R

R[1] is a language and environment for statistical computing and graphics. It is also known as GNU S.

Installed Versions

Default R is Microsoft R Open (MRO)[2] 4.0.2. No modulefile needed.
4.1.3 via modulefile R/4.1.3 This version has been built with Intel MKL (Math Kernel Library)[3] To use:

module load R/4.1.3

4.2.2 via modulefile R/4.2.2 This version has been built with Intel MKL (Math Kernel Library)[4] To use:

module load R/4.2.2

Interactive R

R Studio is not available on Picotte. As an alternative, JupyterLab can be run with an R kernel.

See: R in Jupyter

Third-Party R Packages

NOTE if you have packages from another computer, whether Windows, macOS, or Linux, the packages you installed there cannot be copied and used on Picotte or Proteus. R packages are often compiled, and must be compiled (and linked) on the platform which will be running those packages.

You may install third-party R packages privately into your home directory. From the R prompt (the "repos" and "Ncpus" options are optional):

[juser@picotte001 ~]$ module load R
[juser@picotte001 ~]$ R

R version 4.1.3 (2022-03-10) -- "One Push-Up"
Copyright (C) 2022 The R Foundation for Statistical Computing
Platform: x86_64-pc-linux-gnu (64-bit)
...

> install.packages("package_name", Ncpus=8)
Warning in install.packages("package_name", Ncpus = 8) :
  'lib = "/ifs/opt/R/4.1.3/lib64/R/library"' is not writable
Would you like to use a personal library instead? (yes/No/cancel) yes
Would you like to create a personal library
‘~/R/x86_64-pc-linux-gnu-library/4.1’
to install packages into? (yes/No/cancel) yes
... [compilation output] ...
* DONE (package_name)

The download source packages are in
     '/local/scratch/somethinghere/downloaded_packages'
>

DO NOT use the same library installation location for different versions of R. Just accept the defaults.

The default installation location should be: ~/R/x86_64-pc-linux-gnu-library/$R_VERSION

Configure a default repository

To avoid being asked for a CRAN repository location every time, add the following to ~/.Rprofile:

## Default repo - cloud.r-project.org will get a server near you
local({r <- getOption("repos")
       r["CRAN"] <- "https://cloud.r-project.org"
       options(repos=r)
})

Useful Packages

A couple of useful utility R packages:

minqa

The minqa package has a small bug which prevents it from building successfully on Picotte. A fixed version is available here: https://github.com/prehensilecode/minqa It only modifies the options for compiling the package, and makes no changes to the code which does the computations.

To install, download the .tar.gz file. Then, in the command line:

[juser@picotte001 ~]$ R CMD INSTALL minqa-x.y.z.tar.gz

More Information

Please see R documentation on install.packages[5]

doParallel

It is NOT RECOMMENDED to use doParallel. Instead, use a job array to launch multiple computations on multiple nodes with a single job script. See:

Picotte: Writing Slurm Job Scripts#Job Arrays

Example Job

See: R vs. Stata benchmark

Example R script -- name this file testjob.R

### Name this file: testjob.R
print(sample(1:3))
print(sample(1:3, size=3, replace=FALSE))  # same as previous line
print(sample(c(2,5,3), size=4, replace=TRUE))
print(sample(1:2, size=10, prob=c(1,3), replace=TRUE))

Example job script -- name this file "test_r.sh"

#!/bin/bash
#SBATCH -p def
#SBATCH -n 1
#SBATCH --cpus-per-task=8
#SBATCH --mem=64G
#SBATCH --time=2:00:00

### THIS IS FOR PICOTTE

# To use R 4.1.3, uncomment the following line
# module load R/4.1.3

R CMD BATCH testjob.R