Open zfurman56 opened 1 year ago
Bisected and traced the issue to 1a43e5ea391139a52b634f68432e76f88cb2b1c5, which seems to have introduced a typo while refactoring. Submitted a PR to fix.
Hi @zfurman56, sorry for never replying to this I was on holiday at the time and then forgot to get back to it 😓
Thanks very much for this PR I'll have a look this week!
Just merged your PR into master; thanks very much for this fix!
Using the code provided in this example notebook for either BAOAB or BADODAB gives samples that are constant, even with large dt. I haven't been able to get them to give non-constant output. Is this a bug, or is that notebook just out-of-date?
Output from BADODAB:
Output from SGNHT, for comparison:
Minimal reproducing code, available in this Colab notebook:
Replacing
build_badodab_sampler
withbuild_sgnht_sampler
(and changing dt to 1e-5) works as expected.