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 noticed that various discussion threads have requested the ability to continue a MCMC run. By continue, I mean that after calling sample(), one could call sample again and append the results to generate a longer chain.
I realize that there is the "interactive" sampler, but to me it would be nice for all sampling to have a hot restart / append capabilities.
Feel free to close this if this has already been discussed and deemed undesirable.
The text was updated successfully, but these errors were encountered:
You cannot resume a chain with sample(), but you can sample additional chains that will be started where the last chain left off when using sample() serially, then simply concatenating the chains together. PyMC3 will run multiple chains in parallel by default, which can then be treated as a single sample if desired for inference.
I will add the enhancement tag to this issue, so that restarting is universal for PyMC3.
I noticed that various discussion threads have requested the ability to continue a MCMC run. By continue, I mean that after calling sample(), one could call sample again and append the results to generate a longer chain.
I realize that there is the "interactive" sampler, but to me it would be nice for all sampling to have a hot restart / append capabilities.
Feel free to close this if this has already been discussed and deemed undesirable.
The text was updated successfully, but these errors were encountered: