A few years ago, I was at a Bayesian conference when one of the speakers said, somewhat apologetically, that they had started work on their project two years earlier when “everybody was into Stan, but of course now everybody uses Bouncy Particle Sampler”. I laughed at the absurd fashion-chasing, but…

Nice, thanks for pointing me to that. I especially appreciated contributions there from @maxbiostat and @spinkney . I was less interested in defending Stan the platform (that seems straightforward) as HMC/NUTS.

TL;DR: Dynamic HMC is robust enough and has good diagnostics. You might find an algorithm that outperforms it, but the implementation overhead better be worth it. [image] robertgrant: What are your arguments in favour of HMC? I honestly find this a bit weird. The way I see it, algorithms are…

Good point about diagnostics. Anyone writing their own code for newer sampling algorithms will have to pick over outputs forensically.

I feel confident saying that no-one ever really used the bouncy particle sampler when it came out, and definitely no-one uses it today. It was popular in the more theoretical computational statistics literature because it is more amenable to many of mathematical analyses techniques that are popular…

Thanks Mike, great points there that we can all use when we encounter clients/students/whoever with stars in their eyes about the latest fashionable algorithm. Personally, I think that getting people to simulate some nasty density like high-dimensional banana or multimodal mixtures, then feed that …

Arguments for HMC

General

yizhang January 13, 2021, 5:55pm 2

I believe there have been several discussions on this topic. IIRC This is the most recent one.

Topic		Replies	Views
Stan algorithm General	4	771	October 2, 2017
Tried to Speed Up Stan (HMC) by Rewriting Dynamics, failed Algorithms	2	726	June 17, 2020
Education-oriented, functional, interpreted Stan/HMC implementation Project Proposals education	19	2174	February 14, 2022
Hamiltonian, WHY? Algorithms	5	1231	May 27, 2019
PhD position (fully funded) in Bayesian Statistics, HMC and probabilistic programming Jobs	0	780	April 15, 2023

Arguments for HMC

Related topics