Interfaces to R

This is the reference to the Sagemath R interface, usable from any Sage program.

The %r interface creating an R cell in the sage notebook is decribed in the Notebook manual.

The %R and %%R interface creating an R line or an R cell in the Jupyter notebook are briefly decribed at the end of this page. This documentation will be expanded and placed in the Jupyter notebook manual when this manual exists.

The following examples try to follow “An Introduction to R” which can be found at http://cran.r-project.org/doc/manuals/R-intro.html .

EXAMPLES:

Simple manipulations; numbers and vectors

The simplest data structure in R is the numeric vector which consists of an ordered collection of numbers. To create a vector named \(x\) using the R interface in Sage, you pass the R interpreter object a list or tuple of numbers:

sage: x = r([10.4,5.6,3.1,6.4,21.7]); x  # optional - rpy2
[1] 10.4  5.6  3.1  6.4 21.7

You can invert elements of a vector x in R by using the invert operator or by doing 1/x:

sage: ~x  # optional - rpy2
[1] 0.09615385 0.17857143 0.32258065 0.15625000 0.04608295
sage: 1/x  # optional - rpy2
[1] 0.09615385 0.17857143 0.32258065 0.15625000 0.04608295

The following assignment creates a vector \(y\) with 11 entries which consists of two copies of \(x\) with a 0 in between:

sage: y = r([x,0,x]); y  # optional - rpy2
[1] 10.4  5.6  3.1  6.4 21.7  0.0 10.4  5.6  3.1  6.4 21.7

Vector Arithmetic

The following command generates a new vector \(v\) of length 11 constructed by adding together (element by element) \(2x\) repeated 2.2 times, \(y\) repeated just once, and 1 repeated 11 times:

sage: v = 2*x+y+1; v  # optional - rpy2
[1] 32.2 17.8 10.3 20.2 66.1 21.8 22.6 12.8 16.9 50.8 43.5

One can compute the sum of the elements of an R vector in the following two ways:

sage: sum(x)  # optional - rpy2
[1] 47.2
sage: x.sum()  # optional - rpy2
[1] 47.2

One can calculate the sample variance of a list of numbers:

sage: ((x-x.mean())^2/(x.length()-1)).sum()  # optional - rpy2
[1] 53.853
sage: x.var()  # optional - rpy2
[1] 53.853

sage: x.sort()  # optional - rpy2
[1] 3.1  5.6  6.4 10.4 21.7
sage: x.min()  # optional - rpy2
[1] 3.1
sage: x.max()  # optional - rpy2
[1] 21.7
sage: x  # optional - rpy2
[1] 10.4  5.6  3.1  6.4 21.7

sage: r(-17).sqrt()  # optional - rpy2
[1] NaN
sage: r('-17+0i').sqrt()  # optional - rpy2
[1] 0+4.123106i

Generating an arithmetic sequence:

sage: r('1:10')  # optional - rpy2
[1] 1  2  3  4  5  6  7  8  9 10

Because from is a keyword in Python, it can’t be used as a keyword argument. Instead, from_ can be passed, and R will recognize it as the correct thing:

sage: r.seq(length=10, from_=-1, by=.2)  # optional - rpy2
[1] -1.0 -0.8 -0.6 -0.4 -0.2  0.0  0.2  0.4  0.6  0.8

sage: x = r([10.4,5.6,3.1,6.4,21.7])  # optional - rpy2
sage: x.rep(2)  # optional - rpy2
[1] 10.4  5.6  3.1  6.4 21.7 10.4  5.6  3.1  6.4 21.7
sage: x.rep(times=2)  # optional - rpy2
[1] 10.4  5.6  3.1  6.4 21.7 10.4  5.6  3.1  6.4 21.7
sage: x.rep(each=2)  # optional - rpy2
[1] 10.4 10.4  5.6  5.6  3.1  3.1  6.4  6.4 21.7 21.7

Missing Values:

sage: na = r('NA')  # optional - rpy2
sage: z = r([1,2,3,na])  # optional - rpy2
sage: z  # optional - rpy2
[1]  1  2  3 NA
sage: ind = r.is_na(z)  # optional - rpy2
sage: ind  # optional - rpy2
[1] FALSE FALSE FALSE  TRUE
sage: zero = r(0)  # optional - rpy2
sage: zero / zero  # optional - rpy2
[1] NaN
sage: inf = r('Inf')  # optional - rpy2
sage: inf-inf  # optional - rpy2
[1] NaN
sage: r.is_na(inf)  # optional - rpy2
[1] FALSE
sage: r.is_na(inf-inf)  # optional - rpy2
[1] TRUE
sage: r.is_na(zero/zero)  # optional - rpy2
[1] TRUE
sage: r.is_na(na)  # optional - rpy2
[1] TRUE
sage: r.is_nan(inf-inf)  # optional - rpy2
[1] TRUE
sage: r.is_nan(zero/zero)  # optional - rpy2
[1] TRUE
sage: r.is_nan(na)  # optional - rpy2
[1] FALSE

Character Vectors:

sage: labs = r.paste('c("X","Y")', '1:10', sep='""'); labs  # optional - rpy2
[1] "X1"  "Y2"  "X3"  "Y4"  "X5"  "Y6"  "X7"  "Y8"  "X9"  "Y10"

Index vectors; selecting and modifying subsets of a data set:

sage: na = r('NA')  # optional - rpy2
sage: x = r([10.4,5.6,3.1,6.4,21.7,na]); x  # optional - rpy2
[1] 10.4  5.6  3.1  6.4 21.7   NA
sage: x['!is.na(self)']  # optional - rpy2
[1] 10.4  5.6  3.1  6.4 21.7

sage: x = r([10.4,5.6,3.1,6.4,21.7,na]); x  # optional - rpy2
[1] 10.4  5.6  3.1  6.4 21.7   NA
sage: (x+1)['(!is.na(self)) & self>0']  # optional - rpy2
[1] 11.4  6.6  4.1  7.4 22.7
sage: x = r([10.4,-2,3.1,-0.5,21.7,na]); x  # optional - rpy2
[1] 10.4 -2.0  3.1 -0.5 21.7   NA
sage: (x+1)['(!is.na(self)) & self>0']  # optional - rpy2
[1] 11.4  4.1  0.5 22.7

Distributions:

sage: r.options(width="60")  # optional - rpy2
$width
[1] 80

sage: rr = r.dnorm(r.seq(-3,3,0.1))  # optional - rpy2
sage: rr  # optional - rpy2
 [1] 0.004431848 0.005952532 0.007915452 0.010420935
 [5] 0.013582969 0.017528300 0.022394530 0.028327038
 [9] 0.035474593 0.043983596 0.053990967 0.065615815
[13] 0.078950158 0.094049077 0.110920835 0.129517596
[17] 0.149727466 0.171368592 0.194186055 0.217852177
[21] 0.241970725 0.266085250 0.289691553 0.312253933
[25] 0.333224603 0.352065327 0.368270140 0.381387815
[29] 0.391042694 0.396952547 0.398942280 0.396952547
[33] 0.391042694 0.381387815 0.368270140 0.352065327
[37] 0.333224603 0.312253933 0.289691553 0.266085250
[41] 0.241970725 0.217852177 0.194186055 0.171368592
[45] 0.149727466 0.129517596 0.110920835 0.094049077
[49] 0.078950158 0.065615815 0.053990967 0.043983596
[53] 0.035474593 0.028327038 0.022394530 0.017528300
[57] 0.013582969 0.010420935 0.007915452 0.005952532
[61] 0.004431848

Convert R Data Structures to Python/Sage:

sage: rr = r.dnorm(r.seq(-3,3,0.1))  # optional - rpy2
sage: sum(rr._sage_())  # optional - rpy2
9.9772125168981...

Or you get a dictionary to be able to access all the information:

sage: rs = r.summary(r.c(1,4,3,4,3,2,5,1))  # optional - rpy2
sage: rs  # optional - rpy2
   Min. 1st Qu.  Median    Mean 3rd Qu.    Max.
  1.000   1.750   3.000   2.875   4.000   5.000
  sage: d = rs._sage_()  # optional - rpy2
  sage: d['DATA']  # optional - rpy2
  [1, 1.75, 3, 2.875, 4, 5]
  sage: d['_Names']  # optional - rpy2
  ['Min.', '1st Qu.', 'Median', 'Mean', '3rd Qu.', 'Max.']
  sage: d['_r_class']  # optional - rpy2
  ['summaryDefault', 'table']

It is also possible to access the plotting capabilities of R through Sage. For more information see the documentation of r.plot() or r.png().

THE JUPYTER NOTEBOOK INTERFACE (work in progress).

The %r interface described in the Sage notebook manual is not useful in the Jupyter notebook : it creates a inferior R interpreter which cannot be escaped.

The RPy2 library allows the creation of an R cell in the Jupyter notebook analogous to the %r escape in command line or %r cell in a Sage notebook.

The interface is loaded by a cell containing the sole code:

“%load_ext rpy2.ipython”

After execution of this code, the %R and %%R magics are available:

  • %R allows the execution of a single line of R code. Data exchange is

    possible via the -i and -o options. Do “%R?” in a standalone cell to get the documentation.

  • %%R allows the execution in R of the whole text of a cell, with

    similar options (do “%%R?” in a standalone cell for documentation).

A few important points must be noted:

  • The R interpreter launched by this interface IS (currently) DIFFERENT from the R interpreter used br other r… functions.

  • Data exchanged via the -i and -o options have a format DIFFERENT from the format used by the r… functions (RPy2 mostly uses arrays, and bugs the user to use the pandas Python package).

  • R graphics are (beautifully) displayed in output cells, but are not directly importable. You have to save them as .png, .pdf or .svg files and import them in Sage for further use.

In its current incarnation, this interface is mostly useful to statisticians needing Sage for a few symbolic computations but mostly using R for applied work.

AUTHORS:

  • Mike Hansen (2007-11-01)

  • William Stein (2008-04-19)

  • Harald Schilly (2008-03-20)

  • Mike Hansen (2008-04-19)

  • Emmanuel Charpentier (2015-12-12, RPy2 interface)

class sage.interfaces.r.HelpExpression

Bases: str

Used to improve printing of output of r.help.

class sage.interfaces.r.R(maxread=None, logfile=None, init_list_length=1024, seed=None)

Bases: sage.interfaces.tab_completion.ExtraTabCompletion, sage.interfaces.interface.Interface

An interface to the R interpreter.

R is a comprehensive collection of methods for statistics, modelling, bioinformatics, data analysis and much more. For more details, see http://www.r-project.org/about.html

Resources:

EXAMPLES:

sage: r.summary(r.c(1,2,3,111,2,3,2,3,2,5,4))  # optional - rpy2
Min. 1st Qu.  Median    Mean 3rd Qu.    Max.
1.00    2.00    3.00   12.55    3.50  111.00
available_packages()

Returns a list of all available R package names.

This list is not necessarily sorted.

OUTPUT: list of strings

Note

This requires an internet connection. The CRAN server is that is checked is defined at the top of sage/interfaces/r.py.

EXAMPLES:

sage: ap = r.available_packages()   # optional - internet  # optional - rpy2
sage: len(ap) > 20                  # optional - internet  # optional - rpy2
True
call(function_name, *args, **kwds)

This is an alias for function_call().

EXAMPLES:

sage: r.call('length', [1,2,3])  # optional - rpy2
[1] 3
chdir(dir)

Changes the working directory to dir

INPUT:

  • dir – the directory to change to.

EXAMPLES:

sage: import tempfile  # optional - rpy2
sage: tmpdir = tempfile.mkdtemp()  # optional - rpy2
sage: r.chdir(tmpdir)  # optional - rpy2

Check that tmpdir and r.getwd() refer to the same directory. We need to use realpath() in case $TMPDIR (by default /tmp) is a symbolic link (see trac ticket #10264).

sage: os.path.realpath(tmpdir) == sageobj(r.getwd())  # known bug (trac #9970)  # optional - rpy2
True
completions(s)

Return all commands names that complete the command starting with the string s. This is like typing s[Ctrl-T] in the R interpreter.

INPUT:

  • s – string

OUTPUT: list – a list of strings

EXAMPLES:

sage: dummy = r._tab_completion(use_disk_cache=False)    #clean doctest  # optional - rpy2
sage: 'testInheritedMethods' in r.completions('tes')  # optional - rpy2
True
console()

Runs the R console as a separate new R process.

EXAMPLES:

sage: r.console()                    # not tested  # optional - rpy2
    R version 2.6.1 (2007-11-26)
    Copyright (C) 2007 The R Foundation for Statistical Computing
    ISBN 3-900051-07-0
    ...
convert_r_list(l)

Converts an R list to a Python list.

EXAMPLES:

sage: s = 'c(".GlobalEnv", "package:stats", "package:graphics", "package:grDevices", \n"package:utils", "package:datasets", "package:methods", "Autoloads", \n"package:base")'  # optional - rpy2
sage: r.convert_r_list(s)  # optional - rpy2
['.GlobalEnv',
 'package:stats',
 'package:graphics',
 'package:grDevices',
 'package:utils',
 'package:datasets',
 'package:methods',
 'Autoloads',
 'package:base']
eval(code, *args, **kwds)

Evaluates a command inside the R interpreter and returns the output as a string.

EXAMPLES:

sage: r.eval('1+1')  # optional - rpy2
'[1] 2'
function_call(function, args=None, kwds=None)

Return the result of calling an R function, with given args and keyword args.

OUTPUT: RElement – an object in R

EXAMPLES:

sage: r.function_call('length', args=[ [1,2,3] ])  # optional - rpy2
[1] 3
get(var)

Returns the string representation of the variable var.

INPUT:

  • var – a string

OUTPUT: string

EXAMPLES:

sage: r.set('a', 2)  # optional - rpy2
sage: r.get('a')  # optional - rpy2
'[1] 2'
help(command)

Returns help string for a given command.

INPUT: - command – a string

OUTPUT: HelpExpression – a subclass of string whose __repr__ method is __str__, so it prints nicely

EXAMPLES:

sage: r.help('c')  # optional - rpy2
title
-----

Combine Values into a Vector or List

name
----

c
...
install_packages(package_name)

Install an R package into Sage’s R installation.

EXAMPLES:

sage: r.install_packages('aaMI')       # not tested  # optional - rpy2
...
R is free software and comes with ABSOLUTELY NO WARRANTY.
You are welcome to redistribute it under certain conditions.
Type 'license()' or 'licence()' for distribution details.
...
Please restart Sage in order to use 'aaMI'.
library(library_name)

Load the library library_name into the R interpreter.

This function raises an ImportError if the given library is not known.

INPUT:

  • library_name – string

EXAMPLES:

sage: r.library('grid')  # optional - rpy2
sage: 'grid' in r.eval('(.packages())')  # optional - rpy2
True
sage: r.library('foobar')  # optional - rpy2
Traceback (most recent call last):
...
ImportError: ...
na()

Returns the NA in R.

OUTPUT: RElement – an element of R

EXAMPLES:

sage: r.na()  # optional - rpy2
[1] NA
plot(*args, **kwds)

The R plot function. Type r.help(‘plot’) for much more extensive documentation about this function. See also below for a brief introduction to more plotting with R.

If one simply wants to view an R graphic, using this function is is sufficient (because it calls dev.off() to turn off the device).

However, if one wants to save the graphic to a specific file, it should be used as in the example below to write the output.

EXAMPLES:

This example saves a plot to the standard R output, usually a filename like Rplot001.png - from the command line, in the current directory, and in the cell directory in the notebook:

sage: d=r.setwd('"%s"'%SAGE_TMP)    # for doctesting only; ignore if you are trying this  # optional - rpy2
sage: r.plot("1:10")                # optional -- rgraphics  # optional - rpy2
null device
          1

To save to a specific file name, one should use png() to set the output device to that file. If this is done in the notebook, it must be done in the same cell as the plot itself:

sage: filename = tmp_filename() + '.png'  # optional - rpy2
sage: r.png(filename='"%s"'%filename) # Note the double quotes in single quotes!; optional -- rgraphics  # optional - rpy2
NULL
sage: x = r([1,2,3])  # optional - rpy2
sage: y = r([4,5,6])  # optional - rpy2
sage: r.plot(x,y)         # optional -- rgraphics  # optional - rpy2
null device
          1
sage: import os; os.unlink(filename) # For doctesting, we remove the file; optional -- rgraphics  # optional - rpy2

Please note that for more extensive use of R’s plotting capabilities (such as the lattices package), it is advisable to either use an interactive plotting device or to use the notebook. The following examples are not tested, because they differ depending on operating system:

sage: r.X11() # not tested - opens interactive device on systems with X11 support  # optional - rpy2
sage: r.quartz() # not tested - opens interactive device on OSX  # optional - rpy2
sage: r.hist("rnorm(100)") # not tested - makes a plot  # optional - rpy2
sage: r.library("lattice") # not tested - loads R lattice plotting package  # optional - rpy2
sage: r.histogram(x = "~ wt | cyl", data="mtcars") # not tested - makes a lattice plot  # optional - rpy2
sage: r.dev_off() # not tested, turns off the interactive viewer  # optional - rpy2

In the notebook, one can use r.png() to open the device, but would need to use the following since R lattice graphics do not automatically print away from the command line:

sage: filename = tmp_filename() + '.png' # Not needed in notebook, used for doctesting  # optional - rpy2
sage: r.png(filename='"%s"'%filename) # filename not needed in notebook, used for doctesting; optional -- rgraphics  # optional - rpy2
NULL
sage: r.library("lattice")  # optional - rpy2
sage: r("print(histogram(~wt | cyl, data=mtcars))") # plot should appear; optional -- rgraphics  # optional - rpy2
sage: import os; os.unlink(filename) # We remove the file for doctesting, not needed in notebook; optional -- rgraphics  # optional - rpy2
png(*args, **kwds)

Creates an R PNG device.

This should primarily be used to save an R graphic to a custom file. Note that when using this in the notebook, one must plot in the same cell that one creates the device. See r.plot() documentation for more information about plotting via R in Sage.

These examples won’t work on the many platforms where R still gets built without graphics support.

EXAMPLES:

sage: filename = tmp_filename() + '.png'  # optional - rpy2
sage: r.png(filename='"%s"'%filename)             # optional -- rgraphics  # optional - rpy2
NULL
sage: x = r([1,2,3])  # optional - rpy2
sage: y = r([4,5,6])  # optional - rpy2
sage: r.plot(x,y) # This saves to filename, but is not viewable from command line; optional -- rgraphics  # optional - rpy2
null device
          1
sage: import os; os.unlink(filename) # We remove the file for doctesting; optional -- rgraphics  # optional - rpy2

We want to make sure that we actually can view R graphics, which happens differently on different platforms:

sage: s = r.eval('capabilities("png")') # Should be on Linux and Solaris  # optional - rpy2
sage: t = r.eval('capabilities("aqua")') # Should be on all supported Mac versions  # optional - rpy2
sage: "TRUE" in s+t                      # optional -- rgraphics  # optional - rpy2
True
read(filename)

Read filename into the R interpreter by calling R’s source function on a read-only file connection.

EXAMPLES:

sage: filename = tmp_filename()  # optional - rpy2
sage: f = open(filename, 'w')  # optional - rpy2
sage: _ = f.write('a <- 2+2\n')  # optional - rpy2
sage: f.close()  # optional - rpy2
sage: r.read(filename)  # optional - rpy2
sage: r.get('a')  # optional - rpy2
'[1] 4'
require(library_name)

Load the library library_name into the R interpreter.

This function raises an ImportError if the given library is not known.

INPUT:

  • library_name – string

EXAMPLES:

sage: r.library('grid')  # optional - rpy2
sage: 'grid' in r.eval('(.packages())')  # optional - rpy2
True
sage: r.library('foobar')  # optional - rpy2
Traceback (most recent call last):
...
ImportError: ...
set(var, value)

Set the variable var in R to what the string value evaluates to in R.

INPUT:

  • var – a string

  • value – a string

EXAMPLES:

sage: r.set('a', '2 + 3')  # optional - rpy2
sage: r.get('a')  # optional - rpy2
'[1] 5'
set_seed(seed=None)

Set the seed for R interpreter.

The seed should be an integer.

EXAMPLES:

sage: r = R()  # optional - rpy2
sage: r.set_seed(1)  # optional - rpy2
1
sage: r.sample("1:10", 5) # random  # optional - rpy2
[1] 3 4 5 7 2
source(s)

Display the R source (if possible) about the function named s.

INPUT:

  • s – a string representing the function whose source code you want to see

OUTPUT: string – source code

EXAMPLES:

sage: print(r.source("c"))  # optional - rpy2
function (...)  .Primitive("c")
version()

Return the version of R currently running.

OUTPUT: tuple of ints; string

EXAMPLES:

sage: r.version() # not tested  # optional - rpy2
((3, 0, 1), 'R version 3.0.1 (2013-05-16)')
sage: rint, rstr = r.version()  # optional - rpy2
sage: rint[0] >= 3  # optional - rpy2
True
sage: rstr.startswith('R version')  # optional - rpy2
True
class sage.interfaces.r.RElement(parent, value, is_name=False, name=None)

Bases: sage.interfaces.tab_completion.ExtraTabCompletion, sage.interfaces.interface.InterfaceElement

dot_product(other)

Implements the notation self . other.

INPUT:

  • self, other – R elements

OUTPUT: R element

EXAMPLES:

sage: c = r.c(1,2,3,4)  # optional - rpy2
sage: c.dot_product(c.t())  # optional - rpy2
     [,1] [,2] [,3] [,4]
[1,]    1    2    3    4
[2,]    2    4    6    8
[3,]    3    6    9   12
[4,]    4    8   12   16

sage: v = r([3,-1,8])  # optional - rpy2
sage: v.dot_product(v)  # optional - rpy2
     [,1]
[1,]   74
is_string()

Tell whether this element is a string.

EXAMPLES:

sage: r('"abc"').is_string()  # optional - rpy2
True
sage: r([1,2,3]).is_string()  # optional - rpy2
False
stat_model(x)

The tilde regression operator in R.

EXAMPLES:

sage: x = r([1,2,3,4,5])  # optional - rpy2
sage: y = r([3,5,7,9,11])  # optional - rpy2
sage: a = r.lm( y.tilde(x) ) # lm( y ~ x )  # optional - rpy2
sage: d = a._sage_()  # optional - rpy2
sage: d['DATA']['coefficients']['DATA'][1]  # optional - rpy2
2
tilde(x)

The tilde regression operator in R.

EXAMPLES:

sage: x = r([1,2,3,4,5])  # optional - rpy2
sage: y = r([3,5,7,9,11])  # optional - rpy2
sage: a = r.lm( y.tilde(x) ) # lm( y ~ x )  # optional - rpy2
sage: d = a._sage_()  # optional - rpy2
sage: d['DATA']['coefficients']['DATA'][1]  # optional - rpy2
2
class sage.interfaces.r.RFunction(parent, name, r_name=None)

Bases: sage.interfaces.interface.InterfaceFunction

A Function in the R interface.

INPUT:

  • parent – the R interface

  • name – the name of the function for Python

  • r_name – the name of the function in R itself (which can have dots in it)

EXAMPLES:

sage: length = r.length  # optional - rpy2
sage: type(length)  # optional - rpy2
<class 'sage.interfaces.r.RFunction'>
sage: loads(dumps(length))  # optional - rpy2
length
class sage.interfaces.r.RFunctionElement(obj, name)

Bases: sage.interfaces.interface.InterfaceFunctionElement

sage.interfaces.r.is_RElement(x)

Return True if x is an element in an R interface.

INPUT:

  • x – object

OUTPUT: bool

EXAMPLES:

sage: from sage.interfaces.r import is_RElement  # optional - rpy2
sage: is_RElement(2)  # optional - rpy2
False
sage: is_RElement(r(2))  # optional - rpy2
True
sage.interfaces.r.r_console()

Spawn a new R command-line session.

EXAMPLES:

sage: r.console()                    # not tested  # optional - rpy2
    R version 2.6.1 (2007-11-26)
    Copyright (C) 2007 The R Foundation for Statistical Computing
    ISBN 3-900051-07-0
    ...
sage.interfaces.r.r_version()

Return the R version.

EXAMPLES:

sage: r_version() # not tested  # optional - rpy2
((3, 0, 1), 'R version 3.0.1 (2013-05-16)')
sage: rint, rstr = r_version()  # optional - rpy2
sage: rint[0] >= 3  # optional - rpy2
True
sage: rstr.startswith('R version')  # optional - rpy2
True
sage.interfaces.r.reduce_load_R()

Used for reconstructing a copy of the R interpreter from a pickle.

EXAMPLES:

sage: from sage.interfaces.r import reduce_load_R  # optional - rpy2
sage: reduce_load_R()  # optional - rpy2
R Interpreter