pints Set temperature schedule on population MCMC to be uniform on log scale

Set temperature schedule on population MCMC to be uniform on log scale

Open ben18785 opened this issue 6 years ago • 5 comments

As per SMC (and per Radford Neal's AIS article on p10: ftp://www.cs.toronto.edu/pub/radford/ais-rev.pdf)

Jul 24 '18 15:07 ben18785

PR in #537

Nov 13 '18 14:11 MichaelClerx

Another good one to pick up, has a PR too @ben18785

Sep 24 '19 15:09 MichaelClerx

Are we still interested in this one too, @ben18785 ?

Jan 14 '21 21:01 MichaelClerx

We could be. I've actually just closed the old PR where we had this. It's one of those ones where we probably want to test it out on an example (say, the eggbox) and see which of log vs linear scale works best.

Feb 10 '21 01:02 ben18785

If we do this, we'll also need to think about hyperparameters

https://github.com/pints-team/pints/issues/1327

Aug 11 '21 01:08 MichaelClerx

pints pints copied to clipboard

Set temperature schedule on population MCMC to be uniform on log scale

pints
pints copied to clipboard