I have performed some Stan modeling using Windows and Rstan. The code could work before and but recently I found some error like:
Error: cannot allocate vector of size 944 Kb
It seems that it’s memory allocatin error and the code has not changed while the data size has increased a little bit (like from 25000 observations changing to 27000 observations). My computer have 32 GB memory but specifying
memory.limit(size = 30000)
will not work. Not sure what’s going on and any advice?
Does this occur while the model is sampling, or while you are doing post-sampling operations on the fitted rstan object?
This occured during the model is sampling and thus there is no rstan object outputed.
Can pull up Task manager and watch the RAM usage of the model exe while it is sampling?
It seems to be a good idea but the model takes hours to run, any idea to automatically monitor the RAM usage?
Maybe try perfmon? I don’t think that article shows it, but you can make it process-specific.