Hello all. I was wondering if there was any parallelism built in for MLE estimation in a Stan model? Does reduce_sum assist in the optimization for the MLE?
Any information here would be most welcome!
1 Like
Reduce_sum speeds up the log lok. evaluation…no matter what type of estimation…so it’s also speeding up mle estimation
3 Likes
So this only works if there are more processes to compute the reduce_sum chunks.
How can we specify to use more processes when calling optimize?
What interface are you using?
cmdstanpy. I saw the thread for implementing it using cmdstanr.