The paper[1] defined a multivariate effective sample size (multiESS).
I hope that pymc3 can compute it.
[1] Vats, D., Flegal, J. M., & Jones, G. L.(2017) “Multivariate Output Analysis for Markov chain Monte Carlo”
PRs are welcome!
uh, I’m sorry. what is ‘PRs’ ?
Oh, it means “pull requests”. Contributors “request” to the lead developers to “pull” their proposed changes into the project repository.
Basically if there’s something you’d like inside PyMC3, and the lead devs don’t have enough time to handle it, they’d definitely welcome contributions from users who need things implemented and have gone ahead and implemented it themselves. I’ve done this for a few spots in PyMC3 - documentation, bug fixes for multinomial distribution, and adding a mode to the Weibull distribution.
oh, I see. Good job !
Therefore, we’d love to welcome contributions from you as well! Happy to help you ping the right people to show you around the code base.