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
Section 5.3 of Syed et al., 2021 mentions that it is more efficient, under certain conditions, to run k multiple copies of the PT algorithm rather than increasing N to use more chains.
I might have missed it, but I do not see in Pigeons a direct way to use multiple copies of the PT algorithm. I suppose, one could relatively easily run k copies using Threads.@threads or Distributed.pmap. However, if not implemented yet, it might be sensible to have this implemented inside Pigeons, so to use a consistent interface for the parallelism in all its forms.
The text was updated successfully, but these errors were encountered:
Hello,
Thank you for your message! At the moment, this functionality is not implemented in Pigeons. However, it has been on our to-do list and I think this is a good issue to keep open. What you suggested should work as a temporary workaround.
Section 5.3 of Syed et al., 2021 mentions that it is more efficient, under certain conditions, to run
k
multiple copies of the PT algorithm rather than increasingN
to use more chains.I might have missed it, but I do not see in Pigeons a direct way to use multiple copies of the PT algorithm. I suppose, one could relatively easily run
k
copies usingThreads.@threads
orDistributed.pmap
. However, if not implemented yet, it might be sensible to have this implemented inside Pigeons, so to use a consistent interface for the parallelism in all its forms.The text was updated successfully, but these errors were encountered: