Exact MCMC via Bernoulli factories for proposals with intractable normalizing constants

michaelchchoi · x · 2026-09-22

A thread curating MCMC/Monte Carlo papers from the JJSD special issue "Recent Advances in Dynamical Monte Carlo Methods" (guest edited by Kamatani at the Institute of Statistical Mathematics, noted via Christian Robert's blog).

First up: Kakkad & Vats' "Exact MCMC for intractable proposals". When a proposal distribution has an unknown normalizing function, the Metropolis–Hastings acceptance ratio cannot be evaluated exactly; prior work resorts to approximations or bespoke algorithms. The authors show Bernoulli factory MCMC — originally designed for doubly intractable targets — adapts naturally to yield exact sampling, demonstrated on three diverse examples.

Related event: Special Issue: Bernoulli Factory Exact MCMC and Ensemble Kalman Filter Stability(2 posts)→

Original post →

More from Research

Research channel →