search

joshspeagle / dynesty

Dynamic Nested Sampling package for computing Bayesian posteriors and evidences
https://dynesty.readthedocs.io/
MIT License
357 stars 77 forks source link

stopping criteria check #92

Closed joshspeagle closed 6 years ago

joshspeagle commented 6 years ago

I derived some basic results for the stopping criteria as part of a related project. I should confirm that the basic findings check out in dynesty and adjust the stopping criteria accordingly. This needs to be done before the initial package release since the stopping criteria is a huge selling point.

joshspeagle commented 6 years ago

This seems to be alright, so I'm closing this.

Note: I should include the comparison to MCMC methods in the paper. Also reference the discussion I had with @aimalz concerning what metrics we would ideally want.

aimalz commented 6 years ago

Indeed, let me know if you'd like a hand implementing or writing about the alternative convergence criterion we discussed.