Refactor of jitter_run/utils

joshspeagle / dynesty

Dynamic Nested Sampling package for computing Bayesian posteriors and evidences

https://dynesty.readthedocs.io/

MIT License

347 stars 76 forks source link

Refactor of jitter_run/utils #275

Closed segasai closed 3 years ago

segasai commented 3 years ago

I've started to take a stab at optimizing/understanding/clarifiying jitter_run.

I replaced all the numpy 'int' ,'bool' types by int, bool. I also separated the code that detects decrease in number of live-points. (I verified that it produces the same results).

I think I'd like to vectorize/clarify the jitter_run code, as currently it's a bit of a maze to me. It'd be also good to have some kind of references explaining the maths there...

coveralls commented 3 years ago

Pull Request Test Coverage Report for Build 885276029

29 of 38 (76.32%) changed or added relevant lines in 1 file are covered.
No unchanged relevant lines lost coverage.
Overall coverage increased (+0.07%) to 71.734%

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
py/dynesty/utils.py	29	38	76.32%
<!--	Total:	29	38	76.32%	-->

Totals
Change from base Build 883214612:	0.07%
Covered Lines:	3350
Relevant Lines:	4670

💛 - Coveralls

joshspeagle commented 3 years ago

As a quick reference if you'd like to pursue further vectorization, some of the math is outlined in Appendix A of the dynesty paper, in particular A.4 and A.5.

joshspeagle commented 3 years ago

Okay, this looks good to me. Pulling out the "check if live points are decreasing" part of the code into its own separate internal function is also a good change. Thanks.