You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I'm running some BAO chains using emcee with multiple MPI processes. Some parameters are analytically marginalized over, and the logger is printing this line for every MPI process
Hi, this is because when performing sampling, likelihood evaluations are distributed over MPI processes. Every process get to execute that differentiation part independently on a COMM_SELF (of size 1, so mpicomm.rank is always 0). Would you prefer this not to happen, or were you just wondering why this happened?
I'm running some BAO chains using emcee with multiple MPI processes. Some parameters are analytically marginalized over, and the logger is printing this line for every MPI process
[000008.12] [ 24/128] 08-18 10:55 Differentiation INFO Varied parameters: ['al0_-3', 'al0_-2', 'al0_-1', 'al0_0', 'al0_1', 'al2_-3', 'al2_-2', 'al2_-1', 'al2_0', 'al2_1']
I'm guessing the mesage is coming from
desilike/desilike/differentiation.py
Line 234 in e42e89c
which is supposed to only print it for the root process. Not sure why the logger is bypassing this conditional...
The text was updated successfully, but these errors were encountered: