plot
Graphics3d Object

We plot a fairly complicated Cayley graph:

sage: A5 = AlternatingGroup(5); A5                                          # needs sage.groups
Alternating group of order 5!/2 as a permutation group
sage: G = A5.cayley_graph()                                                 # needs sage.groups
sage: G.plot3d(vertex_size=0.03, edge_size=0.01,    # long time             # needs sage.groups sage.plot
....:          vertex_colors={(1,1,1): list(G)}, bgcolor=(0,0,0),
....:          color_by_label=True, iterations=200)
Graphics3d Object

Some Tachyon examples:

sage: D = graphs.DodecahedralGraph()
sage: P3D = D.plot3d(engine='tachyon')                                      # needs sage.plot
sage: P3D.show()                    # long time                             # needs sage.plot
sage: G = graphs.PetersenGraph()
sage: G.plot3d(engine='tachyon',    # long time                             # needs sage.plot
....:          vertex_colors={(0,0,1): list(G)}).show()
sage: C = graphs.CubeGraph(4)
sage: C.plot3d(engine='tachyon',    # long time                             # needs sage.plot
....:          edge_colors={(0,1,0): C.edges(sort=False)},
....:          vertex_colors={(1,1,1): list(C)}, bgcolor=(0,0,0)).show()
sage: K = graphs.CompleteGraph(3)
sage: K.plot3d(engine='tachyon',    # long time                             # needs sage.plot
....:          edge_colors={(1,0,0): [(0,1,None)],
....:                       (0,1,0): [(0,2,None)],
....:                       (0,0,1): [(1,2,None)]}).show()

A directed version of the dodecahedron

sage: D = DiGraph({0: [1, 10, 19], 1: [8, 2], 2: [3, 6], 3: [19, 4],
....:              4: [17, 5], 5: [6, 15], 6: [7], 7: [8, 14],
....:              8: [9], 9: [10, 13], 10: [11], 11: [12, 18],
....:              12: [16, 13], 13: [14], 14: [15], 15: [16], 16: [17],
....:              17: [18], 18: [19], 19: []})
sage: D.plot3d().show()             # long time                             # needs sage.plot
sage: P = graphs.PetersenGraph().to_directed()
sage: from sage.plot.colors import rainbow                                  # needs sage.plot
sage: R = rainbow(P.size(), 'rgbtuple')                                     # needs sage.plot
sage: edge_colors = {R[i]: [e] for i, e in enumerate(P.edge_iterator())}    # needs sage.plot
sage: P.plot3d(engine='tachyon', edge_colors=edge_colors).show()    # long time, needs sage.plot
sage: G = Graph({'a': ['a','b','b','b','e'], 'b': ['c','d','e'],
....:            'c': ['c','d','d','d'], 'd': ['e']}, sparse=True)
sage: G.show3d()
Traceback (most recent call last):
...
NotImplementedError: 3D plotting of multiple edges or loops not implemented

Using the partition keyword:

sage: G = graphs.WheelGraph(7)
sage: G.plot3d(partition=[[0], [1, 2, 3, 4, 5, 6]])                         # needs sage.plot
Graphics3d Object
random_edge(**kwds)#

Return a random edge of self.

INPUT:

EXAMPLES:

The returned value is an edge of self:

sage: g = graphs.PetersenGraph()
sage: u,v = g.random_edge(labels=False)
sage: g.has_edge(u,v)
True

As the edges() method would, this function returns by default a triple (u, v, l) of values, in which l is the label of edge (u, v):

sage: g.random_edge()  # random
(3, 4, None)
random_edge_iterator(*args, **kwds)#

Return an iterator over random edges of self.

The returned iterator enables to amortize the cost of accessing random edges, as can be done with multiple calls to method random_edge().

INPUT:

  • *args and **kwds – arguments to be passed down to the edge_iterator() method.

EXAMPLES:

The returned value is an iterator over the edges of self:

sage: g = graphs.PetersenGraph()
sage: it = g.random_edge_iterator()
sage: [g.has_edge(next(it)) for _ in range(5)]
[True, True, True, True, True]

As the edges() method would, this function returns by default a triple (u, v, l) of values, in which l is the label of edge (u,v):

sage: print(next(g.random_edge_iterator())) # random
(0, 5, None)
sage: print(next(g.random_edge_iterator(labels=False))) # random
(5, 7)
random_subgraph(p, inplace=False)#

Return a random subgraph containing each vertex with probability p.

INPUT:

  • p – the probability of choosing a vertex

  • inplace – boolean (default: False); using inplace=True will simply delete the extra vertices and edges from the current graph. This will modify the graph.

EXAMPLES:

sage: P = graphs.PetersenGraph()
sage: P.random_subgraph(.25)
Subgraph of (Petersen graph): Graph on ... vert...
random_vertex(**kwds)#

Return a random vertex of self.

INPUT:

EXAMPLES:

The returned value is a vertex of self:

sage: g = graphs.PetersenGraph()
sage: v = g.random_vertex()
sage: v in g
True
random_vertex_iterator(*args, **kwds)#

Return an iterator over random vertices of self.

The returned iterator enables to amortize the cost of accessing random vertices, as can be done with multiple calls to method random_vertex().

INPUT:

EXAMPLES:

The returned value is an iterator over the vertices of self:

sage: g = graphs.PetersenGraph()
sage: it = g.random_vertex_iterator()
sage: [next(it) in g for _ in range(5)]
[True, True, True, True, True]
relabel(perm=None, inplace=True, return_map=False, check_input=True, complete_partial_function=True, immutable=None)#

Relabels the vertices of self

INPUT:

  • perm – a function, dictionary, iterable, permutation, or None (default: None)

  • inplace – a boolean (default: True)

  • return_map – a boolean (default: False)

  • check_input (boolean) – whether to test input for correctness. This can potentially be very time-consuming !.

  • complete_partial_function (boolean) – whether to automatically complete the permutation if some elements of the graph are not associated with any new name. In this case, those elements are not relabeled This can potentially be very time-consuming !.

  • immutable (boolean) – with inplace=False, whether to create a mutable/immutable relabelled copy. immutable=None (default) means that the graph and its copy will behave the same way.

If perm is a function f, then each vertex v is relabeled to f(v).

If perm is a dictionary d, then each vertex v (which should be a key of d) is relabeled to d[v].

If perm is a list (or more generally, any iterable) of length n, then the first vertex returned by G.vertices(sort=True) is relabeled to l[0], the second to l[1], …

If perm is a permutation, then each vertex v is relabeled to perm(v). Caveat: this assumes that the vertices are labelled \(\{0,1,...,n-1\}\); since permutations act by default on the set \(\{1,2,...,n\}\), this is achieved by identifying \(n\) and \(0\).

If perm is None, the graph is relabeled to be on the vertices \(\{0,1,...,n-1\}\). This is not any kind of canonical labeling, but it is consistent (relabeling twice will give the same result).

If inplace is True, the graph is modified in place and None is returned. Otherwise a relabeled copy of the graph is returned.

If return_map is True a dictionary representing the relabelling map is returned (incompatible with inplace==False).

EXAMPLES:

sage: G = graphs.PathGraph(3)
sage: G.am()                                                                # needs sage.modules
[0 1 0]
[1 0 1]
[0 1 0]

Relabeling using a dictionary. Note that the dictionary does not define the new label of vertex \(0\):

sage: G.relabel({1:2,2:1}, inplace=False).am()                              # needs sage.modules
[0 0 1]
[0 0 1]
[1 1 0]

This is because the method automatically “extends” the relabeling to the missing vertices (whose label will not change). Checking that all vertices have an image can require some time, and this feature can be disabled (at your own risk):

sage: G.relabel({1:2,2:1}, inplace=False,                                   # needs sage.modules
....:           complete_partial_function=False).am()
Traceback (most recent call last):
...
KeyError: 0

Relabeling using a list:

sage: G.relabel([0,2,1], inplace=False).am()                                # needs sage.modules
[0 0 1]
[0 0 1]
[1 1 0]

Relabeling using an iterable:

sage: G.relabel(iter((0,2,1)), inplace=False).am()                          # needs sage.modules
[0 0 1]
[0 0 1]
[1 1 0]

Relabeling using a Sage permutation:

sage: G = graphs.PathGraph(3)
sage: from sage.groups.perm_gps.permgroup_named import SymmetricGroup       # needs sage.groups
sage: S = SymmetricGroup(3)                                                 # needs sage.groups
sage: gamma = S('(1,2)')                                                    # needs sage.groups
sage: G.relabel(gamma, inplace=False).am()                                  # needs sage.groups sage.modules
[0 0 1]
[0 0 1]
[1 1 0]

A way to get a random relabeling:

sage: set_random_seed(0)  # Results are reproducible
sage: D = DiGraph({1: [2], 3: [4]})
sage: D.relabel(Permutations(D.vertices(sort=True)).random_element())
sage: D.sources()
[1, 4]

Relabeling using an injective function:

sage: G.edges(sort=True)
[(0, 1, None), (1, 2, None)]
sage: H = G.relabel(lambda i: i+10, inplace=False)
sage: H.vertices(sort=True)
[10, 11, 12]
sage: H.edges(sort=True)
[(10, 11, None), (11, 12, None)]

Relabeling using a non injective function has no meaning:

sage: G.edges(sort=True)
[(0, 1, None), (1, 2, None)]
sage: G.relabel(lambda i: 0, inplace=False)
Traceback (most recent call last):
...
NotImplementedError: Non injective relabeling

But this test can be disabled, which can lead to … problems:

sage: G.edges(sort=True)
[(0, 1, None), (1, 2, None)]
sage: G.relabel(lambda i: 0, check_input = False)
sage: G.edges(sort=True)
[]

Recovering the relabeling with return_map:

sage: G = graphs.CubeGraph(3)
sage: G.relabel(range(8), return_map=True)
{'000': 0,
 '001': 1,
 '010': 2,
 '011': 3,
 '100': 4,
 '101': 5,
 '110': 6,
 '111': 7}

When no permutation is given, the relabeling is done to integers from 0 to N-1 but in an arbitrary order:

sage: G = graphs.CubeGraph(3)
sage: G.vertices(sort=True)
['000', '001', '010', '011', '100', '101', '110', '111']
sage: G.relabel()
sage: G.vertices(sort=True)
[0, 1, 2, 3, 4, 5, 6, 7]

In the above case, the mapping is arbitrary but consistent:

sage: map1 = G.relabel(inplace=False, return_map=True)
sage: map2 = G.relabel(inplace=False, return_map=True)
sage: map1 == map2
True
sage: G = graphs.PathGraph(3)
sage: G.relabel(lambda i: i+10, return_map=True)
{0: 10, 1: 11, 2: 12}
remove_loops(vertices=None)#

Remove loops on vertices in vertices.

If vertices is None, removes all loops.

EXAMPLES:

sage: G = Graph(4, loops=True)
sage: G.add_edges([(0, 0), (1, 1), (2, 2), (3, 3), (2, 3)])
sage: G.edges(sort=True, labels=False)
[(0, 0), (1, 1), (2, 2), (2, 3), (3, 3)]
sage: G.remove_loops()
sage: G.edges(sort=True, labels=False)
[(2, 3)]
sage: G.allows_loops()
True
sage: G.has_loops()
False

sage: D = DiGraph(4, loops=True)
sage: D.add_edges([(0, 0), (1, 1), (2, 2), (3, 3), (2, 3)])
sage: D.edges(sort=True, labels=False)
[(0, 0), (1, 1), (2, 2), (2, 3), (3, 3)]
sage: D.remove_loops()
sage: D.edges(sort=True, labels=False)
[(2, 3)]
sage: D.allows_loops()
True
sage: D.has_loops()
False
remove_multiple_edges()#

Remove all multiple edges, retaining one edge for each.

See also

See also allow_multiple_edges()

EXAMPLES:

sage: G = Graph(multiedges=True, sparse=True)
sage: G.add_edges([(0,1), (0,1), (0,1), (0,1), (1,2)])
sage: G.edges(sort=True, labels=False)
[(0, 1), (0, 1), (0, 1), (0, 1), (1, 2)]
sage: G.remove_multiple_edges()
sage: G.edges(sort=True, labels=False)
[(0, 1), (1, 2)]
sage: D = DiGraph(multiedges=True, sparse=True)
sage: D.add_edges([(0, 1, 1), (0, 1, 2), (0, 1, 3), (0, 1, 4), (1, 2, None)])
sage: D.edges(sort=True, labels=False)
[(0, 1), (0, 1), (0, 1), (0, 1), (1, 2)]
sage: D.remove_multiple_edges()
sage: D.edges(sort=True, labels=False)
[(0, 1), (1, 2)]
rooted_product(G, H, root=None)#

Return the rooted product of \(G\) and \(H\).

The rooted product of two graphs \(G\) and \(H\) is the graph \(R\) defined as follows: take a copy of \(G\) and \(|V(G)|\) copies of \(H\), and for every vertex \(g_i\) of \(G\), identify \(g_i\) with the root of the \(i\)-th copy of \(H\). Mode formally, let \(V(G) = \{g_1, g_2, \ldots, g_n\}\), \(V(H) = \{h_1, h_2, \ldots, h_m\}\), and let \(h_1\) be the root vertex of \(H\). The vertex set \(V(R)\) is equal to the cartesian product of the sets of vertices \(V(G)\) and \(V(H)\), that is \(V(R) = \{(g_i, h_j) : g_i \in V(G), h_j \in V(H)\}\). The edge set \(E(R)\) is the union of the edges of a copy of \(G\), that is \(\{((g_i, h_1), (g_j, h_1)) : (g_i, g_j) \in E(G)\}\), and the edges of the copies of \(H\) for every \(g_i \in V(G)\), that is \(\{((g_i, h_j), (g_i, h_k)) : (h_j, h_k) \in V(H)\}\).

See Wikipedia article Rooted_product_of_graphs for more details.

See also

  • cartesian_product() – return the cartesian product of two graphs

  • graph_products – a module on graph products

EXAMPLES:

The rooted product of two trees is a tree:

sage: T1 = graphs.RandomTree(7)
sage: T2 = graphs.RandomTree(8)
sage: T = T1.rooted_product(T2)
sage: T.is_tree()
True

The rooted product of \(G\) and \(H\) depends on the selected root in \(H\):

sage: G = graphs.CycleGraph(4)
sage: H = graphs.PathGraph(3)
sage: R1 = G.rooted_product(H, root=0)
sage: R2 = G.rooted_product(H, root=1)
sage: R1.is_isomorphic(R2)
False
sage: sorted(R1.degree())
[1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3]
sage: sorted(R2.degree())
[1, 1, 1, 1, 1, 1, 1, 1, 4, 4, 4, 4]

The domination number of the rooted product of any graph \(G\) and a path of order 2 is the order of \(G\):

sage: G = graphs.RandomGNP(20, .3)
sage: P = graphs.PathGraph(2)
sage: R = G.rooted_product(P)
sage: len(R.dominating_set()) == G.order()                                      # needs sage.numerical.mip
True
sage: G = digraphs.RandomDirectedGNP(20, .3)
sage: P = digraphs.Path(2)
sage: R = G.rooted_product(P)
sage: len(R.dominating_set()) == G.order()                                      # needs sage.numerical.mip
True

The rooted product of two graphs is a subgraph of the cartesian product of the same two graphs:

sage: G = graphs.RandomGNP(6, .4)
sage: H = graphs.RandomGNP(7, .4)
sage: R = G.rooted_product(H)
sage: C = G.cartesian_product(H)
sage: R.is_subgraph(C, induced=False)
True

Corner cases:

sage: Graph().rooted_product(Graph())
Rooted product of Graph on 0 vertices and Graph on 0 vertices: Graph on 0 vertices
sage: Graph(1).rooted_product(Graph())
Rooted product of Graph on 1 vertex and Graph on 0 vertices: Graph on 0 vertices
sage: Graph().rooted_product(Graph(1))
Rooted product of Graph on 0 vertices and Graph on 1 vertex: Graph on 0 vertices
sage: Graph(1).rooted_product(Graph(1))
Rooted product of Graph on 1 vertex and Graph on 1 vertex: Graph on 1 vertex
set_edge_label(u, v, l)#

Set the edge label of a given edge.

Note

There can be only one edge from u to v for this to make sense. Otherwise, an error is raised.

INPUT:

  • u, v – the vertices (and direction if digraph) of the edge

  • l – the new label

EXAMPLES:

sage: d = {1: [18, 2], 2: [5, 3], 3: [4, 6], 4: [7, 2], 5: [4],
....:      6: [13, 12], 7: [18, 8, 10], 8: [6, 9, 10], 9: [6],
....:      10: [11, 13], 11: [12], 12: [13], 13: [17, 14],
....:      14: [16, 15], 15: [2], 16: [13], 17: [15, 13], 18: [13]}
sage: SD = DiGraph(d, sparse=True)
sage: SD.set_edge_label(1, 18, 'discrete')
sage: SD.set_edge_label(4, 7, 'discrete')
sage: SD.set_edge_label(2, 5, 'h = 0')
sage: SD.set_edge_label(7, 18, 'h = 0')
sage: SD.set_edge_label(7, 10, 'aut')
sage: SD.set_edge_label(8, 10, 'aut')
sage: SD.set_edge_label(8, 9, 'label')
sage: SD.set_edge_label(8, 6, 'no label')
sage: SD.set_edge_label(13, 17, 'k > h')
sage: SD.set_edge_label(13, 14, 'k = h')
sage: SD.set_edge_label(17, 15, 'v_k finite')
sage: SD.set_edge_label(14, 15, 'v_k m.c.r.')
sage: posn = {1: [3, -3], 2: [0, 2], 3: [0, 13], 4: [3, 9],
....:         5: [3, 3], 6: [16, 13], 7: [6, 1], 8: [6, 6],
....:         9: [6, 11], 10: [9, 1], 11: [10, 6], 12: [13, 6],
....:         13: [16, 2], 14: [10, -6], 15: [0, -10], 16: [14, -6],
....:         17: [16, -10], 18: [6, -4]}
sage: SD.plot(pos=posn, vertex_size=400, vertex_colors={'#FFFFFF':list(range(1,19))}, edge_labels=True).show()  # long time, needs sage.plot
sage: G = graphs.HeawoodGraph()
sage: for u,v,l in G.edges(sort=False):
....:  G.set_edge_label(u, v, '(' + str(u) + ',' + str(v) + ')')
sage: G.edges(sort=True)
    [(0, 1, '(0,1)'),
     (0, 5, '(0,5)'),
     (0, 13, '(0,13)'),
     ...
     (11, 12, '(11,12)'),
     (12, 13, '(12,13)')]
sage: g = Graph({0: [0, 1, 1, 2]}, loops=True, multiedges=True, sparse=True)
sage: g.set_edge_label(0, 0, 'test')
sage: g.edges(sort=True)
[(0, 0, 'test'), (0, 1, None), (0, 1, None), (0, 2, None)]
sage: g.add_edge(0, 0, 'test2')
sage: g.set_edge_label(0,0,'test3')
Traceback (most recent call last):
...
RuntimeError: cannot set edge label, since there are multiple edges from 0 to 0
sage: dg = DiGraph({0: [1], 1: [0]}, sparse=True)
sage: dg.set_edge_label(0, 1, 5)
sage: dg.set_edge_label(1, 0, 9)
sage: dg.outgoing_edges(1)
[(1, 0, 9)]
sage: dg.incoming_edges(1)
[(0, 1, 5)]
sage: dg.outgoing_edges(0)
[(0, 1, 5)]
sage: dg.incoming_edges(0)
[(1, 0, 9)]
sage: G = Graph({0: {1: 1}}, sparse=True)
sage: G.num_edges()
1
sage: G.set_edge_label(0, 1, 1)
sage: G.num_edges()
1
set_embedding(embedding)#

Set a combinatorial embedding dictionary to _embedding attribute.

The dictionary embedding represents a combinatorial embedding of self and is organized as a mapping from vertex labels to list of vertex neighbors in clockwise order.

Parameter embedding is error-checked for validity.

Warning

Combinatorial embeddings are defined for simple graphs only (i.e., without loops or multiple edges). Therefore, an error is raised when this method is used for a graph with loops or multiple edges.

INPUT:

  • embedding – dictionary representing a combinatorial embedding of self. Format: “{v1: [v2,v3], v2: [v1], v3: [v1]}” (clockwise ordering of neighbors at each vertex).

EXAMPLES:

sage: G = graphs.PetersenGraph()
sage: G.set_embedding({0: [1, 5, 4], 1: [0, 2, 6], 2: [1, 3, 7],
....:                  3: [8, 2, 4], 4: [0, 9, 3], 5: [0, 8, 7],
....:                  6: [8, 1, 9], 7: [9, 2, 5], 8: [3, 5, 6],
....:                  9: [4, 6, 7]})
sage: G.set_embedding({'s': [1, 5, 4], 1: [0, 2, 6], 2: [1, 3, 7],
....:                  3: [8, 2, 4], 4: [0, 9, 3], 5: [0, 8, 7],
....:                  6: [8, 1, 9], 7: [9, 2, 5], 8: [3, 5, 6],
....:                  9: [4, 6, 7]})
Traceback (most recent call last):
...
ValueError: vertices in ['s'] from the embedding do not belong to the graph
set_latex_options(**kwds)#

Set multiple options for rendering a graph with LaTeX.

INPUT:

  • kwds – any number of option/value pairs to set many graph latex options at once (a variable number, in any order). Existing values are overwritten, new values are added. Existing values can be cleared by setting the value to None. Possible options are documented at sage.graphs.graph_latex.GraphLatex.set_option().

This method is a convenience for setting the options of a graph directly on an instance of the graph. For a full explanation of how to use LaTeX to render graphs, see the introduction to the graph_latex module.

EXAMPLES:

sage: g = graphs.PetersenGraph()
sage: g.set_latex_options(tkz_style='Welsh')                                # needs sage.plot
sage: opts = g.latex_options()                                              # needs sage.plot
sage: opts.get_option('tkz_style')                                          # needs sage.plot
'Welsh'
set_pos(pos, dim=2)#

Set the position dictionary.

The position dictionary specifies the coordinates of each vertex.

INPUT:

  • pos – a position dictionary for the vertices of the (di)graph

  • dim – integer (default: 2); the number of coordinates per vertex

EXAMPLES:

sage: G = graphs.PetersenGraph()
sage: G.get_pos()
{0: (..., ...),
 ...
 9: (..., ...)}

The method get_pos() check the position dictionary so that invalid positioning are ignored:

sage: G.set_pos(dict(enumerate('abcdefghi')))
sage: P = G.plot()  # positions are ignored                                 # needs sage.plot
sage: G.get_pos() is None                                                   # needs sage.plot
True
set_vertex(vertex, object)#

Associate an arbitrary object with a vertex.

INPUT:

  • vertex – which vertex

  • object – object to associate to vertex

EXAMPLES:

sage: T = graphs.TetrahedralGraph()
sage: T.vertices(sort=True)
[0, 1, 2, 3]
sage: T.set_vertex(1, graphs.FlowerSnark())
sage: T.get_vertex(1)
Flower Snark: Graph on 20 vertices
sage: T.set_vertex(4, 'foo')
Traceback (most recent call last):
...
ValueError: vertex (4) not in the graph
set_vertices(vertex_dict)#

Associate arbitrary objects with each vertex, via an association dictionary.

INPUT:

  • vertex_dict – the association dictionary

EXAMPLES:

sage: d = {0: graphs.DodecahedralGraph(), 1: graphs.FlowerSnark(), 2: graphs.MoebiusKantorGraph(), 3: graphs.PetersenGraph()}
sage: d[2]
Moebius-Kantor Graph: Graph on 16 vertices
sage: T = graphs.TetrahedralGraph()
sage: T.vertices(sort=True)
[0, 1, 2, 3]
sage: T.set_vertices(d)
sage: T.get_vertex(1)
Flower Snark: Graph on 20 vertices
shortest_path(u, v, by_weight=False, algorithm=None, weight_function=None, check_weight=True)#

Return a list of vertices representing some shortest path from u to v.

If there is no path from \(u\) to \(v\), the returned list is empty.

For more information and more examples, see shortest_paths() (the inputs are very similar).

INPUT:

  • u, v – the start and the end vertices of the paths

  • by_weight – boolean (default: False); if True, the edges in the graph are weighted, otherwise all edges have weight 1

  • algorithm – string (default: None); one of the following algorithms:

    • 'BFS': performs a BFS from u. Does not work with edge weights.

    • 'BFS_Bid': performs a BFS from u and from v. Does not work with edge weights.

    • 'Dijkstra_NetworkX': the Dijkstra algorithm, implemented in NetworkX. Works only with positive weights.

    • 'Dijkstra_Bid_NetworkX': performs a Dijkstra visit from u and from v (NetworkX implementation). Works only with positive weights.

    • 'Dijkstra_Bid': a Cython implementation that performs a Dijkstra visit from u and from v. Works only with positive weights.

    • 'Bellman-Ford_Boost': the Bellman-Ford algorithm, implemented in Boost. Works also with negative weights, if there is no negative cycle.

    • None (default): Sage chooses the best algorithm: 'BFS_Bid' if by_weight is False, 'Dijkstra_Bid' otherwise.

    Note

    If there are negative weights and algorithm is None, the result is not reliable. This occurs because, for performance reasons, we cannot check whether there are edges with negative weights before running the algorithm. If there are, the user should explicitly input algorithm='Bellman-Ford_Boost'.

  • weight_function – function (default: None); a function that takes as input an edge (u, v, l) and outputs its weight. If not None, by_weight is automatically set to True. If None and by_weight is True, we use the edge label l, if l is not None, else 1 as a weight.

  • check_weight – boolean (default: True); if True, we check that the weight_function outputs a number for each edge

EXAMPLES:

sage: D = graphs.DodecahedralGraph()
sage: D.shortest_path(4, 9)
[4, 17, 16, 12, 13, 9]
sage: D.shortest_path(4, 9, algorithm='BFS')
[4, 3, 2, 1, 8, 9]
sage: D.shortest_path(4, 8, algorithm='Dijkstra_NetworkX')                  # needs networkx
[4, 3, 2, 1, 8]
sage: D.shortest_path(4, 8, algorithm='Dijkstra_Bid_NetworkX')              # needs networkx
[4, 3, 2, 1, 8]
sage: D.shortest_path(4, 9, algorithm='Dijkstra_Bid')
[4, 3, 19, 0, 10, 9]
sage: D.shortest_path(5, 5)
[5]
sage: D.delete_edges(D.edges_incident(13))
sage: D.shortest_path(13, 4)
[]
sage: G = Graph({0: {1: 1}, 1: {2: 1}, 2: {3: 1}, 3: {4: 2}, 4: {0: 2}},
....:           sparse=True)
sage: G.plot(edge_labels=True).show()       # long time                     # needs sage.plot
sage: G.shortest_path(0, 3)
[0, 4, 3]
sage: G.shortest_path(0, 3, by_weight=True)
[0, 1, 2, 3]
sage: G.shortest_path(0, 3, by_weight=True,                                 # needs networkx
....:                 algorithm='Dijkstra_NetworkX')
[0, 1, 2, 3]
sage: G.shortest_path(0, 3, by_weight=True,                                 # needs networkx
....:                 algorithm='Dijkstra_Bid_NetworkX')
[0, 1, 2, 3]

Todo

Add options to return a path as a list of edges with or without edge labels. This can be useful in (di)graphs with multiple edges.

shortest_path_all_pairs(by_weight=False, algorithm=None, weight_function=None, check_weight=True)#

Return a shortest path between each pair of vertices.

INPUT:

  • by_weight – boolean (default: False); if True, the edges in the graph are weighted, otherwise all edges have weight 1

  • algorithm – string (default: None); one of the following algorithms:

    • 'BFS': the computation is done through a BFS centered on each vertex successively. Works only if by_weight==False.

    • 'Floyd-Warshall-Cython': the Cython implementation of the Floyd-Warshall algorithm. Works only if by_weight==False.

    • 'Floyd-Warshall-Python': the Python implementation of the Floyd-Warshall algorithm. Works also with weighted graphs, even with negative weights (but no negative cycle is allowed).

    • 'Floyd-Warshall_Boost': the Boost implementation of the Floyd-Warshall algorithm. Works also with weighted graphs, even with negative weights (but no negative cycle is allowed).

    • 'Floyd-Warshall_SciPy': the SciPy implementation of the Floyd-Warshall algorithm. Works also with weighted graphs, even with negative weights (but no negative cycle is allowed).

    • 'Dijkstra_NetworkX': the Dijkstra algorithm, implemented in NetworkX. It works with weighted graphs, but no negative weight is allowed.

    • 'Dijkstra_Boost': the Dijkstra algorithm, implemented in Boost (works only with positive weights).

    • 'Johnson_Boost': the Johnson algorithm, implemented in Boost (works also with negative weights, if there is no negative cycle).

    • None (default): Sage chooses the best algorithm: 'BFS' if by_weight is False, 'Dijkstra_Boost' if all weights are positive, 'Floyd-Warshall_Boost' otherwise.

  • weight_function – function (default: None); a function that takes as input an edge (u, v, l) and outputs its weight. If not None, by_weight is automatically set to True. If None and by_weight is True, we use the edge label l, if l is not None, else 1 as a weight.

  • check_weight – boolean (default: True); if True, we check that the weight_function outputs a number for each edge

OUTPUT:

A tuple (dist, pred). They are both dicts of dicts. The first indicates the length dist[u][v] of the shortest weighted path from \(u\) to \(v\). The second is a compact representation of all the paths - it indicates the predecessor pred[u][v] of \(v\) in the shortest path from \(u\) to \(v\).

Note

Only reachable vertices are present in the dictionaries.

Note

There is a Cython version of this method that is usually much faster for large graphs, as most of the time is actually spent building the final double dictionary. Everything on the subject is to be found in the distances_all_pairs module.

EXAMPLES:

Some standard examples (see shortest_paths() for more examples on how to use the input variables):

sage: G = Graph({0: {1: 1}, 1: {2: 1}, 2: {3: 1}, 3: {4: 2}, 4: {0: 2}}, sparse=True)
sage: G.plot(edge_labels=True).show()       # long time                     # needs sage.plot
sage: dist, pred = G.shortest_path_all_pairs(by_weight = True)
sage: dist
{0: {0: 0, 1: 1, 2: 2, 3: 3, 4: 2},
 1: {0: 1, 1: 0, 2: 1, 3: 2, 4: 3},
 2: {0: 2, 1: 1, 2: 0, 3: 1, 4: 3},
 3: {0: 3, 1: 2, 2: 1, 3: 0, 4: 2},
 4: {0: 2, 1: 3, 2: 3, 3: 2, 4: 0}}
sage: pred
{0: {0: None, 1: 0, 2: 1, 3: 2, 4: 0},
 1: {0: 1, 1: None, 2: 1, 3: 2, 4: 0},
 2: {0: 1, 1: 2, 2: None, 3: 2, 4: 3},
 3: {0: 1, 1: 2, 2: 3, 3: None, 4: 3},
 4: {0: 4, 1: 0, 2: 3, 3: 4, 4: None}}
sage: pred[0]
{0: None, 1: 0, 2: 1, 3: 2, 4: 0}
sage: G = Graph( { 0: {1: {'weight':1}}, 1: {2: {'weight':1}}, 2: {3: {'weight':1}}, 3: {4: {'weight':2}}, 4: {0: {'weight':2}} }, sparse=True)
sage: dist, pred = G.shortest_path_all_pairs(weight_function = lambda e:e[2]['weight'])
sage: dist
{0: {0: 0, 1: 1, 2: 2, 3: 3, 4: 2},
 1: {0: 1, 1: 0, 2: 1, 3: 2, 4: 3},
 2: {0: 2, 1: 1, 2: 0, 3: 1, 4: 3},
 3: {0: 3, 1: 2, 2: 1, 3: 0, 4: 2},
 4: {0: 2, 1: 3, 2: 3, 3: 2, 4: 0}}
sage: pred
{0: {0: None, 1: 0, 2: 1, 3: 2, 4: 0},
 1: {0: 1, 1: None, 2: 1, 3: 2, 4: 0},
 2: {0: 1, 1: 2, 2: None, 3: 2, 4: 3},
 3: {0: 1, 1: 2, 2: 3, 3: None, 4: 3},
 4: {0: 4, 1: 0, 2: 3, 3: 4, 4: None}}

So for example the shortest weighted path from \(0\) to \(3\) is obtained as follows. The predecessor of \(3\) is pred[0][3] == 2, the predecessor of \(2\) is pred[0][2] == 1, and the predecessor of \(1\) is pred[0][1] == 0.

sage: G = Graph( { 0: {1:None}, 1: {2:None}, 2: {3: 1}, 3: {4: 2}, 4: {0: 2} }, sparse=True )
sage: G.shortest_path_all_pairs()
({0: {0: 0, 1: 1, 2: 2, 3: 2, 4: 1},
1: {0: 1, 1: 0, 2: 1, 3: 2, 4: 2},
2: {0: 2, 1: 1, 2: 0, 3: 1, 4: 2},
3: {0: 2, 1: 2, 2: 1, 3: 0, 4: 1},
4: {0: 1, 1: 2, 2: 2, 3: 1, 4: 0}},
{0: {0: None, 1: 0, 2: 1, 3: 4, 4: 0},
1: {0: 1, 1: None, 2: 1, 3: 2, 4: 0},
2: {0: 1, 1: 2, 2: None, 3: 2, 4: 3},
3: {0: 4, 1: 2, 2: 3, 3: None, 4: 3},
4: {0: 4, 1: 0, 2: 3, 3: 4, 4: None}})
sage: G.shortest_path_all_pairs(weight_function=lambda e:(e[2] if e[2] is not None else 1))
({0: {0: 0, 1: 1, 2: 2, 3: 3, 4: 2},
1: {0: 1, 1: 0, 2: 1, 3: 2, 4: 3},
2: {0: 2, 1: 1, 2: 0, 3: 1, 4: 3},
3: {0: 3, 1: 2, 2: 1, 3: 0, 4: 2},
4: {0: 2, 1: 3, 2: 3, 3: 2, 4: 0}},
{0: {0: None, 1: 0, 2: 1, 3: 2, 4: 0},
1: {0: 1, 1: None, 2: 1, 3: 2, 4: 0},
2: {0: 1, 1: 2, 2: None, 3: 2, 4: 3},
3: {0: 1, 1: 2, 2: 3, 3: None, 4: 3},
4: {0: 4, 1: 0, 2: 3, 3: 4, 4: None}})

Checking that distances are equal regardless of the algorithm used:

sage: g = graphs.Grid2dGraph(5,5)
sage: d1, _ = g.shortest_path_all_pairs(algorithm="BFS")
sage: d2, _ = g.shortest_path_all_pairs(algorithm="Floyd-Warshall-Cython")
sage: d3, _ = g.shortest_path_all_pairs(algorithm="Floyd-Warshall-Python")
sage: d4, _ = g.shortest_path_all_pairs(algorithm="Dijkstra_NetworkX")      # needs networkx
sage: d5, _ = g.shortest_path_all_pairs(algorithm="Dijkstra_Boost")
sage: d6, _ = g.shortest_path_all_pairs(algorithm="Johnson_Boost")
sage: d7, _ = g.shortest_path_all_pairs(algorithm="Floyd-Warshall_Boost")
sage: d8, _ = g.shortest_path_all_pairs(algorithm="Floyd-Warshall_SciPy")   # needs scipy
sage: d1 == d2 == d3 == d4 == d5 == d6 == d7 == d8                          # needs networkx scipy
True

Checking that distances are equal regardless of the algorithm used:

sage: g = digraphs.RandomDirectedGNM(6,12)
sage: d1, _ = g.shortest_path_all_pairs(algorithm="BFS")
sage: d2, _ = g.shortest_path_all_pairs(algorithm="Floyd-Warshall-Cython")
sage: d3, _ = g.shortest_path_all_pairs(algorithm="Floyd-Warshall-Python")
sage: d4, _ = g.shortest_path_all_pairs(algorithm="Dijkstra_NetworkX")      # needs networkx
sage: d5, _ = g.shortest_path_all_pairs(algorithm="Dijkstra_Boost")
sage: d6, _ = g.shortest_path_all_pairs(algorithm="Johnson_Boost")
sage: d7, _ = g.shortest_path_all_pairs(algorithm="Floyd-Warshall_Boost")
sage: d8, _ = g.shortest_path_all_pairs(algorithm="Floyd-Warshall_SciPy")   # needs scipy
sage: d1 == d2 == d3 == d4 == d5 == d6 == d7 == d8                          # needs networkx scipy
True

Checking that weighted distances are equal regardless of the algorithm used:

sage: g = graphs.CompleteGraph(5)
sage: import random
sage: for v, w in g.edges(labels=False, sort=False):
....:     g.add_edge(v, w, random.uniform(1, 10))
sage: d1, _ = g.shortest_path_all_pairs(algorithm="Floyd-Warshall-Python")
sage: d2, _ = g.shortest_path_all_pairs(algorithm="Dijkstra_NetworkX")      # needs networkx
sage: d3, _ = g.shortest_path_all_pairs(algorithm="Dijkstra_Boost")
sage: d4, _ = g.shortest_path_all_pairs(algorithm="Johnson_Boost")
sage: d5, _ = g.shortest_path_all_pairs(algorithm="Floyd-Warshall_Boost")
sage: d6, _ = g.shortest_path_all_pairs(algorithm="Floyd-Warshall_SciPy")   # needs scipy
sage: d1 == d2 == d3 == d4 == d5 == d6                                      # needs networkx scipy
True

Checking a random path is valid:

sage: dist, path = g.shortest_path_all_pairs(algorithm="BFS")
sage: u,v = g.random_vertex(), g.random_vertex()
sage: p = [v]
sage: while p[0] is not None:
....:   p.insert(0,path[u][p[0]])
sage: len(p) == dist[u][v] + 2
True

Negative weights:

sage: g = DiGraph([(0,1,-2),(1,0,1)], weighted=True)
sage: g.shortest_path_all_pairs(by_weight=True)
Traceback (most recent call last):
...
ValueError: the graph contains a negative cycle

Unreachable vertices are not present in the dictionaries:

sage: g = DiGraph([(0,1,1),(1,2,2)])
sage: g.shortest_path_all_pairs(algorithm='BFS')
({0: {0: 0, 1: 1, 2: 2}, 1: {1: 0, 2: 1}, 2: {2: 0}},
 {0: {0: None, 1: 0, 2: 1}, 1: {1: None, 2: 1}, 2: {2: None}})
sage: g.shortest_path_all_pairs(algorithm='Dijkstra_NetworkX')              # needs networkx
({0: {0: 0, 1: 1, 2: 2}, 1: {1: 0, 2: 1}, 2: {2: 0}},
 {0: {0: None, 1: 1, 2: 1}, 1: {1: None, 2: 2}, 2: {2: None}})
sage: g.shortest_path_all_pairs(algorithm='Dijkstra_Boost')
({0: {0: 0, 1: 1, 2: 2}, 1: {1: 0, 2: 1}, 2: {2: 0}},
 {0: {0: None, 1: 0, 2: 1}, 1: {1: None, 2: 1}, 2: {2: None}})
sage: g.shortest_path_all_pairs(algorithm='Floyd-Warshall-Python')
({0: {0: 0, 1: 1, 2: 2}, 1: {1: 0, 2: 1}, 2: {2: 0}},
 {0: {0: None, 1: 0, 2: 1}, 1: {1: None, 2: 1}, 2: {2: None}})
sage: g.shortest_path_all_pairs(algorithm='Floyd-Warshall-Cython')
({0: {0: 0, 1: 1, 2: 2}, 1: {1: 0, 2: 1}, 2: {2: 0}},
 {0: {0: None, 1: 0, 2: 1}, 1: {1: None, 2: 1}, 2: {2: None}})
sage: g.shortest_path_all_pairs(algorithm='Floyd-Warshall_SciPy')           # needs scipy
({0: {0: 0.0, 1: 1.0, 2: 2.0}, 1: {1: 0.0, 2: 1.0}, 2: {2: 0.0}},
 {0: {0: None, 1: 0, 2: 1}, 1: {1: None, 2: 1}, 2: {2: None}})

In order to change the default behavior if the graph is disconnected, we can use default values with dictionaries:

sage: G = 2*graphs.PathGraph(2)
sage: d,_ = G.shortest_path_all_pairs()
sage: import itertools
sage: from sage.rings.infinity import Infinity
sage: for u,v in itertools.combinations(G.vertex_iterator(), 2):
....:     print("dist({}, {}) = {}".format(u,v, d[u].get(v,+Infinity)))
dist(0, 1) = 1
dist(0, 2) = +Infinity
dist(0, 3) = +Infinity
dist(1, 2) = +Infinity
dist(1, 3) = +Infinity
dist(2, 3) = 1
shortest_path_length(u, v, by_weight=False, algorithm=None, weight_function=None, check_weight=True)#

Return the minimal length of a path from u to v.

If there is no path from \(u\) to \(v\), returns Infinity.

For more information and more examples, we refer to shortest_path() and shortest_paths(), which have very similar inputs.

INPUT:

  • u, v – the start and the end vertices of the paths

  • by_weight – boolean (default: False); if True, the edges in the graph are weighted, otherwise all edges have weight 1

  • algorithm – string (default: None); one of the following algorithms:

    • 'BFS': performs a BFS from u. Does not work with edge weights.

    • 'BFS_Bid': performs a BFS from u and from v. Does not work with edge weights.

    • 'Dijkstra_NetworkX': the Dijkstra algorithm, implemented in NetworkX. Works only with positive weights.

    • 'Dijkstra_Bid_NetworkX': performs a Dijkstra visit from u and from v (NetworkX implementation). Works only with positive weights.

    • 'Dijkstra_Bid': a Cython implementation that performs a Dijkstra visit from u and from v. Works only with positive weights.

    • 'Bellman-Ford_Boost': the Bellman-Ford algorithm, implemented in Boost. Works also with negative weights, if there is no negative cycle.

    • None (default): Sage chooses the best algorithm: 'BFS_Bid' if by_weight is False, 'Dijkstra_Bid' otherwise.

    Note

    If there are negative weights and algorithm is None, the result is not reliable. This occurs because, for performance reasons, we cannot check whether there are edges with negative weights before running the algorithm. If there are, the user should explicitly input algorithm='Bellman-Ford_Boost'.

  • weight_function – function (default: None); a function that takes as input an edge (u, v, l) and outputs its weight. If not None, by_weight is automatically set to True. If None and by_weight is True, we use the edge label l, if l is not None, else 1 as a weight.

  • check_weight – boolean (default: True); if True, we check that the weight_function outputs a number for each edge

EXAMPLES:

Standard examples:

sage: D = graphs.DodecahedralGraph()
sage: D.shortest_path_length(4, 9)
5
sage: D.shortest_path_length(4, 9, algorithm='BFS')
5
sage: D.shortest_path_length(4, 9, algorithm='Dijkstra_NetworkX')           # needs networkx
5
sage: D.shortest_path_length(4, 9, algorithm='Dijkstra_Bid_NetworkX')       # needs networkx
5
sage: D.shortest_path_length(4, 9, algorithm='Dijkstra_Bid')
5
sage: D.shortest_path_length(4, 9, algorithm='Bellman-Ford_Boost')
5
sage: D.shortest_path_length(5, 5)
0
sage: D.delete_edges(D.edges_incident(13))
sage: D.shortest_path_length(13, 4)
+Infinity
sage: G = Graph({0: {1: 1}, 1: {2: 1}, 2: {3: 1}, 3: {4: 2}, 4: {0: 2}},
....:           sparse=True)
sage: G.plot(edge_labels=True).show()       # long time                     # needs sage.plot
sage: G.shortest_path_length(0, 3)
2
sage: G.shortest_path_length(0, 3, by_weight=True)
3
sage: G.shortest_path_length(0, 3, by_weight=True,                          # needs networkx
....:                        algorithm='Dijkstra_NetworkX')
3
sage: G.shortest_path_length(0, 3, by_weight=True,                          # needs networkx
....:                        algorithm='Dijkstra_Bid_NetworkX')
3

If Dijkstra is used with negative weights, usually it raises an error:

sage: G = DiGraph({0: {1: 1}, 1: {2: 1}, 2: {3: 1}, 3: {4: 2}, 4: {0: -2}},
....:             sparse=True)
sage: G.shortest_path_length(4, 1, by_weight=True, algorithm=None)
Traceback (most recent call last):
...
ValueError: the graph contains an edge with negative weight
sage: G.shortest_path_length(4, 1, by_weight=True,
....:                        algorithm='Bellman-Ford_Boost')
-1

However, sometimes the result may be wrong, and no error is raised:

sage: G = DiGraph([(0,1,1), (1,2,1), (0,3,1000), (3,4,-3000), (4,2,1000)])
sage: G.shortest_path_length(0, 2, by_weight=True,
....:                        algorithm='Bellman-Ford_Boost')
-1000
sage: G.shortest_path_length(0, 2, by_weight=True)
2
shortest_path_lengths(u, by_weight=False, algorithm=None, weight_function=None, check_weight=True)#

Return the length of a shortest path from u to any other vertex.

Returns a dictionary of shortest path lengths keyed by targets, excluding all vertices that are not reachable from \(u\).

For more information on the input variables and more examples, we refer to shortest_paths() which has the same input variables.

INPUT:

  • u – the starting vertex

  • by_weight – boolean (default: False); if True, the edges in the graph are weighted, otherwise all edges have weight 1

  • algorithm – string (default: None); one of the following algorithms:

    • 'BFS': performs a BFS from u. Does not work with edge weights.

    • 'Dijkstra_NetworkX': the Dijkstra algorithm, implemented in NetworkX (works only with positive weights).

    • 'Dijkstra_Boost': the Dijkstra algorithm, implemented in Boost (works only with positive weights).

    • 'Bellman-Ford_Boost': the Bellman-Ford algorithm, implemented in Boost (works also with negative weights, if there is no negative cycle).

    • None (default): Sage chooses the best algorithm: 'BFS' if by_weight is False, 'Dijkstra_Boost' if all weights are positive, 'Bellman-Ford_Boost' otherwise.

  • weight_function – function (default: None); a function that takes as input an edge (u, v, l) and outputs its weight. If not None, by_weight is automatically set to True. If None and by_weight is True, we use the edge label l, if l is not None, else 1 as a weight.

  • check_weight – boolean (default: True); if True, we check that the weight_function outputs a number for each edge

EXAMPLES:

Unweighted case:

sage: D = graphs.DodecahedralGraph()
sage: D.shortest_path_lengths(0)
{0: 0, 1: 1, 2: 2, 3: 2, 4: 3, 5: 4, 6: 3, 7: 3, 8: 2, 9: 2, 10: 1,
 11: 2, 12: 3, 13: 3, 14: 4, 15: 5, 16: 4, 17: 3, 18: 2, 19: 1}

Weighted case:

sage: G = Graph({0: {1: 1}, 1: {2: 1}, 2: {3: 1}, 3: {4: 2}, 4: {0: 2}},
....:           sparse=True)
sage: G.plot(edge_labels=True).show()       # long time                     # needs sage.plot
sage: G.shortest_path_lengths(0, by_weight=True)
{0: 0, 1: 1, 2: 2, 3: 3, 4: 2}

Using a weight function:

sage: D = DiGraph([(0,1,{'weight':1}), (1,2,{'weight':3}), (0,2,{'weight':5})])
sage: weight_function = lambda e: e[2]['weight']
sage: D.shortest_path_lengths(1, algorithm='Dijkstra_NetworkX',             # needs networkx
....:                         by_weight=False)
{1: 0, 2: 1}
sage: D.shortest_path_lengths(0, weight_function=weight_function)
{0: 0, 1: 1, 2: 4}
sage: D.shortest_path_lengths(1, weight_function=weight_function)
{1: 0, 2: 3}

Negative weights:

sage: D = DiGraph([(0,1,{'weight':-1}), (1,2,{'weight':3}), (0,2,{'weight':5})])
sage: D.shortest_path_lengths(0, weight_function=weight_function)
{0: 0, 1: -1, 2: 2}

Negative cycles:

sage: D = DiGraph([(0,1,{'weight':-5}), (1,2,{'weight':3}), (2,0,{'weight':1})])
sage: D.shortest_path_lengths(0, weight_function=weight_function)
Traceback (most recent call last):
...
ValueError: the graph contains a negative cycle

Checking that distances are equal regardless of the algorithm used:

sage: g = graphs.Grid2dGraph(5,5)
sage: d1 = g.shortest_path_lengths((0,0), algorithm="BFS")
sage: d2 = g.shortest_path_lengths((0,0), algorithm="Dijkstra_NetworkX")    # needs networkx
sage: d3 = g.shortest_path_lengths((0,0), algorithm="Dijkstra_Boost")
sage: d4 = g.shortest_path_lengths((0,0), algorithm="Bellman-Ford_Boost")
sage: d1 == d2 == d3 == d4                                                  # needs networkx
True
shortest_paths(u, by_weight=False, algorithm=None, weight_function=None, check_weight=True, cutoff=None)#

Return a dictionary associating to each vertex v a shortest path from u to v, if it exists.

If \(u\) and \(v\) are not connected, vertex \(v\) is not present in the dictionary.

INPUT:

  • u – the starting vertex

  • by_weight – boolean (default: False); if True, the edges in the graph are weighted, otherwise all edges have weight 1

  • algorithm – string (default: None); one of the following algorithms:

    • 'BFS': performs a BFS from u. Does not work with edge weights.

    • 'Dijkstra_NetworkX': the Dijkstra algorithm, implemented in NetworkX (works only with positive weights).

    • 'Dijkstra_Boost': the Dijkstra algorithm, implemented in Boost (works only with positive weights).

    • 'Bellman-Ford_Boost': the Bellman-Ford algorithm, implemented in Boost (works also with negative weights, if there is no negative cycle).

    • None (default): Sage chooses the best algorithm: 'BFS' if by_weight is False, 'Dijkstra_Boost' if all weights are positive, 'Bellman-Ford_Boost' otherwise.

  • weight_function – function (default: None); a function that takes as input an edge (u, v, l) and outputs its weight. If not None, by_weight is automatically set to True. If None and by_weight is True, we use the edge label l, if l is not None, else 1 as a weight.

  • check_weight – boolean (default: True); if True, we check that the weight_function outputs a number for each edge

  • cutoff – integer (default: None); integer depth to stop search (used only if algorithm=='BFS')

EXAMPLES:

Standard example:

sage: D = graphs.DodecahedralGraph()
sage: D.shortest_paths(0)
{0: [0], 1: [0, 1], 2: [0, 1, 2], 3: [0, 19, 3], 4: [0, 19, 3, 4],
 5: [0, 1, 2, 6, 5], 6: [0, 1, 2, 6], 7: [0, 1, 8, 7], 8: [0, 1, 8],
 9: [0, 10, 9], 10: [0, 10], 11: [0, 10, 11], 12: [0, 10, 11, 12],
 13: [0, 10, 9, 13], 14: [0, 1, 8, 7, 14],
 15: [0, 19, 18, 17, 16, 15], 16: [0, 19, 18, 17, 16],
 17: [0, 19, 18, 17], 18: [0, 19, 18], 19: [0, 19]}

All these paths are obviously induced graphs:

sage: all(D.subgraph(p).is_isomorphic(graphs.PathGraph(len(p))) for p in D.shortest_paths(0).values())
True
sage: D.shortest_paths(0, cutoff=2)
{0: [0], 1: [0, 1], 2: [0, 1, 2], 3: [0, 19, 3], 8: [0, 1, 8],
 9: [0, 10, 9], 10: [0, 10], 11: [0, 10, 11], 18: [0, 19, 18],
 19: [0, 19]}
sage: G = Graph( { 0: {1: 1}, 1: {2: 1}, 2: {3: 1}, 3: {4: 2}, 4: {0: 2} }, sparse=True)
sage: G.plot(edge_labels=True).show()       # long time                     # needs sage.plot
sage: G.shortest_paths(0, by_weight=True)
{0: [0], 1: [0, 1], 2: [0, 1, 2], 3: [0, 1, 2, 3], 4: [0, 4]}

Weighted shortest paths:

sage: D = DiGraph([(0,1,1),(1,2,3),(0,2,5)])
sage: D.shortest_paths(0)
{0: [0], 1: [0, 1], 2: [0, 2]}
sage: D.shortest_paths(0, by_weight=True)
{0: [0], 1: [0, 1], 2: [0, 1, 2]}

Using a weight function (this way, by_weight is set to True):

sage: D = DiGraph([(0,1,{'weight':1}),(1,2,{'weight':3}),(0,2,{'weight':5})])
sage: weight_function = lambda e:e[2]['weight']
sage: D.shortest_paths(0, weight_function=weight_function)
{0: [0], 1: [0, 1], 2: [0, 1, 2]}

If the weight function does not match the label:

sage: D.shortest_paths(0, weight_function=lambda e:e[2])
Traceback (most recent call last):
...
ValueError: the weight function cannot find the weight of (0, 1, {'weight': 1})

However, if check_weight is set to False, unexpected behavior may occur:

sage: D.shortest_paths(0, algorithm='Dijkstra_NetworkX',                    # needs networkx
....:                  weight_function=lambda e:e[2], check_weight=False)
Traceback (most recent call last):
...
TypeError: unsupported operand type(s) for +: 'int' and 'dict'

Negative weights:

sage: D = DiGraph([(0,1,1),(1,2,-2),(0,2,4)])
sage: D.shortest_paths(0, by_weight=True)
{0: [0], 1: [0, 1], 2: [0, 1, 2]}

Negative cycles:

sage: D.add_edge(2,0,0)
sage: D.shortest_paths(0, by_weight=True)
Traceback (most recent call last):
...
ValueError: the graph contains a negative cycle
shortest_simple_paths(source, target, weight_function=None, by_weight=False, check_weight=True, algorithm=None, report_edges=False, labels=False, report_weight=False)#

Return an iterator over the simple paths between a pair of vertices.

This method returns an iterator over the simple paths (i.e., without repetition) from source to target. By default (by_weight is False), the paths are reported by increasing number of edges. When by_weight is True, the paths are reported by increasing weights.

In case of weighted graphs negative weights are not allowed.

If source is the same vertex as target, then [[source]] is returned – a list containing the 1-vertex, 0-edge path source.

By default Yen's algorithm [Yen1970] is used for undirected graphs and Feng's algorithm is used for directed graphs [Feng2014].

The loops and the multiedges if present in the given graph are ignored and only minimum of the edge labels is kept in case of multiedges.

INPUT:

  • source – a vertex of the graph, where to start

  • target – a vertex of the graph, where to end

  • weight_function – function (default: None); a function that takes as input an edge (u, v, l) and outputs its weight. If not None, by_weight is automatically set to True. If None and by_weight is True, we use the edge label l as a weight.

  • by_weight – boolean (default: False); if True, the edges in the graph are weighted, otherwise all edges have weight 1

  • check_weight – boolean (default: True); whether to check that the weight_function outputs a number for each edge.

  • algorithm – string (default: None); the algorithm to use in computing k shortest paths of self. The following algorithms are supported:

    • "Yen" – Yen’s algorithm [Yen1970]

    • "Feng" – an improved version of Yen’s algorithm but that works only for directed graphs [Feng2014]

  • report_edges – boolean (default: False); whether to report paths as list of vertices (default) or list of edges. When set to False, the labels parameter is ignored.

  • labels – boolean (default: False); if False, each edge is simply a pair (u, v) of vertices. Otherwise a list of edges along with its edge labels are used to represent the path.

  • report_weight – boolean (default: False); if False, just the path between source and target is returned. Otherwise a tuple of path length and path is returned.

EXAMPLES:

sage: g = DiGraph([(1, 2, 20), (1, 3, 10), (1, 4, 30),
....:              (2, 5, 20), (3, 5, 10), (4, 5, 30)])
sage: list(g.shortest_simple_paths(1, 5, by_weight=True, algorithm="Yen"))
[[1, 3, 5], [1, 2, 5], [1, 4, 5]]
sage: list(g.shortest_simple_paths(1, 5, algorithm="Yen"))
[[1, 2, 5], [1, 3, 5], [1, 4, 5]]
sage: list(g.shortest_simple_paths(1, 1))
[[1]]
sage: list(g.shortest_simple_paths(1, 5, by_weight=True,
....:                              report_edges=True, report_weight=True, labels=True))
[(20, [(1, 3, 10), (3, 5, 10)]),
 (40, [(1, 2, 20), (2, 5, 20)]),
 (60, [(1, 4, 30), (4, 5, 30)])]
sage: list(g.shortest_simple_paths(1, 5, by_weight=True, algorithm="Feng",
....:                              report_edges=True, report_weight=True))
[(20, [(1, 3), (3, 5)]), (40, [(1, 2), (2, 5)]), (60, [(1, 4), (4, 5)])]
sage: list(g.shortest_simple_paths(1, 5, report_edges=True, report_weight=True))
[(2, [(1, 4), (4, 5)]), (2, [(1, 3), (3, 5)]), (2, [(1, 2), (2, 5)])]
sage: list(g.shortest_simple_paths(1, 5, by_weight=True, report_edges=True))
[[(1, 3), (3, 5)], [(1, 2), (2, 5)], [(1, 4), (4, 5)]]
sage: list(g.shortest_simple_paths(1, 5, by_weight=True, algorithm="Feng",
....:                              report_edges=True, labels=True))
[[(1, 3, 10), (3, 5, 10)], [(1, 2, 20), (2, 5, 20)], [(1, 4, 30), (4, 5, 30)]]
sage: g = Graph([(1, 2, 20), (1, 3, 10), (1, 4, 30), (2, 5, 20),
....:            (3, 5, 10), (4, 5, 30), (1, 6, 100), (5, 6, 5)])
sage: list(g.shortest_simple_paths(1, 6, by_weight = True))
[[1, 3, 5, 6], [1, 2, 5, 6], [1, 4, 5, 6], [1, 6]]
sage: list(g.shortest_simple_paths(1, 6, algorithm="Yen"))
[[1, 6], [1, 2, 5, 6], [1, 3, 5, 6], [1, 4, 5, 6]]
sage: list(g.shortest_simple_paths(1, 6,
....:                              report_edges=True, report_weight=True, labels=True))
[(1, [(1, 6, 100)]),
 (3, [(1, 2, 20), (2, 5, 20), (5, 6, 5)]),
 (3, [(1, 3, 10), (3, 5, 10), (5, 6, 5)]),
 (3, [(1, 4, 30), (4, 5, 30), (5, 6, 5)])]
sage: list(g.shortest_simple_paths(1, 6, by_weight=True,
....:                              report_edges=True, report_weight=True, labels=True))
[(25, [(1, 3, 10), (3, 5, 10), (5, 6, 5)]),
 (45, [(1, 2, 20), (2, 5, 20), (5, 6, 5)]),
 (65, [(1, 4, 30), (4, 5, 30), (5, 6, 5)]),
 (100, [(1, 6, 100)])]
sage: list(g.shortest_simple_paths(1, 6, by_weight=True,
....:                              report_edges=True, labels=True))
[[(1, 3, 10), (3, 5, 10), (5, 6, 5)],
 [(1, 2, 20), (2, 5, 20), (5, 6, 5)],
 [(1, 4, 30), (4, 5, 30), (5, 6, 5)],
 [(1, 6, 100)]]
sage: list(g.shortest_simple_paths(1, 6, report_edges=True, labels=True))
[[(1, 6, 100)],
 [(1, 2, 20), (2, 5, 20), (5, 6, 5)],
 [(1, 3, 10), (3, 5, 10), (5, 6, 5)],
 [(1, 4, 30), (4, 5, 30), (5, 6, 5)]]
show(method='matplotlib', **kwds)#

Show the (di)graph.

INPUT:

EXAMPLES:

sage: C = graphs.CubeGraph(8)
sage: P = C.plot(vertex_labels=False, vertex_size=0, graph_border=True)     # needs sage.plot
sage: P.show()                      # long time (3s on sage.math, 2011), needs sage.plot
show3d(bgcolor=(1, 1, 1), vertex_colors=None, vertex_size=0.06, edge_colors=None, edge_size=0.02, edge_size2=0.0325, pos3d=None, color_by_label=False, engine='threejs', **kwds)#

Plot the graph and show the resulting plot.

INPUT:

  • bgcolor – rgb tuple (default: (1,1,1))

  • vertex_size – float (default: 0.06)

  • vertex_labels – a boolean (default: False); whether to display vertices using text labels instead of spheres

  • vertex_colors – dictionary (default: None); optional dictionary to specify vertex colors: each key is a color recognizable by tachyon (rgb tuple (default: (1,0,0))), and each corresponding entry is a list of vertices. If a vertex is not listed, it looks invisible on the resulting plot (it doesn’t get drawn).

  • edge_colors – dictionary (default: None); a dictionary specifying edge colors: each key is a color recognized by tachyon (default: (0,0,0)), and each entry is a list of edges.

  • color_by_label – a boolean or dictionary or function (default: False) whether to color each edge with a different color according to its label; the colors are chosen along a rainbow, unless they are specified by a function or dictionary mapping labels to colors; this option is incompatible with edge_color and edge_colors.

  • edge_size – float (default: 0.02)

  • edge_size2 – float (default: 0.0325); used for Tachyon sleeves

  • pos3d – a position dictionary for the vertices

  • layout, iterations, … – layout options; see layout()

  • engine – string (default: 'threejs'); the renderer to use among:

    • 'threejs': interactive web-based 3D viewer using JavaScript and a WebGL renderer

    • 'jmol': interactive 3D viewer using Java

    • 'tachyon': ray tracer generating a static PNG image

  • xres – resolution

  • yres – resolution

  • **kwds – passed on to the rendering engine

EXAMPLES:

sage: G = graphs.CubeGraph(5)
sage: G.show3d(iterations=500, edge_size=None, vertex_size=0.04)    # long time, needs sage.plot

We plot a fairly complicated Cayley graph:

sage: A5 = AlternatingGroup(5); A5                                          # needs sage.groups
Alternating group of order 5!/2 as a permutation group
sage: G = A5.cayley_graph()                                                 # needs sage.groups
sage: G.show3d(vertex_size=0.03,    # long time                             # needs sage.groups sage.plot
....:          edge_size=0.01, edge_size2=0.02,
....:          vertex_colors={(1,1,1): list(G)}, bgcolor=(0,0,0),
....:          color_by_label=True, iterations=200)

Some Tachyon examples:

sage: D = graphs.DodecahedralGraph()
sage: D.show3d(engine='tachyon')    # long time                             # needs sage.plot
sage: G = graphs.PetersenGraph()
sage: G.show3d(engine='tachyon',    # long time                             # needs sage.plot
....:          vertex_colors={(0,0,1): list(G)})
sage: C = graphs.CubeGraph(4)
sage: C.show3d(engine='tachyon',    # long time                             # needs sage.plot
....:          edge_colors={(0,1,0): C.edges(sort=False)},
....:          vertex_colors={(1,1,1): list(C)}, bgcolor=(0,0,0))
sage: K = graphs.CompleteGraph(3)
sage: K.show3d(engine='tachyon',    # long time                             # needs sage.plot
....:          edge_colors={(1,0,0): [(0, 1, None)],
....:                       (0, 1, 0): [(0, 2, None)],
....:                       (0, 0, 1): [(1, 2, None)]})
size()#

Return the number of edges.

Note that num_edges() also returns the number of edges in \(G\).

EXAMPLES:

sage: G = graphs.PetersenGraph()
sage: G.size()
15
spanning_trees_count(root_vertex=None)#

Return the number of spanning trees in a graph.

In the case of a digraph, counts the number of spanning out-trees rooted in root_vertex. Default is to set first vertex as root.

This computation uses Kirchhoff’s Matrix Tree Theorem [1] to calculate the number of spanning trees. For complete graphs on \(n\) vertices the result can also be reached using Cayley’s formula: the number of spanning trees are \(n^(n-2)\).

For digraphs, the augmented Kirchhoff Matrix as defined in [2] is used for calculations. Here the result is the number of out-trees rooted at a specific vertex.

INPUT:

  • root_vertex – a vertex (default: None); the vertex that will be used as root for all spanning out-trees if the graph is a directed graph. Otherwise, the first vertex returned by vertex_iterator() is used. This argument is ignored if the graph is not a digraph.

See also

spanning_trees() – enumerates all spanning trees of a graph

REFERENCES:

AUTHORS:

  • Anders Jonsson (2009-10-10)

EXAMPLES:

sage: G = graphs.PetersenGraph()
sage: G.spanning_trees_count()                                              # needs sage.modules
2000
sage: n = 11
sage: G = graphs.CompleteGraph(n)
sage: ST = G.spanning_trees_count()                                         # needs sage.modules
sage: ST == n ^ (n - 2)                                                     # needs sage.modules
True
sage: # needs sage.modules
sage: M = matrix(3, 3, [0, 1, 0, 0, 0, 1, 1, 1, 0])
sage: D = DiGraph(M)
sage: D.spanning_trees_count()
1
sage: D.spanning_trees_count(0)
1
sage: D.spanning_trees_count(2)
2
spectral_radius(G, prec=1e-10)#

Return an interval of floating point number that encloses the spectral radius of this graph

The input graph G must be strongly connected.

INPUT:

  • prec – (default 1e-10) an upper bound for the relative precision of the interval

The algorithm is iterative and uses an inequality valid for non-negative matrices. Namely, if \(A\) is a non-negative square matrix with Perron-Frobenius eigenvalue \(\lambda\) then the following inequality is valid for any vector \(x\)

\[\min_i \frac{(Ax)_i}{x_i} \leq \lambda \leq \max_i \frac{(Ax)_i}{x_i}\]

Note

The speed of convergence of the algorithm is governed by the spectral gap (the distance to the second largest modulus of other eigenvalues). If this gap is small, then this function might not be appropriate.

The algorithm is not smart and not parallel! It uses basic interval arithmetic and native floating point arithmetic.

EXAMPLES:

sage: from sage.graphs.base.static_sparse_graph import spectral_radius

sage: G = DiGraph([(0,0),(0,1),(1,0)], loops=True)
sage: phi = (RR(1) + RR(5).sqrt() ) / 2
sage: phi  # abs tol 1e-14
1.618033988749895
sage: e_min, e_max = spectral_radius(G, 1e-14)
sage: e_min, e_max     # abs tol 1e-14
(1.618033988749894, 1.618033988749896)
sage: (e_max - e_min)  # abs tol 1e-14
1e-14
sage: e_min < phi < e_max
True

This function also works for graphs:

sage: G = Graph([(0,1),(0,2),(1,2),(1,3),(2,4),(3,4)])
sage: e_min, e_max = spectral_radius(G, 1e-14)
sage: e = max(G.adjacency_matrix().charpoly().roots(AA, multiplicities=False))  # needs sage.modules sage.rings.number_field
sage: e_min < e < e_max                                                         # needs sage.modules sage.rings.number_field sage.symbolic
True

sage: G.spectral_radius()  # abs tol 1e-9
(2.48119430408, 2.4811943041)

A larger example:

sage: # needs sage.modules
sage: G = DiGraph()
sage: G.add_edges((i,i+1) for i in range(200))
sage: G.add_edge(200,0)
sage: G.add_edge(1,0)
sage: e_min, e_max = spectral_radius(G, 0.00001)
sage: p = G.adjacency_matrix(sparse=True).charpoly()
sage: p
x^201 - x^199 - 1
sage: r = p.roots(AA, multiplicities=False)[0]                                  # needs sage.rings.number_field
sage: e_min < r < e_max                                                         # needs sage.rings.number_field
True

A much larger example:

sage: G = DiGraph(100000)
sage: r = list(range(100000))
sage: while not G.is_strongly_connected():
....:     shuffle(r)
....:     G.add_edges(enumerate(r), loops=False)
sage: spectral_radius(G, 1e-10)  # random
(1.9997956006500042, 1.9998043797692782)

The algorithm takes care of multiple edges:

sage: G = DiGraph(2,loops=True,multiedges=True)
sage: G.add_edges([(0,0),(0,0),(0,1),(1,0)])
sage: spectral_radius(G, 1e-14)  # abs tol 1e-14
(2.414213562373094, 2.414213562373095)
sage: max(G.adjacency_matrix().eigenvalues(AA))                                 # needs sage.modules sage.rings.number_field
2.414213562373095?

Some bipartite graphs:

sage: G = Graph([(0,1),(0,3),(2,3)])
sage: G.spectral_radius()  # abs tol 1e-10
(1.6180339887253428, 1.6180339887592732)

sage: G = DiGraph([(0,1),(0,3),(2,3),(3,0),(1,0),(1,2)])
sage: G.spectral_radius() # abs tol 1e-10
(1.5537739740270458, 1.553773974033029)

sage: G = graphs.CompleteBipartiteGraph(1,3)
sage: G.spectral_radius()  # abs tol 1e-10
(1.7320508075688772, 1.7320508075688774)
spectrum(laplacian=False)#

Return a list of the eigenvalues of the adjacency matrix.

INPUT:

  • laplacian – boolean (default: False); if True, use the Laplacian matrix (see kirchhoff_matrix())

OUTPUT:

A list of the eigenvalues, including multiplicities, sorted with the largest eigenvalue first.

See also

The method spectral_radius() returns floating point approximation of the maximum eigenvalue.

EXAMPLES:

sage: P = graphs.PetersenGraph()
sage: P.spectrum()                                                          # needs sage.modules sage.rings.number_field
[3, 1, 1, 1, 1, 1, -2, -2, -2, -2]
sage: P.spectrum(laplacian=True)                                            # needs sage.modules sage.rings.number_field
[5, 5, 5, 5, 2, 2, 2, 2, 2, 0]
sage: D = P.to_directed()
sage: D.delete_edge(7, 9)
sage: D.spectrum()                                                          # needs sage.modules sage.rings.number_field
[2.9032119259..., 1, 1, 1, 1, 0.8060634335..., -1.7092753594..., -2, -2, -2]
sage: C = graphs.CycleGraph(8)
sage: C.spectrum()                                                          # needs sage.modules sage.rings.number_field
[2, 1.4142135623..., 1.4142135623..., 0, 0, -1.4142135623..., -1.4142135623..., -2]

A digraph may have complex eigenvalues. Previously, the complex parts of graph eigenvalues were being dropped. For a 3-cycle, we have:

sage: T = DiGraph({0: [1], 1: [2], 2: [0]})
sage: T.spectrum()                                                          # needs sage.modules sage.rings.number_field
[1, -0.5000000000... + 0.8660254037...*I, -0.5000000000... - 0.8660254037...*I]
steiner_tree(vertices, weighted, solver=False, verbose=None, integrality_tolerance=0)#

Return a tree of minimum weight connecting the given set of vertices.

Definition :

Computing a minimum spanning tree in a graph can be done in \(n \log(n)\) time (and in linear time if all weights are equal) where \(n = V + E\). On the other hand, if one is given a large (possibly weighted) graph and a subset of its vertices, it is NP-Hard to find a tree of minimum weight connecting the given set of vertices, which is then called a Steiner Tree.

See the Wikipedia article Steiner_tree_problem for more information.

INPUT:

  • vertices – the vertices to be connected by the Steiner Tree.

  • weighted – boolean (default: False); whether to consider the graph as weighted, and use each edge’s label as a weight, considering None as a weight of \(1\). If weighted=False (default) all edges are considered to have a weight of \(1\).

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

Note

  • This problem being defined on undirected graphs, the orientation is not considered if the current graph is actually a digraph.

  • The graph is assumed not to have multiple edges.

ALGORITHM:

Solved through Linear Programming.

COMPLEXITY:

NP-Hard.

Note that this algorithm first checks whether the given set of vertices induces a connected graph, returning one of its spanning trees if weighted is set to False, and thus answering very quickly in some cases

EXAMPLES:

The Steiner Tree of the first 5 vertices in a random graph is, of course, always a tree:

sage: g = graphs.RandomGNP(30, .5)
sage: first5 = g.vertices(sort=True)[:5]
sage: st = g.steiner_tree(first5)                                           # needs sage.numerical.mip
sage: st.is_tree()                                                          # needs sage.numerical.mip
True

And all the 5 vertices are contained in this tree

sage: all(v in st for v in first5)                                          # needs sage.numerical.mip
True

An exception is raised when the problem is impossible, i.e. if the given vertices are not all included in the same connected component:

sage: g = 2 * graphs.PetersenGraph()
sage: st = g.steiner_tree([5, 15])
Traceback (most recent call last):
...
EmptySetError: the given vertices do not all belong to the same connected component. This problem has no solution !
strong_product(other)#

Return the strong product of self and other.

The strong product of \(G\) and \(H\) is the graph \(L\) with vertex set \(V(L)=V(G)\times V(H)\), and \(((u,v), (w,x))\) is an edge of \(L\) iff either :

  • \((u, w)\) is an edge of \(G\) and \(v = x\), or

  • \((v, x)\) is an edge of \(H\) and \(u = w\), or

  • \((u, w)\) is an edge of \(G\) and \((v, x)\) is an edge of \(H\).

In other words, the edges of the strong product is the union of the edges of the tensor and Cartesian products.

EXAMPLES:

sage: Z = graphs.CompleteGraph(2)
sage: C = graphs.CycleGraph(5)
sage: S = C.strong_product(Z); S
Graph on 10 vertices
sage: S.plot()                      # long time                             # needs sage.plot
Graphics object consisting of 36 graphics primitives
sage: D = graphs.DodecahedralGraph()
sage: P = graphs.PetersenGraph()
sage: S = D.strong_product(P); S
Graph on 200 vertices
sage: S.plot()                      # long time                             # needs sage.plot
Graphics object consisting of 1701 graphics primitives
subdivide_edge(*args)#

Subdivide an edge \(k\) times.

INPUT:

The following forms are all accepted to subdivide \(8\) times the edge between vertices \(1\) and \(2\) labeled with "my_label".

  • G.subdivide_edge( 1, 2, 8 )

  • G.subdivide_edge( (1, 2), 8 )

  • G.subdivide_edge( (1, 2, "my_label"), 8 )

Note

  • If the given edge is labelled with \(l\), all the edges created by the subdivision will have the same label

  • If no label is given, the label used will be the one returned by the method edge_label() on the pair u,v

EXAMPLES:

Subdividing \(5\) times an edge in a path of length \(3\) makes it a path of length \(8\):

sage: g = graphs.PathGraph(3)
sage: edge = next(g.edge_iterator())
sage: g.subdivide_edge(edge, 5)
sage: g.is_isomorphic(graphs.PathGraph(8))
True

Subdividing a labelled edge in two ways:

sage: g = Graph()
sage: g.add_edge(0, 1, "label1")
sage: g.add_edge(1, 2, "label2")
sage: print(g.edges(sort=True))
[(0, 1, 'label1'), (1, 2, 'label2')]

Specifying the label:

sage: g.subdivide_edge(0, 1, "label1", 3)
sage: print(g.edges(sort=True))
[(0, 3, 'label1'), (1, 2, 'label2'), (1, 5, 'label1'), (3, 4, 'label1'), (4, 5, 'label1')]

The lazy way:

sage: g.subdivide_edge(1, 2, "label2", 5)
sage: print(g.edges(sort=True))
[(0, 3, 'label1'), (1, 5, 'label1'), (1, 6, 'label2'),
 (2, 10, 'label2'), (3, 4, 'label1'), (4, 5, 'label1'),
 (6, 7, 'label2'), (7, 8, 'label2'), (8, 9, 'label2'),
 (9, 10, 'label2')]

If too many arguments are given, an exception is raised

sage: g.subdivide_edge(0,1,1,1,1,1,1,1,1,1,1)
Traceback (most recent call last):
...
ValueError: this method takes at most 4 arguments

The same goes when the given edge does not exist:

sage: g.subdivide_edge(0, 1, "fake_label", 5)
Traceback (most recent call last):
...
ValueError: the given edge does not exist

See also

subdivide_edges(edges, k)#

Subdivide \(k\) times edges from an iterable container.

For more information on the behaviour of this method, please refer to the documentation of subdivide_edge().

INPUT:

  • edges – a list of edges

  • k – integer; common length of the subdivisions

Note

If a given edge is labelled with \(l\), all the edges created by its subdivision will have the same label.

EXAMPLES:

If we are given the disjoint union of several paths:

sage: paths = [2, 5, 9]
sage: paths = map(graphs.PathGraph, paths)
sage: g = Graph()
sage: for P in paths:
....:   g = g + P

Subdividing edges in each of them will only change their lengths:

sage: edges = [next(P.edge_iterator()) for P in g.connected_components_subgraphs()]
sage: k = 6
sage: g.subdivide_edges(edges, k)

Let us check this by creating the graph we expect to have built through subdivision:

sage: paths2 = [2 + k, 5 + k, 9 + k]
sage: paths2 = map(graphs.PathGraph, paths2)
sage: g2 = Graph()
sage: for P in paths2:
....:   g2 = g2 + P
sage: g.is_isomorphic(g2)
True

See also

subgraph(vertices=None, edges=None, inplace=False, vertex_property=None, edge_property=None, algorithm=None, immutable=None)#

Return the subgraph containing the given vertices and edges.

If either vertices or edges are not specified, they are assumed to be all vertices or edges. If edges are not specified, returns the subgraph induced by the vertices.

INPUT:

  • inplace – boolean (default: False); using inplace=True will simply delete the extra vertices and edges from the current graph. This will modify the graph.

  • vertices – a single vertex or an iterable container of vertices, e.g. a list, set, graph, file or numeric array. If not passed (i.e., None), defaults to the entire graph.

  • edges – as with vertices, edges can be a single edge or an iterable container of edges (e.g., a list, set, file, numeric array, etc.). By default (edges=None), all edges are assumed and the returned graph is an induced subgraph. In the case of multiple edges, specifying an edge as \((u,v)\) means to keep all edges \((u,v)\), regardless of the label.

  • vertex_property – function (default: None); a function that inputs a vertex and outputs a boolean value, i.e., a vertex v in vertices is kept if vertex_property(v) == True

  • edge_property – function (default: None); a function that inputs an edge and outputs a boolean value, i.e., a edge e in edges is kept if edge_property(e) == True

  • algorithm – string (default: None); one of the following:

    • If algorithm="delete" or inplace=True, then the graph is constructed by deleting edges and vertices

    • If algorithm="add", then the graph is constructed by building a new graph from the appropriate vertices and edges. Implies inplace=False.

    • If algorithm=None, then the algorithm is chosen based on the number of vertices in the subgraph.

  • immutable – boolean (default: None); whether to create a mutable/immutable subgraph. immutable=None (default) means that the graph and its subgraph will behave the same way.

EXAMPLES:

sage: G = graphs.CompleteGraph(9)
sage: H = G.subgraph([0, 1, 2]); H
Subgraph of (Complete graph): Graph on 3 vertices
sage: G
Complete graph: Graph on 9 vertices
sage: J = G.subgraph(edges=[(0, 1)])
sage: J.edges(sort=True, labels=False)
[(0, 1)]
sage: set(J) == set(G)
True
sage: G.subgraph([0, 1, 2], inplace=True); G
Subgraph of (Complete graph): Graph on 3 vertices
sage: G.subgraph() == G
True
sage: D = digraphs.Complete(9)
sage: H = D.subgraph([0, 1, 2]); H
Subgraph of (Complete digraph): Digraph on 3 vertices
sage: H = D.subgraph(edges=[(0, 1), (0, 2)])
sage: H.edges(sort=True, labels=False)
[(0, 1), (0, 2)]
sage: set(H) == set(D)
True
sage: D
Complete digraph: Digraph on 9 vertices
sage: D.subgraph([0, 1, 2], inplace=True); D
Subgraph of (Complete digraph): Digraph on 3 vertices
sage: D.subgraph() == D
True

A more complicated example involving multiple edges and labels:

sage: G = Graph(multiedges=True, sparse=True)
sage: G.add_edges([(0, 1, 'a'), (0, 1, 'b'), (1, 0, 'c'), (0, 2, 'd'), (0, 2, 'e'), (2, 0, 'f'), (1, 2, 'g')])
sage: G.subgraph(edges=[(0, 1), (0, 2,'d'), (0, 2, 'not in graph')]).edges(sort=True)
[(0, 1, 'a'), (0, 1, 'b'), (0, 1, 'c'), (0, 2, 'd')]
sage: J = G.subgraph(vertices=[0, 1], edges=[(0, 1, 'a'), (0, 2, 'c')])
sage: J.edges(sort=True)
[(0, 1, 'a')]
sage: J.vertices(sort=True)
[0, 1]
sage: G.subgraph(vertices=G) == G
True
sage: D = DiGraph(multiedges=True, sparse=True)
sage: D.add_edges([(0, 1, 'a'), (0, 1, 'b'), (1, 0, 'c'), (0, 2, 'd'), (0, 2, 'e'), (2, 0, 'f'), (1, 2, 'g')])
sage: D.subgraph(edges=[(0, 1), (0, 2, 'd'), (0, 2, 'not in graph')]).edges(sort=True)
[(0, 1, 'a'), (0, 1, 'b'), (0, 2, 'd')]
sage: H = D.subgraph(vertices=[0, 1], edges=[(0, 1, 'a'), (0, 2, 'c')])
sage: H.edges(sort=True)
[(0, 1, 'a')]
sage: H.vertices(sort=True)
[0, 1]

Using the property arguments:

sage: P = graphs.PetersenGraph()
sage: S = P.subgraph(vertex_property=lambda v: not (v % 2))
sage: S.vertices(sort=True)
[0, 2, 4, 6, 8]
sage: C = graphs.CubeGraph(2)
sage: S = C.subgraph(edge_property=(lambda e: e[0][0] == e[1][0]))
sage: C.edges(sort=True)
[('00', '01', None), ('00', '10', None), ('01', '11', None), ('10', '11', None)]
sage: S.edges(sort=True)
[('00', '01', None), ('10', '11', None)]

The algorithm is not specified, then a reasonable choice is made for speed:

sage: g = graphs.PathGraph(1000)
sage: g.subgraph(list(range(10)))  # uses the 'add' algorithm
Subgraph of (Path graph): Graph on 10 vertices

Return a copy of G in self.

INPUT:

  • G – the (di)graph whose copy we are looking for in self

  • induced – boolean (default: False); whether or not to search for an induced copy of G in self

OUTPUT:

If induced=False, return a copy of G in this graph. Otherwise, return an induced copy of G in self. If G is the empty graph, return the empty graph since it is a subgraph of every graph. Now suppose G is not the empty graph. If there is no copy (induced or otherwise) of G in self, we return None.

Note

The vertex labels and the edge labels in the graph are ignored.

See also

ALGORITHM:

See the documentation of SubgraphSearch.

EXAMPLES:

The Petersen graph contains the path graph \(P_5\):

sage: g = graphs.PetersenGraph()
sage: h1 = g.subgraph_search(graphs.PathGraph(5)); h1                      # needs sage.modules
Subgraph of (Petersen graph): Graph on 5 vertices
sage: h1.vertices(sort=True); h1.edges(sort=True, labels=False)            # needs sage.modules
[0, 1, 2, 3, 4]
[(0, 1), (1, 2), (2, 3), (3, 4)]
sage: I1 = g.subgraph_search(graphs.PathGraph(5), induced=True); I1        # needs sage.modules
Subgraph of (Petersen graph): Graph on 5 vertices
sage: I1.vertices(sort=True); I1.edges(sort=True, labels=False)            # needs sage.modules
[0, 1, 2, 3, 8]
[(0, 1), (1, 2), (2, 3), (3, 8)]

It also contains the claw \(K_{1,3}\):

sage: # needs sage.modules
sage: h2 = g.subgraph_search(graphs.ClawGraph()); h2
Subgraph of (Petersen graph): Graph on 4 vertices
sage: h2.vertices(sort=True); h2.edges(sort=True, labels=False)
[0, 1, 4, 5]
[(0, 1), (0, 4), (0, 5)]
sage: I2 = g.subgraph_search(graphs.ClawGraph(), induced=True); I2
Subgraph of (Petersen graph): Graph on 4 vertices
sage: I2.vertices(sort=True); I2.edges(sort=True, labels=False)
[0, 1, 4, 5]
[(0, 1), (0, 4), (0, 5)]

Of course the induced copies are isomorphic to the graphs we were looking for:

sage: I1.is_isomorphic(graphs.PathGraph(5))                                # needs sage.modules
True
sage: I2.is_isomorphic(graphs.ClawGraph())                                 # needs sage.modules
True

However, the Petersen graph does not contain a subgraph isomorphic to \(K_3\):

sage: g.subgraph_search(graphs.CompleteGraph(3)) is None                   # needs sage.modules
True

Nor does it contain a nonempty induced subgraph isomorphic to \(P_6\):

sage: g.subgraph_search(graphs.PathGraph(6), induced=True) is None         # needs sage.modules
True

The empty graph is a subgraph of every graph:

sage: g.subgraph_search(graphs.EmptyGraph())                               # needs sage.modules
Graph on 0 vertices
sage: g.subgraph_search(graphs.EmptyGraph(), induced=True)                 # needs sage.modules
Graph on 0 vertices

The subgraph may just have edges missing:

sage: k3 = graphs.CompleteGraph(3); p3 = graphs.PathGraph(3)
sage: k3.relabel(list('abc'))
sage: s = k3.subgraph_search(p3)                                            # needs sage.modules
sage: s.edges(sort=True, labels=False)                                      # needs sage.modules
[('a', 'b'), ('b', 'c')]

Of course, \(P_3\) is not an induced subgraph of \(K_3\), though:

sage: k3 = graphs.CompleteGraph(3); p3 = graphs.PathGraph(3)
sage: k3.relabel(list('abc'))
sage: k3.subgraph_search(p3, induced=True) is None                          # needs sage.modules
True

If the graph has labels, the labels are just ignored:

sage: g.set_vertex(0, 'foo')
sage: c = g.subgraph_search(graphs.PathGraph(5))                            # needs sage.modules
sage: c.get_vertices()                                                      # needs sage.modules
{0: 'foo', 1: None, 2: None, 3: None, 4: None}
subgraph_search_count(G, induced=False)#

Return the number of labelled occurrences of G in self.

INPUT:

  • G – the (di)graph whose copies we are looking for in self

  • induced – boolean (default: False); whether or not to count induced copies of G in self

Note

The vertex labels and the edge labels in the graph are ignored.

ALGORITHM:

See the documentation of SubgraphSearch.

See also

EXAMPLES:

Counting the number of paths \(P_5\) in a PetersenGraph:

sage: g = graphs.PetersenGraph()
sage: g.subgraph_search_count(graphs.PathGraph(5))                          # needs sage.modules
240

Requiring these subgraphs be induced:

sage: g.subgraph_search_count(graphs.PathGraph(5), induced=True)            # needs sage.modules
120

If we define the graph \(T_k\) (the transitive tournament on \(k\) vertices) as the graph on \(\{0, ..., k-1\}\) such that \(ij \in T_k\) if \(i<j\), how many directed triangles can be found in \(T_5\) ? The answer is of course \(0\):

sage: T5 = digraphs.TransitiveTournament(5)
sage: T5.subgraph_search_count(digraphs.Circuit(3))                        # needs sage.modules
0

If we count instead the number of \(T_3\) in \(T_5\), we expect the answer to be \(\binom{5}{3}\):

sage: T3 = digraphs.TransitiveTournament(3)
sage: T5.subgraph_search_count(T3)                                         # needs sage.modules
10
sage: binomial(5,3)                                                        # needs sage.symbolic
10
sage: T3.is_isomorphic(T5.subgraph(vertices=[0, 1, 2]))                    # needs sage.modules
True

The empty graph is a subgraph of every graph:

sage: g.subgraph_search_count(graphs.EmptyGraph())                          # needs sage.modules
1

If the graph has vertex labels or edge labels, the label is just ignored:

sage: g.set_vertex(0, 'foo')
sage: g.subgraph_search_count(graphs.PathGraph(5))                          # needs sage.modules
240
subgraph_search_iterator(G, induced=False, return_graphs=True)#

Return an iterator over the labelled copies of G in self.

INPUT:

  • G – the graph whose copies we are looking for in self

  • induced – boolean (default: False); whether or not to iterate over the induced copies of G in self

  • return_graphs – boolean (default: True); whether to return (di)graphs or only the list of vertices of the found (di)graphs

Note

The vertex labels and the edge labels in the graph are ignored.

ALGORITHM:

See the documentation of SubgraphSearch.

OUTPUT:

Iterator over the labelled copies of G in self, as lists or (di)graphs. For each value \((v_1, v_2, ..., v_k)\) returned, the first vertex of \(G\) is associated with \(v_1\), the second with \(v_2\), etc.

Note

This method works on Graph, DiGraph and BipartiteGraph.

See also

EXAMPLES:

Iterating through all the labelled \(P_3\) of \(P_5\):

sage: g = graphs.PathGraph(5)
sage: P3 = graphs.PathGraph(3)
sage: for p in g.subgraph_search_iterator(P3, return_graphs=False):         # needs sage.modules
....:     print(p)
[0, 1, 2]
[1, 2, 3]
[2, 1, 0]
[2, 3, 4]
[3, 2, 1]
[4, 3, 2]
sage: for p in g.subgraph_search_iterator(P3, return_graphs=True):          # needs sage.modules
....:     print(p)
Subgraph of (Path graph)
Subgraph of (Path graph)
Subgraph of (Path graph)
Subgraph of (Path graph)
Subgraph of (Path graph)
Subgraph of (Path graph)
sage: all(h.is_isomorphic(P3) for h in g.subgraph_search_iterator(P3))      # needs sage.modules
True

If the graph has vertex labels or edge labels, the label is just ignored:

sage: g.set_vertex(0, 'foo')
sage: for p in g.subgraph_search_iterator(P3, return_graphs=False):         # needs sage.modules
....:     print(p)
[0, 1, 2]
[1, 2, 3]
[2, 1, 0]
[2, 3, 4]
[3, 2, 1]
[4, 3, 2]

Search for induced subgraphs:

sage: H = graphs.HouseGraph()
sage: P4 = graphs.PathGraph(4)
sage: all(h.is_isomorphic(P4)                                               # needs sage.modules
....:     for h in H.subgraph_search_iterator(P4, induced=True))
True
sage: sum(1 for h in H.subgraph_search_iterator(P4, induced=True))          # needs sage.modules
4
sage: sum(1 for h in H.subgraph_search_iterator(P4, induced=False))         # needs sage.modules
20

Search for subdigraphs:

sage: H = digraphs.Complete(5)
sage: P4 = digraphs.Path(4)
sage: sum(1 for _ in H.subgraph_search_iterator(P4, induced=True))          # needs sage.modules
0
sage: sum(1 for _ in H.subgraph_search_iterator(P4, induced=False))         # needs sage.modules
120

This method also works for bipartite graphs:

sage: K33 = BipartiteGraph(graphs.CompleteBipartiteGraph(3, 3))
sage: K22 = BipartiteGraph(graphs.CompleteBipartiteGraph(2, 2))
sage: sum(1 for _ in K33.subgraph_search_iterator(K22))                     # needs sage.modules
72
symmetric_edge_polytope(backend=None)#

Return the symmetric edge polytope of self.

The symmetric edge polytope (SEP) of a Graph on \(n\) vertices is the polytope in \(\ZZ^{n}\) defined as the convex hull of \(e_i - e_j\) and \(e_j - e_i\) for each edge \((i, j)\). Here \(e_1, \dots, e_n\) denotes the standard basis.

INPUT:

  • backend – string or None (default); the backend to use; see sage.geometry.polyhedron.constructor.Polyhedron()

EXAMPLES:

The SEP of a \(4\)-cycle is a cube:

sage: G = graphs.CycleGraph(4)
sage: P = G.symmetric_edge_polytope(); P                                    # needs sage.geometry.polyhedron
A 3-dimensional polyhedron in ZZ^4 defined as the convex hull of 8 vertices
sage: P.is_combinatorially_isomorphic(polytopes.cube())                     # needs sage.geometry.polyhedron
True

The SEP of a complete graph on \(4\) vertices is a cuboctahedron:

sage: G = graphs.CompleteGraph(4)
sage: P = G.symmetric_edge_polytope(); P                                    # needs sage.geometry.polyhedron
A 3-dimensional polyhedron in ZZ^4 defined as the convex hull of 12 vertices
sage: P.is_combinatorially_isomorphic(polytopes.cuboctahedron())            # needs sage.geometry.polyhedron
True

The SEP of a graph with edges on \(n\) vertices has dimension \(n\) minus the number of connected components:

sage: n = randint(5, 12)
sage: G = Graph()
sage: while not G.num_edges():                                              # needs networkx
....:     G = graphs.RandomGNP(n, 0.2)
sage: P = G.symmetric_edge_polytope()                                       # needs networkx sage.geometry.polyhedron
sage: P.ambient_dim() == n                                                  # needs networkx sage.geometry.polyhedron
True
sage: P.dim() == n - G.connected_components_number()                        # needs networkx sage.geometry.polyhedron
True

The SEP of a graph is isomorphic to the subdirect sum of its connected components SEP’s:

sage: n = randint(3, 6)
sage: G1 = graphs.RandomGNP(n, 0.2)                                         # needs networkx
sage: n = randint(3, 6)
sage: G2 = graphs.RandomGNP(n, 0.2)                                         # needs networkx
sage: G = G1.disjoint_union(G2)                                             # needs networkx
sage: P = G.symmetric_edge_polytope()                                       # needs networkx sage.geometry.polyhedron
sage: P1 = G1.symmetric_edge_polytope()                                     # needs networkx sage.geometry.polyhedron
sage: P2 = G2.symmetric_edge_polytope()                                     # needs networkx sage.geometry.polyhedron
sage: P.is_combinatorially_isomorphic(P1.subdirect_sum(P2))                 # needs networkx sage.geometry.polyhedron
True

All trees on \(n\) vertices have isomorphic SEPs:

sage: n = randint(4, 10)
sage: G1 = graphs.RandomTree(n)
sage: G2 = graphs.RandomTree(n)
sage: P1 = G1.symmetric_edge_polytope()                                     # needs sage.geometry.polyhedron
sage: P2 = G2.symmetric_edge_polytope()                                     # needs sage.geometry.polyhedron
sage: P1.is_combinatorially_isomorphic(P2)                                  # needs sage.geometry.polyhedron
True

However, there are still many different SEPs:

sage: len(list(graphs(5)))
34
sage: polys = []
sage: for G in graphs(5):                                                   # needs sage.geometry.polyhedron
....:     P = G.symmetric_edge_polytope()
....:     for P1 in polys:
....:         if P.is_combinatorially_isomorphic(P1):
....:             break
....:     else:
....:         polys.append(P)
sage: len(polys)                                                            # needs sage.geometry.polyhedron
25

A non-trivial example of two graphs with isomorphic SEPs:

sage: G1 = graphs.CycleGraph(4)
sage: G1.add_edges([[0, 5], [5, 2], [1, 6], [6, 2]])
sage: G2 = copy(G1)
sage: G1.add_edges([[2, 7], [7, 3]])
sage: G2.add_edges([[0, 7], [7, 3]])
sage: G1.is_isomorphic(G2)
False
sage: P1 = G1.symmetric_edge_polytope()                                     # needs sage.geometry.polyhedron
sage: P2 = G2.symmetric_edge_polytope()                                     # needs sage.geometry.polyhedron
sage: P1.is_combinatorially_isomorphic(P2)                                  # needs sage.geometry.polyhedron
True

Apparently, glueing two graphs together on a vertex gives isomorphic SEPs:

sage: n = randint(3, 7)
sage: g1 = graphs.RandomGNP(n, 0.2)                                         # needs networkx
sage: g2 = graphs.RandomGNP(n, 0.2)                                         # needs networkx
sage: G = g1.disjoint_union(g2)                                             # needs networkx
sage: H = copy(G)                                                           # needs networkx
sage: G.merge_vertices(((0, randrange(n)), (1, randrange(n))))              # needs networkx
sage: H.merge_vertices(((0, randrange(n)), (1, randrange(n))))              # needs networkx
sage: PG = G.symmetric_edge_polytope()                                      # needs networkx sage.geometry.polyhedron
sage: PH = H.symmetric_edge_polytope()                                      # needs networkx sage.geometry.polyhedron
sage: PG.is_combinatorially_isomorphic(PH)                                  # needs networkx sage.geometry.polyhedron
True
szeged_index(G, algorithm=None)#

Return the Szeged index of the graph \(G\).

Let \(G = (V, E)\) be a connected graph, and for any \(uv\in E\), let \(N_u(uv) = \{w\in V:d(u,w)<d(v,w)\}\) and \(n_u(uv)=|N_u(uv)|\). The Szeged index of \(G\) is then defined as [KRG1996]

\[`\sum_{uv \in E(G)}n_u(uv)\times n_v(uv)`\]

See the Wikipedia article Szeged_index for more details.

INPUT:

  • G – a Sage graph

  • algorithm – string (default: None); algorithm to use among:

    • "low" – algorithm with time complexity in \(O(nm)\) and space complexity in \(O(m)\). This implementation is currently valid only for simple (without loops or multiple edges) connected graphs.

    • "high" – algorithm with time complexity in \(O(nm)\) and space complexity in \(O(n^2)\). It cannot be used on graphs with more than \(65536 = 2^{16}\) vertices.

    By default (None), the "low" algorithm is used for graphs and the "high" algorithm for digraphs.

EXAMPLES:

True for any connected graph [KRG1996]:

sage: from sage.graphs.distances_all_pairs import szeged_index
sage: g = graphs.PetersenGraph()
sage: g.wiener_index() <= szeged_index(g)
True

True for all trees [KRG1996]:

sage: g = Graph()
sage: g.add_edges(graphs.CubeGraph(5).min_spanning_tree())
sage: g.wiener_index() == szeged_index(g)
True

Check that both algorithms return same value:

sage: # long time, needs networkx
sage: G = graphs.RandomBarabasiAlbert(100, 2)
sage: a = szeged_index(G, algorithm='low')
sage: b = szeged_index(G, algorithm='high')
sage: a == b
True

The Szeged index of a directed circuit of order \(n\) is \((n-1)^2\):

sage: [digraphs.Circuit(n).szeged_index() for n in range(1, 8)]
[0, 1, 4, 9, 16, 25, 36]
tensor_product(other)#

Return the tensor product of self and other.

The tensor product of \(G\) and \(H\) is the graph \(L\) with vertex set \(V(L)\) equal to the Cartesian product of the vertices \(V(G)\) and \(V(H)\), and \(((u,v), (w,x))\) is an edge iff - \((u, w)\) is an edge of self, and - \((v, x)\) is an edge of other.

The tensor product is also known as the categorical product and the Kronecker product (referring to the Kronecker matrix product). See the Wikipedia article Kronecker_product.

EXAMPLES:

sage: Z = graphs.CompleteGraph(2)
sage: C = graphs.CycleGraph(5)
sage: T = C.tensor_product(Z); T
Graph on 10 vertices
sage: T.size()
10
sage: T.plot()                      # long time                             # needs sage.plot
Graphics object consisting of 21 graphics primitives
sage: D = graphs.DodecahedralGraph()
sage: P = graphs.PetersenGraph()
sage: T = D.tensor_product(P); T
Graph on 200 vertices
sage: T.size()
900
sage: T.plot()                      # long time                             # needs sage.plot
Graphics object consisting of 1101 graphics primitives
to_dictionary(edge_labels=False, multiple_edges=False)#

Return the graph as a dictionary.

INPUT:

  • edge_labels – boolean (default: False); whether to include edge labels in the output

  • multiple_edges – boolean (default: False); whether to include multiple edges in the output

OUTPUT:

The output depends on the input:

  • If edge_labels == False and multiple_edges == False, the output is a dictionary associating to each vertex the list of its neighbors.

  • If edge_labels == False and multiple_edges == True, the output is a dictionary the same as previously with one difference: the neighbors are listed with multiplicity.

  • If edge_labels == True and multiple_edges == False, the output is a dictionary associating to each vertex \(u\) [a dictionary associating to each vertex \(v\) incident to \(u\) the label of edge \((u,v)\)].

  • If edge_labels == True and multiple_edges == True, the output is a dictionary associating to each vertex \(u\) [a dictionary associating to each vertex \(v\) incident to \(u\) [the list of labels of all edges between \(u\) and \(v\)]].

Note

When used on directed graphs, the explanations above can be understood by replacing the word “neighbors” by “out-neighbors”

EXAMPLES:

sage: g = graphs.PetersenGraph().to_dictionary()
sage: [(key, sorted(g[key])) for key in g]
[(0, [1, 4, 5]),
 (1, [0, 2, 6]),
 (2, [1, 3, 7]),
 (3, [2, 4, 8]),
 (4, [0, 3, 9]),
 (5, [0, 7, 8]),
 (6, [1, 8, 9]),
 (7, [2, 5, 9]),
 (8, [3, 5, 6]),
 (9, [4, 6, 7])]
sage: graphs.PetersenGraph().to_dictionary(multiple_edges=True)
{0: [1, 4, 5], 1: [0, 2, 6],
 2: [1, 3, 7], 3: [2, 4, 8],
 4: [0, 3, 9], 5: [0, 7, 8],
 6: [1, 8, 9], 7: [2, 5, 9],
 8: [3, 5, 6], 9: [4, 6, 7]}
sage: graphs.PetersenGraph().to_dictionary(edge_labels=True)
{0: {1: None, 4: None, 5: None},
 1: {0: None, 2: None, 6: None},
 2: {1: None, 3: None, 7: None},
 3: {2: None, 4: None, 8: None},
 4: {0: None, 3: None, 9: None},
 5: {0: None, 7: None, 8: None},
 6: {1: None, 8: None, 9: None},
 7: {2: None, 5: None, 9: None},
 8: {3: None, 5: None, 6: None},
 9: {4: None, 6: None, 7: None}}
sage: graphs.PetersenGraph().to_dictionary(edge_labels=True,multiple_edges=True)
{0: {1: [None], 4: [None], 5: [None]},
 1: {0: [None], 2: [None], 6: [None]},
 2: {1: [None], 3: [None], 7: [None]},
 3: {2: [None], 4: [None], 8: [None]},
 4: {0: [None], 3: [None], 9: [None]},
 5: {0: [None], 7: [None], 8: [None]},
 6: {1: [None], 8: [None], 9: [None]},
 7: {2: [None], 5: [None], 9: [None]},
 8: {3: [None], 5: [None], 6: [None]},
 9: {4: [None], 6: [None], 7: [None]}}
to_simple(to_undirected=True, keep_label='any', immutable=None)#

Return a simple version of the self.

In particular, loops and multiple edges are removed, and the graph might optionally be converted to an undirected graph.

INPUT:

  • to_undirected – boolean (default: True); if True, the graph is also converted to an undirected graph

  • keep_label – string (default: 'any'); if there are multiple edges with different labels, this variable defines which label should be kept:

    • 'any' – any label

    • 'min' – the smallest label

    • 'max' – the largest label

Warning

'min' and 'max' only works if the labels can be compared. A TypeError might be raised when working with non-comparable objects in Python 3.

  • immutable – boolean (default: Non); whether to create a mutable/immutable copy. immutable=None (default) means that the graph and its copy will behave the same way.

EXAMPLES:

sage: G = DiGraph(loops=True, multiedges=True, sparse=True)
sage: G.add_edges([(0, 0, None), (1, 1, None), (2, 2, None),
....:              (2, 3, 1), (2, 3, 2), (3, 2, None)])
sage: G.edges(sort=True, labels=False)
[(0, 0), (1, 1), (2, 2), (2, 3), (2, 3), (3, 2)]
sage: H = G.to_simple()
sage: H.edges(sort=True, labels=False)
[(2, 3)]
sage: H.is_directed()
False
sage: H.allows_loops()
False
sage: H.allows_multiple_edges()
False
sage: G.to_simple(to_undirected=False, keep_label='min').edges(sort=True)
[(2, 3, 1), (3, 2, None)]
sage: G.to_simple(to_undirected=False, keep_label='max').edges(sort=True)
[(2, 3, 2), (3, 2, None)]
transitive_closure(loops=True)#

Return the transitive closure of the (di)graph.

The transitive closure of a graph \(G\) has an edge \((x, y)\) if and only if there is a path between \(x\) and \(y\) in \(G\).

The transitive closure of any (strongly) connected component of a (di)graph is a complete graph. The transitive closure of a directed acyclic graph is a directed acyclic graph representing the full partial order.

Note

If the (di)graph allows loops, its transitive closure will by default have one loop edge per vertex. This can be prevented by disallowing loops in the (di)graph (self.allow_loops(False)).

EXAMPLES:

sage: g = graphs.PathGraph(4)
sage: g.transitive_closure()
Transitive closure of Path graph: Graph on 4 vertices
sage: g.transitive_closure().is_isomorphic(graphs.CompleteGraph(4))
True
sage: g = DiGraph({0: [1, 2], 1: [3], 2: [4, 5]})
sage: g.transitive_closure().edges(sort=True, labels=False)
[(0, 1), (0, 2), (0, 3), (0, 4), (0, 5), (1, 3), (2, 4), (2, 5)]

On an immutable digraph:

sage: digraphs.Path(5).copy(immutable=True).transitive_closure()
Transitive closure of Path: Digraph on 5 vertices

The transitive closure of a (di)graph allowing loops has by default a loop edge per vertex. Parameter loops allows to prevent that:

sage: G = digraphs.Circuit(3)
sage: G.transitive_closure().loop_edges(labels=False)
[]
sage: G.allow_loops(True)
sage: G.transitive_closure().loop_edges(labels=False)
[(0, 0), (1, 1), (2, 2)]
sage: G = graphs.CycleGraph(3)
sage: G.transitive_closure().loop_edges(labels=False)
[]
sage: G.allow_loops(True)
sage: G.transitive_closure().loop_edges(labels=False)
[(0, 0), (1, 1), (2, 2)]
transitive_reduction()#

Return a transitive reduction of a graph.

A transitive reduction \(H\) of \(G\) has a path from \(x\) to \(y\) if and only if there was a path from \(x\) to \(y\) in \(G\). Deleting any edge of \(H\) destroys this property. A transitive reduction is not unique in general. A transitive reduction has the same transitive closure as the original graph.

A transitive reduction of a complete graph is a tree. A transitive reduction of a tree is itself.

EXAMPLES:

sage: g = graphs.PathGraph(4)
sage: g.transitive_reduction() == g
True
sage: g = graphs.CompleteGraph(5)
sage: h = g.transitive_reduction(); h.size()
4
sage: g = DiGraph({0: [1, 2], 1: [2, 3, 4, 5], 2: [4, 5]})
sage: g.transitive_reduction().size()
5
traveling_salesman_problem(use_edge_labels, maximize=False, solver=False, constraint_generation=None, verbose=None, verbose_constraints=0, integrality_tolerance=False)#

Solve the traveling salesman problem (TSP)

Given a graph (resp. a digraph) \(G\) with weighted edges, the traveling salesman problem consists in finding a Hamiltonian cycle (resp. circuit) of the graph of minimum cost.

This TSP is one of the most famous NP-Complete problems, this function can thus be expected to take some time before returning its result.

INPUT:

  • use_edge_labels – boolean (default: False); whether to solve the weighted traveling salesman problem where the weight of an edge is defined by its label (a label set to None or {} being considered as a weight of \(1\)), or the non-weighted version (i.e., the Hamiltonian cycle problem)

  • maximize – boolean (default: False); whether to compute a minimum (default) or a maximum (when maximize == True) weight tour (or Hamiltonian cycle). This parameter is considered only if use_edge_labels == True.

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

  • constraint_generation – boolean (default: None); whether to use constraint generation when solving the Mixed Integer Linear Program.

    When constraint_generation = None, constraint generation is used whenever the graph has a density larger than 70%.

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

  • verbose_constraints – boolean (default: False); whether to display which constraints are being generated

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

OUTPUT:

A solution to the TSP, as a Graph object whose vertex set is \(V(G)\), and whose edges are only those of the solution.

ALGORITHM:

This optimization problem is solved through the use of Linear Programming.

Note

This function is correctly defined for both graph and digraphs. In the second case, the returned cycle is a circuit of optimal cost.

EXAMPLES:

The Heawood graph is known to be Hamiltonian:

sage: g = graphs.HeawoodGraph()
sage: tsp = g.traveling_salesman_problem(); tsp                             # needs sage.numerical.mip
TSP from Heawood graph: Graph on 14 vertices

The solution to the TSP has to be connected:

sage: tsp.is_connected()                                                    # needs sage.numerical.mip
True

It must also be a \(2\)-regular graph:

sage: tsp.is_regular(k=2)                                                   # needs sage.numerical.mip
True

And obviously it is a subgraph of the Heawood graph:

sage: tsp.is_subgraph(g, induced=False)                                     # needs sage.numerical.mip
True

On the other hand, the Petersen Graph is known not to be Hamiltonian:

sage: g = graphs.PetersenGraph()
sage: tsp = g.traveling_salesman_problem()                                  # needs sage.numerical.mip
Traceback (most recent call last):
...
EmptySetError: the given graph is not Hamiltonian

One easy way to change it is obviously to add to this graph the edges corresponding to a Hamiltonian cycle. If we do this by setting the cost of these new edges to \(2\), while the others are set to \(1\), we notice that not all the edges we added are used in the optimal solution

sage: for u, v in g.edge_iterator(labels=None):
....:    g.set_edge_label(u, v, 1)

sage: cycle = graphs.CycleGraph(10)
sage: for u,v in cycle.edges(labels=None, sort=False):
....:    if not g.has_edge(u, v):
....:        g.add_edge(u, v)
....:    g.set_edge_label(u, v, 2)

sage: tsp = g.traveling_salesman_problem(use_edge_labels=True)              # needs sage.numerical.mip
sage: sum( tsp.edge_labels() ) < 2 * 10                                     # needs sage.numerical.mip
True

If we pick \(1/2\) instead of \(2\) as a cost for these new edges, they clearly become the optimal solution:

sage: for u, v in cycle.edges(labels=None, sort=False):
....:    g.set_edge_label(u,v,1/2)

sage: tsp = g.traveling_salesman_problem(use_edge_labels=True)              # needs sage.numerical.mip
sage: sum(tsp.edge_labels()) == (1/2) * 10                                  # needs sage.numerical.mip
True

Search for a minimum and a maximum weight Hamiltonian cycle:

sage: # needs sage.numerical.mip
sage: G = Graph([(0, 1, 1), (0, 2, 2), (0, 3, 1), (1, 2, 1), (1, 3, 2), (2, 3, 1)])
sage: tsp = G.traveling_salesman_problem(use_edge_labels=True,
....:                                    maximize=False)
sage: print(sum(tsp.edge_labels()))
4
sage: tsp = G.traveling_salesman_problem(use_edge_labels=True,
....:                                    maximize=True)
sage: print(sum(tsp.edge_labels()))
6
triangles_count(algorithm=None)#

Return the number of triangles in the (di)graph.

For digraphs, we count the number of directed circuit of length 3.

INPUT:

  • algorithm – string (default: None); specifies the algorithm to use (note that only 'iter' is available for directed graphs):

EXAMPLES:

The Petersen graph is triangle free and thus:

sage: G = graphs.PetersenGraph()
sage: G.triangles_count()
0

Any triple of vertices in the complete graph induces a triangle so we have:

sage: G = graphs.CompleteGraph(15)
sage: G.triangles_count() == binomial(15, 3)                                # needs sage.symbolic
True

The 2-dimensional DeBruijn graph of 2 symbols has 2 directed \(C_3\):

sage: G = digraphs.DeBruijn(2,2)                                            # needs sage.combinat
sage: G.triangles_count()                                                   # needs sage.combinat
2

The directed \(n\)-cycle is trivially triangle free for \(n > 3\):

sage: G = digraphs.Circuit(10)
sage: G.triangles_count()                                                   # needs sage.modules
0
union(other, immutable=None)#

Return the union of self and other.

If the graphs have common vertices, the common vertices will be identified.

If one of the two graphs allows loops (or multiple edges), the resulting graph will allow loops (or multiple edges).

If both graphs are weighted the resulting graphs is weighted.

If both graphs are immutable, the resulting graph is immutable, unless requested otherwise.

INPUT:

  • immutable – boolean (default: None); whether to create a mutable/immutable union. immutable=None (default) means that the graphs and their union will behave the same way.

EXAMPLES:

sage: G = graphs.CycleGraph(3)
sage: H = graphs.CycleGraph(4)
sage: J = G.union(H); J
Graph on 4 vertices
sage: J.vertices(sort=True)
[0, 1, 2, 3]
sage: J.edges(sort=True, labels=False)
[(0, 1), (0, 2), (0, 3), (1, 2), (2, 3)]
vertex_boundary(vertices1, vertices2=None)#

Return a list of all vertices in the external boundary of vertices1, intersected with vertices2.

If vertices2 is None, then vertices2 is the complement of vertices1. This is much faster if vertices1 is smaller than vertices2.

The external boundary of a set of vertices is the union of the neighborhoods of each vertex in the set. Note that in this implementation, since vertices2 defaults to the complement of vertices1, if a vertex \(v\) has a loop, then vertex_boundary(v) will not contain \(v\).

In a digraph, the external boundary of a vertex \(v\) are those vertices \(u\) with an arc \((v, u)\).

EXAMPLES:

sage: G = graphs.CubeGraph(4)
sage: l = ['0111', '0000', '0001', '0011', '0010', '0101', '0100', '1111', '1101', '1011', '1001']
sage: sorted(G.vertex_boundary(['0000', '1111'], l))
['0001', '0010', '0100', '0111', '1011', '1101']
sage: D = DiGraph({0: [1, 2], 3: [0]})
sage: D.vertex_boundary([0])
[1, 2]
vertex_connectivity(G, value_only=True, sets=False, k=None, solver=None, verbose=0, integrality_tolerance=0.001)#

Return the vertex connectivity of the graph.

For more information, see the Wikipedia article Connectivity_(graph_theory) and the Wikipedia article K-vertex-connected_graph.

Note

  • When the graph is directed, this method actually computes the strong connectivity, (i.e. a directed graph is strongly \(k\)-connected if there are \(k\) vertex disjoint paths between any two vertices \(u, v\)). If you do not want to consider strong connectivity, the best is probably to convert your DiGraph object to a Graph object, and compute the connectivity of this other graph.

  • By convention, a complete graph on \(n\) vertices is \(n-1\) connected. In this case, no certificate can be given as there is no pair of vertices split by a cut of order \(k-1\). For this reason, the certificates returned in this situation are empty.

INPUT:

  • G – the input Sage (Di)Graph

  • value_only – boolean (default: True)

    • When set to True (default), only the value is returned.

    • When set to False, both the value and a minimum vertex cut are returned.

  • sets – boolean (default: False); whether to also return the two

    sets of vertices that are disconnected by the cut (implies value_only=False)

  • k – integer (default: None); when specified, check if the vertex connectivity of the (di)graph is larger or equal to \(k\). The method thus outputs a boolean only.

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

EXAMPLES:

A basic application on a PappusGraph:

sage: from sage.graphs.connectivity import vertex_connectivity
sage: g = graphs.PappusGraph()
sage: vertex_connectivity(g)                                                     # needs sage.numerical.mip
3
sage: g.vertex_connectivity()                                                    # needs sage.numerical.mip
3

In a grid, the vertex connectivity is equal to the minimum degree, in which case one of the two sets is of cardinality \(1\):

sage: g = graphs.GridGraph([ 3,3 ])
sage: [value, cut, [ setA, setB ]] = vertex_connectivity(g, sets=True)           # needs sage.numerical.mip
sage: len(setA) == 1 or len(setB) == 1                                           # needs sage.numerical.mip
True

A vertex cut in a tree is any internal vertex:

sage: tree = graphs.RandomTree(15)
sage: val, [cut_vertex] = vertex_connectivity(tree, value_only=False)            # needs sage.numerical.mip
sage: tree.degree(cut_vertex) > 1                                                # needs sage.numerical.mip
True

When value_only = True, this function is optimized for small connectivity values and does not need to build a linear program.

It is the case for connected graphs which are not connected:

sage: g = 2 * graphs.PetersenGraph()
sage: vertex_connectivity(g)                                                     # needs sage.numerical.mip
0

Or if they are just 1-connected:

sage: g = graphs.PathGraph(10)
sage: vertex_connectivity(g)                                                     # needs sage.numerical.mip
1

For directed graphs, the strong connectivity is tested through the dedicated function:

sage: g = digraphs.ButterflyGraph(3)
sage: vertex_connectivity(g)                                                     # needs sage.numerical.mip
0

A complete graph on \(10\) vertices is \(9\)-connected:

sage: g = graphs.CompleteGraph(10)
sage: vertex_connectivity(g)                                                     # needs sage.numerical.mip
9

A complete digraph on \(10\) vertices is \(9\)-connected:

sage: g = DiGraph(graphs.CompleteGraph(10))
sage: vertex_connectivity(g)                                                     # needs sage.numerical.mip
9

When parameter k is set, we only check for the existence of a vertex cut of order at least k:

sage: g = graphs.PappusGraph()
sage: vertex_connectivity(g, k=3)                                                # needs sage.numerical.mip
True
sage: vertex_connectivity(g, k=4)                                                # needs sage.numerical.mip
False
vertex_cut(s, t, value_only, vertices=True, solver=False, verbose=None, integrality_tolerance=0)#

Return a minimum vertex cut between non-adjacent vertices \(s\) and \(t\) represented by a list of vertices.

A vertex cut between two non-adjacent vertices is a set \(U\) of vertices of self such that the graph obtained by removing \(U\) from self is disconnected. For more information, see the Wikipedia article Cut_(graph_theory).

INPUT:

  • value_only – boolean (default: True); whether to return only the size of the minimum cut, or to also return the set \(U\) of vertices of the cut

  • vertices – boolean (default: False); whether to also return the two sets of vertices that are disconnected by the cut. Implies value_only set to False.

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

OUTPUT:

Real number or tuple, depending on the given arguments (examples are given below).

EXAMPLES:

A basic application in the Pappus graph:

sage: g = graphs.PappusGraph()
sage: g.vertex_cut(1, 16, value_only=True)                                   # needs sage.numerical.mip
3

In the bipartite complete graph \(K_{2,8}\), a cut between the two vertices in the size \(2\) part consists of the other \(8\) vertices:

sage: g = graphs.CompleteBipartiteGraph(2, 8)
sage: [value, vertices] = g.vertex_cut(0, 1, value_only=False)               # needs sage.numerical.mip
sage: print(value)                                                           # needs sage.numerical.mip
8
sage: vertices == list(range(2, 10))                                         # needs sage.numerical.mip
True

Clearly, in this case the two sides of the cut are singletons:

sage: [value, vertices, [set1, set2]] = g.vertex_cut(0, 1, vertices=True)    # needs sage.numerical.mip
sage: len(set1) == 1                                                         # needs sage.numerical.mip
True
sage: len(set2) == 1                                                         # needs sage.numerical.mip
True
vertex_disjoint_paths(s, t, solver, verbose=None, integrality_tolerance=0)#

Return a list of vertex-disjoint paths between two vertices.

The vertex version of Menger’s theorem asserts that the size of the minimum vertex cut between two vertices \(s\) and \(t\) (the minimum number of vertices whose removal disconnects \(s\) and \(t\)) is equal to the maximum number of pairwise vertex-independent paths from \(s\) to \(t\).

This function returns a list of such paths.

INPUT:

  • s,t – two vertices of the graph.

  • solver – string (default: None); specify a Mixed Integer Linear Programming (MILP) solver to be used. If set to None, the default one is used. For more information on MILP solvers and which default solver is used, see the method solve of the class MixedIntegerLinearProgram.

  • verbose – integer (default: 0); sets the level of verbosity. Set to 0 by default, which means quiet.

  • integrality_tolerance – float; parameter for use with MILP solvers over an inexact base ring; see MixedIntegerLinearProgram.get_values().

EXAMPLES:

In a complete bipartite graph

sage: g = graphs.CompleteBipartiteGraph(2, 3)
sage: g.vertex_disjoint_paths(0, 1)                                         # needs sage.numerical.mip
[[0, 2, 1], [0, 3, 1], [0, 4, 1]]
vertex_iterator(vertices=None, degree=None, vertex_property=None)#

Return an iterator over the given vertices.

Returns False if not given a vertex, sequence, iterator or None. None is equivalent to a list of every vertex. Note that for v in G syntax is allowed.

INPUT:

  • vertices – iterated vertices are these intersected with the vertices of the (di)graph

  • degree – a nonnegative integer (default: None); a vertex v is kept if degree(v) == degree

  • vertex_property – function (default: None); a function that inputs a vertex and outputs a boolean value, i.e., a vertex v is kept if vertex_property(v) == True

EXAMPLES:

sage: P = graphs.PetersenGraph()
sage: for v in P.vertex_iterator():
....:     print(v)
0
1
2
...
8
9
sage: G = graphs.TetrahedralGraph()
sage: for i in G:
....:     print(i)
0
1
2
3
sage: H = graphs.PathGraph(5)
sage: prop = lambda l: l % 3 == 1
sage: for v in H.vertex_iterator(degree=1, vertex_property=prop):
....:     print(v)
4

Note that since the intersection option is available, the vertex_iterator() function is sub-optimal, speed-wise, but note the following optimization:

sage: timeit V = P.vertices(sort=False)         # not tested
100000 loops, best of 3: 8.85 [micro]s per loop
sage: timeit V = list(P.vertex_iterator())      # not tested
100000 loops, best of 3: 5.74 [micro]s per loop
vertices(sort=False, key=None, degree=None, vertex_property=None)#

Return a list of the vertices.

INPUT:

  • sort – boolean (default: False); whether to sort vertices according the ordering specified with parameter key. If False (default), vertices are not sorted.

  • key – a function (default: None); a function that takes a vertex as its one argument and returns a value that can be used for comparisons in the sorting algorithm (we must have sort=True)

  • degree – a nonnegative integer (default: None); a vertex v is kept if degree(v) == degree

  • vertex_property – function (default: None); a function that inputs a vertex and outputs a boolean value, i.e., a vertex v is kept if vertex_property(v) == True

OUTPUT:

The list of vertices of the (di)graph.

Warning

Since any object may be a vertex, there is no guarantee that any two vertices will be comparable. With default objects for vertices (all integers), or when all the vertices are of the same simple type, then there should not be a problem with how the vertices will be sorted. However, if you need to guarantee a total order for the sorting of the edges, use the key argument, as illustrated in the examples below.

EXAMPLES:

sage: P = graphs.PetersenGraph()
sage: P.vertices(sort=True)
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9]

If you do not care about sorted output and you are concerned about the time taken to sort, consider the following alternative:

sage: # not tested
sage: timeit V = P.vertices(sort=True)
625 loops, best of 3: 3.86 [micro]s per loop
sage: timeit V = P.vertices(sort=False)
625 loops, best of 3: 2.06 [micro]s per loop
sage: timeit V = list(P.vertex_iterator())
625 loops, best of 3: 2.05 [micro]s per loop
sage: timeit('V = list(P)')
625 loops, best of 3: 1.98 [micro]s per loop

We illustrate various ways to use a key to sort the list:

sage: H = graphs.HanoiTowerGraph(3, 3, labels=False)
sage: H.vertices(sort=True)
[0, 1, 2, 3, 4, ... 22, 23, 24, 25, 26]
sage: H.vertices(sort=True, key=lambda x: -x)
[26, 25, 24, 23, 22, ... 4, 3, 2, 1, 0]
sage: G = graphs.HanoiTowerGraph(3, 3)
sage: G.vertices(sort=True)
[(0, 0, 0), (0, 0, 1), (0, 0, 2), (0, 1, 0), ... (2, 2, 1), (2, 2, 2)]
sage: G.vertices(sort=True, key = lambda x: (x[1], x[2], x[0]))
[(0, 0, 0), (1, 0, 0), (2, 0, 0), (0, 0, 1), ... (1, 2, 2), (2, 2, 2)]

The discriminant of a polynomial is a function that returns an integer. We build a graph whose vertices are polynomials, and use the discriminant function to provide an ordering. Note that since functions are first-class objects in Python, we can specify precisely the function from the Sage library that we wish to use as the key:

sage: # needs sage.libs.flint
sage: t = polygen(QQ, 't')
sage: K = Graph({5*t: [t^2], t^2: [t^2+2], t^2+2: [4*t^2-6], 4*t^2-6: [5*t]})
sage: from sage.rings.polynomial.polynomial_rational_flint import Polynomial_rational_flint
sage: dsc = Polynomial_rational_flint.discriminant
sage: verts = K.vertices(sort=True, key=dsc); verts
[t^2 + 2, t^2, 5*t, 4*t^2 - 6]
sage: [x.discriminant() for x in verts]
[-8, 0, 1, 96]
weighted(new=None)#

Whether the (di)graph is to be considered as a weighted (di)graph.

INPUT:

  • new – boolean (default: None); if it is provided, then the weightedness flag is set accordingly. This is not allowed for immutable graphs.

Note

Changing the weightedness flag changes the ==-class of a graph and is thus not allowed for immutable graphs.

Edge weightings can still exist for (di)graphs G where G.weighted() is False.

EXAMPLES:

Here we have two graphs with different labels, but weighted() is False for both, so we just check for the presence of edges:

sage: G = Graph({0: {1: 'a'}}, sparse=True)
sage: H = Graph({0: {1: 'b'}}, sparse=True)
sage: G == H
True

Now one is weighted and the other is not, and thus the graphs are not equal:

sage: G.weighted(True)
sage: H.weighted()
False
sage: G == H
False

However, if both are weighted, then we finally compare ‘a’ to ‘b’:

sage: H.weighted(True)
sage: G == H
False
weighted_adjacency_matrix(sparse, vertices=True, default_weight=None, base_ring=None, **kwds)#

Return the weighted adjacency matrix of the graph.

By default, each vertex is represented by its position in the list returned by method vertices().

INPUT:

  • sparse – boolean (default: True); whether to use a sparse or a dense matrix

  • vertices – list (default: None); when specified, each vertex is represented by its position in the list vertices, otherwise each vertex is represented by its position in the list returned by method vertices()

  • default_weight – (default: None); specifies the weight to replace any None edge label. When not specified an error is raised if the label of an edge is None.

  • base_ring – a ring (default: determined from the weights); the base ring of the matrix space to use.

  • **kwds – other keywords to pass to matrix()

EXAMPLES:

sage: G = Graph(sparse=True, weighted=True)
sage: G.add_edges([(0, 1, 1), (1, 2, 2), (0, 2, 3), (0, 3, 4)])
sage: M = G.weighted_adjacency_matrix(); M                                  # needs sage.modules
[0 1 3 4]
[1 0 2 0]
[3 2 0 0]
[4 0 0 0]
sage: H = Graph(data=M, format='weighted_adjacency_matrix', sparse=True)    # needs sage.modules
sage: H == G                                                                # needs sage.modules
True
sage: G.weighted_adjacency_matrix(vertices=[3, 2, 1, 0])                    # needs sage.modules
[0 0 0 4]
[0 0 2 3]
[0 2 0 1]
[4 3 1 0]

Using a different matrix implementation:

sage: M = G.weighted_adjacency_matrix(sparse=False, base_ring=ZZ,           # needs numpy sage.modules
....:                                 implementation='numpy'); M
[0 1 3 4]
[1 0 2 0]
[3 2 0 0]
[4 0 0 0]

As an immutable matrix:

sage: M = G.weighted_adjacency_matrix(immutable=True); M                    # needs sage.modules
[0 1 3 4]
[1 0 2 0]
[3 2 0 0]
[4 0 0 0]
sage: M[2, 2] = 1                                                           # needs sage.modules
Traceback (most recent call last):
...
ValueError: matrix is immutable; please change a copy instead
(i.e., use copy(M) to change a copy of M).
wiener_index(by_weight=False, algorithm=None, weight_function=None, check_weight=True)#

Return the Wiener index of self.

The graph is expected to have no cycles of negative weight.

The Wiener index of a undirected graph \(G\) is \(W(G) = \frac{1}{2} \sum_{u,v\in G} d(u,v)\) where \(d(u,v)\) denotes the distance between vertices \(u\) and \(v\) (see [KRG1996]).

The Wiener index of a directed graph \(G\) is defined as the sum of the distances between each pairs of vertices, i.e., \(W(G) = \sum_{u,v\in G} d(u,v)\).

For more information on the input variables and more examples, we refer to shortest_paths() and shortest_path_all_pairs(), which have very similar input variables.

INPUT:

  • by_weight – boolean (default: False); if True, the edges in the graph are weighted, otherwise all edges have weight 1

  • algorithm – string (default: None); one of the following algorithms:

    • For by_weight==False only:

      • 'BFS' - the computation is done through a BFS centered on each vertex successively.

      • 'Floyd-Warshall-Cython' - the Cython implementation of the Floyd-Warshall algorithm. Usually slower than 'BFS'.

    • For graphs without negative weights:

      • 'Dijkstra_Boost': the Dijkstra algorithm, implemented in Boost.

      • 'Dijkstra_NetworkX': the Dijkstra algorithm, implemented in NetworkX. Usually slower than 'Dijkstra_Boost'.

    • For graphs with negative weights:

      • 'Bellman-Ford_Boost': the Bellman-Ford algorithm, implemented in Boost.

      • 'Johnson_Boost': the Johnson algorithm, implemented in Boost.

      • 'Floyd-Warshall-Python' - the Python implementation of the Floyd-Warshall algorithm. Usually slower than 'Johnson_Boost'.

    • None (default): Sage chooses the best algorithm: 'BFS' for unweighted graphs, 'Dijkstra_Boost' if all weights are positive, 'Johnson_Boost', otherwise.

  • weight_function – function (default: None); a function that takes as input an edge (u, v, l) and outputs its weight. If not None, by_weight is automatically set to True. If None and by_weight is True, we use the edge label l as a weight, if l is not None, else 1 as a weight.

  • check_weight – boolean (default: True); if True, we check that the weight_function outputs a number for each edge

Note

Some algorithms (e.g., Boost algorithms) use floating point numbers for internal computations. Whenever the solution is integral, we try to convert the returned value to an integer.

EXAMPLES:

sage: G = Graph( { 0: {1: None}, 1: {2: None}, 2: {3: 1}, 3: {4: 2}, 4: {0: 2} }, sparse=True)
sage: G.wiener_index()
15
sage: G.wiener_index(weight_function=lambda e:(e[2] if e[2] is not None else 1))
20
sage: G.wiener_index(weight_function=lambda e:(e[2] if e[2] is not None else 200))
820
sage: G.wiener_index(algorithm='BFS')
15
sage: G.wiener_index(algorithm='Floyd-Warshall-Cython')
15
sage: G.wiener_index(algorithm='Floyd-Warshall-Python')
15
sage: G.wiener_index(algorithm='Dijkstra_Boost')
15
sage: G.wiener_index(algorithm='Bellman-Ford_Boost')
15
sage: G.wiener_index(algorithm='Johnson_Boost')
15
sage: G.wiener_index(algorithm='Dijkstra_NetworkX')                         # needs networkx
15

Wiener index of complete (di)graphs:

sage: n = 5
sage: g = graphs.CompleteGraph(n)
sage: g.wiener_index() == (n * (n - 1)) / 2
True
sage: g = digraphs.Complete(n)
sage: g.wiener_index() == n * (n - 1)
True

Wiener index of circuit digraphs:

sage: n = 7
sage: g = digraphs.Circuit(n)
sage: w = lambda x: (x*x*(x-1))/2
sage: g.wiener_index(algorithm='Dijkstra_Boost') == w(n)
True

Wiener index of a graph of order 1:

sage: Graph(1).wiener_index()
0

The Wiener index is not defined on the empty graph:

sage: Graph().wiener_index()
Traceback (most recent call last):
...
ValueError: Wiener index is not defined for the empty graph
sage.graphs.generic_graph.graph_isom_equivalent_non_edge_labeled_graph(g, partition=None, standard_label=None, return_relabeling=False, return_edge_labels=False, inplace=False, ignore_edge_labels=False)#

Helper function for canonical labeling of edge labeled (di)graphs.

Translates to a bipartite incidence-structure type graph appropriate for computing canonical labels of edge labeled and/or multi-edge graphs. Note that this is actually computationally equivalent to implementing a change on an inner loop of the main algorithm – namely making the refinement procedure sort for each label.

If the graph is a multigraph, it is translated to a non-multigraph, where each instance of multiple edges is converted to a single edge labeled with a list [[label1, multiplicity], [label2, multiplicity], ...] describing how many edges of each label were originally there. Then in either case we are working on a graph without multiple edges. At this point, we create another (partially bipartite) graph, whose left vertices are the original vertices of the graph, and whose right vertices represent the labeled edges. Any unlabeled edges in the original graph are also present in the new graph, and – this is the bipartite aspect – for every labeled edge \(e\) from \(v\) to \(w\) in the original graph, there is an edge between the right vertex corresponding to \(e\) and each of the left vertices corresponding to \(v\) and \(w\). We partition the left vertices as they were originally, and the right vertices by common labels: only automorphisms taking edges to like-labeled edges are allowed, and this additional partition information enforces this on the new graph.

INPUT:

  • g – Graph or DiGraph

  • partition – list (default: None); a partition of the vertices as a list of lists of vertices. If given, the partition of the vertices is as well relabeled

  • standard_label – (default: None); edges in g with this label are preserved in the new graph

  • return_relabeling – boolean (default: False); whether to return a dictionary containing the relabeling

  • return_edge_labels – boolean (default: False); whether the different edge_labels are returned (useful if inplace is True)

  • inplace – boolean (default: False); whether the input (di)graph g is modified or the return a new (di)graph. Note that attributes of g are not copied for speed issues, only edges and vertices.

  • ignore_edge_labels – boolean (default: False): if True, ignore edge labels, so when constructing the new graph, only multiple edges are replaced with vertices. Labels on multiple edges are ignored – only the multiplicity is relevant, so multiple edges with the same multiplicity in the original graph correspond to right vertices in the same partition in the new graph.

OUTPUT:

  • if inplace is False: the unlabeled graph without multiple edges

  • the partition of the vertices

  • if return_relabeling is True: a dictionary containing the relabeling

  • if return_edge_labels is True: the list of (former) edge labels is returned

EXAMPLES:

sage: from sage.graphs.generic_graph import graph_isom_equivalent_non_edge_labeled_graph

sage: G = Graph(multiedges=True,sparse=True)
sage: G.add_edges((0, 1, i) for i in range(10))
sage: G.add_edge(1,2,'string')
sage: G.add_edge(2,123)
sage: graph_isom_equivalent_non_edge_labeled_graph(G, partition=[[0,123],[1,2]])
[Graph on 6 vertices, [[1, 0], [2, 3], [5], [4]]]

sage: g, part = graph_isom_equivalent_non_edge_labeled_graph(G)
sage: g, sorted(part)
(Graph on 6 vertices, [[0, 1, 2, 3], [4], [5]])
sage: g.edges(sort=True)
[(0, 3, None), (1, 4, None), (2, 4, None), (2, 5, None), (3, 5, None)]

sage: g = graph_isom_equivalent_non_edge_labeled_graph(G, standard_label='string',
....:                                                  return_edge_labels=True)
sage: g[0]
Graph on 6 vertices
sage: g[0].edges(sort=True)
[(0, 5, None), (1, 4, None), (2, 3, None), (2, 4, None), (3, 5, None)]
sage: g[1]
[[0, 1, 2, 3], [4], [5]]
sage: g[2]
[[['string', 1]], [[0, 1], [1, 1], [2, 1], [3, 1], [4, 1], [5, 1], [6, 1], [7, 1], [8, 1], [9, 1]], [[None, 1]]]

sage: graph_isom_equivalent_non_edge_labeled_graph(G, inplace=True)
[[[0, 1, 2, 3], [5], [4]]]
sage: G.edges(sort=True)
[(0, 3, None), (1, 4, None), (2, 4, None), (2, 5, None), (3, 5, None)]

sage: G = Graph(multiedges=True,sparse=True)
sage: G.add_edges((0, 1) for i in range(10))
sage: G.add_edge(1, 2, 'a')
sage: G.add_edge(1, 3, 'b')
sage: G.add_edge(2, 3, 'b')
sage: graph_isom_equivalent_non_edge_labeled_graph(G)[0]
Graph on 8 vertices
sage: graph_isom_equivalent_non_edge_labeled_graph(G, ignore_edge_labels=True)[0]
Graph on 5 vertices

sage: G = Graph(multiedges=True,sparse=True)
sage: G.add_edges((0, 1, i) for i in range(5))
sage: G.add_edges((0, 2, i+10) for i in range(5))
sage: G.add_edges((0, 3) for i in range(4))
sage: g0 = graph_isom_equivalent_non_edge_labeled_graph(G)
sage: g1 = graph_isom_equivalent_non_edge_labeled_graph(G, ignore_edge_labels=True)
sage: g0
[Graph on 7 vertices, [[0, 1, 2, 3], [4], [5], [6]]]
sage: g1
[Graph on 7 vertices, [[0, 1, 2, 3], [6], [4, 5]]]
sage.graphs.generic_graph.tachyon_vertex_plot(g, bgcolor=(1, 1, 1), vertex_colors=None, vertex_size=0.06, pos3d=None, **kwds)#

Helper function for plotting graphs in 3d with Tachyon.

Returns a plot containing only the vertices, as well as the 3d position dictionary used for the plot.

INPUT:
  • pos3d – a 3D layout of the vertices

  • various rendering options

EXAMPLES:

sage: G = graphs.TetrahedralGraph()
sage: from sage.graphs.generic_graph import tachyon_vertex_plot
sage: T,p = tachyon_vertex_plot(G, pos3d=G.layout(dim=3))                       # needs sage.plot
sage: type(T)                                                                   # needs sage.plot
<class 'sage.plot.plot3d.tachyon.Tachyon'>
sage: type(p)                                                                   # needs sage.plot
<... 'dict'>