High performance R
I have tried the solution using Intel MKL, and for me the speedup was negligible. This was for a script that does Markov-chain Monte Carlo studies of traits related to mating on a moose population. The libraries plyr, MuMIn, gdata, parallel and MCMCglmm were used. Does anyone have experience with significant speedupt using MKL? Perhaps in applications using lots of linear algebra? --AsmundEr (talk) 08:22, 25 August 2014 (UTC)
I think the section on optimized builds should be dropped from the wiki, or moved to a less proninent location. openblas commonly causes problems with R HPC packages for parallel computations, and the r-mkl AUR package has been orphaned due to licencing issues. In short getting an optimized blas working with R is a fairly advanced and error-prone process, and I don't think it should be the second thing people see when they look up R on the Archlinux wiki. Thoughts? Izahn (talk) 14:55, 18 December 2014 (UTC)
Since there was no objection here I've moved the "Optimized package" section to the end. Izahn
Also there seems to be some evidence that MKL does not severely outperform OpenBLAS: Benchmarking Single- and Multi-Core BLAS --Matmo (talk) 20:13, 22 January 2016 (UTC)
Revolution R Open
Now available in the AUR: revolution-r-open. Unikum (talk) 21:41, 21 August 2015 (UTC)
R CMD check problems with LaTeX
When creating packages you need to run the command R CMD check 'package-name' to check for errors. Using the arch R install, the process fails with a message
LaTeX errors when creating PDF version. This typically indicates Rd problems. LaTeX errors found: ! LaTeX Error: File `Rd.sty' not found.
The problem seems to be that the arch R installation puts Rd.sty in
whereas R CMD check is looking for it in
It may be possible to set a switch to tell R where it is, but I just created a /usr/share/R/texmf and copied it over, including
/usr/share/R/texmf/bibtex /usr/share/R/texmf/bibtex/bib /usr/share/R/texmf/bibtex/bib/RJournal.bib /usr/share/R/texmf/bibtex/bib/Rnews.bib /usr/share/R/texmf/bibtex/bst /usr/share/R/texmf/bibtex/bst/jss.bst /usr/share/R/texmf/tex /usr/share/R/texmf/tex/latex /usr/share/R/texmf/tex/latex/omscmtt.fd /usr/share/R/texmf/tex/latex/omsaett.fd /usr/share/R/texmf/tex/latex/Rd.sty /usr/share/R/texmf/tex/latex/jss.cls /usr/share/R/texmf/tex/latex/omsaer.fd /usr/share/R/texmf/tex/latex/ts1aett.fd /usr/share/R/texmf/tex/latex/ts1aer.fd /usr/share/R/texmf/tex/latex/Sweave.sty
just to cover all bets with respect to creating vignettes, etc.
I installed R on a new system and raised errors when trying "install.packages("hello_world")" both on console and in rstudio. The solution was to install the package tk. Maybe this dependency should be named some more directly (maybe next to gcc-fortran)? Kirjava (talk) 19:49, 22 December 2016 (UTC)