Stan on ARC: Difference between revisions
Jump to navigation
Jump to search
m (Added categories) |
|||
| Line 4: | Line 4: | ||
= Installing Stan and rstan on ARC = | = Installing Stan and rstan on ARC = | ||
This requires building your own copy of R from source in your home directory. See the page [[R on ARC | R on ARC]] for more details. | This requires building your own copy of R from source in your home directory. See the page [[R on ARC | R on ARC]] for more details. | ||
== Troubleshooting == | |||
If you are trying to build RStan on ARC's login node and the build process fails with an error like this: | |||
<pre> | |||
g++: fatal error: Killed signal terminated program cc1plus | |||
compilation terminated. | |||
make: *** [/home/maryam.mahmoudighara/software/R-4.2.2/lib64/R/etc/Makeconf:178: grammar_inst.o] Error 1 | |||
ERROR: compilation failed for package ‘rstan’ | |||
</pre> | |||
This means that there is not enough RAM on the login node to build RStan. | |||
The login node limits all processes to 5 GB belonging to one user, due to its multi-user nature. | |||
You will have to request an interactive job with sufficient memory (RAM) to do this. | |||
If you request a job with 16 or 32 GB this should be sufficient to build RStan. | |||
You only have to do this once. | |||
= Submitting jobs that use rstan on ARC = | = Submitting jobs that use rstan on ARC = | ||
Revision as of 17:01, 8 September 2023
Intro to Stan
Installing Stan and rstan on ARC
This requires building your own copy of R from source in your home directory. See the page R on ARC for more details.
Troubleshooting
If you are trying to build RStan on ARC's login node and the build process fails with an error like this:
g++: fatal error: Killed signal terminated program cc1plus compilation terminated. make: *** [/home/maryam.mahmoudighara/software/R-4.2.2/lib64/R/etc/Makeconf:178: grammar_inst.o] Error 1 ERROR: compilation failed for package ‘rstan’
This means that there is not enough RAM on the login node to build RStan. The login node limits all processes to 5 GB belonging to one user, due to its multi-user nature.
You will have to request an interactive job with sufficient memory (RAM) to do this. If you request a job with 16 or 32 GB this should be sufficient to build RStan. You only have to do this once.