Statistics > Computation
[Submitted on 11 Sep 2019 (this version), latest version 15 Mar 2021 (v6)]
Title:Exploring Discrete Analogs of Hamiltonian Monte Carlo
View PDFAbstract:Hamiltonian Monte Carlo (HMC) has emerged as a powerful way to sample from continuous distributions. However, a fundamental limitation of HMC is that it can't be applied to discrete distributions. In this paper, we propose the momentum sampler as a general framework for exploring the idea of building discrete analogs of HMC. The momentum sampler is a novel family of Markov Chain Monte Carlo (MCMC) algorithms that extends arbitrary single-site Metropolis Hastings (MH) samplers for discrete distributions with an auxiliary momentum process. It connects and generalizes a variety of existing methods, and points to a more general framework where we can mix stochastic evolution of discrete variables with deterministic evolution of continuous variables. The auxiliary momentum process can be integrated exactly, resulting in a rejection-free sampling process. The various trade-offs involved in picking different proposal distributions and kinetic energies have clear physical interpretations, and are demonstrated by numerical experiments on a simple Potts model.
Submission history
From: Guangyao Zhou [view email][v1] Wed, 11 Sep 2019 05:03:08 UTC (535 KB)
[v2] Fri, 21 Feb 2020 05:06:05 UTC (3,584 KB)
[v3] Mon, 15 Jun 2020 23:48:24 UTC (3,267 KB)
[v4] Wed, 21 Oct 2020 06:33:59 UTC (3,534 KB)
[v5] Thu, 18 Feb 2021 05:45:02 UTC (3,317 KB)
[v6] Mon, 15 Mar 2021 04:17:42 UTC (5,228 KB)
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.