Pseudo-Marginal Inference for CTMCs on Infinite Spaces via Monotonic Likelihood Approximations

DOI10.1080/10618600.2022.2118750arXiv2105.13566OpenAlexW3166610467WikidataQ114099318 ScholiaQ114099318MaRDI QIDQ6094080FDOQ6094080

Authors: Alexandre Bouchard-Côté, Trevor Campbell 

Publication date: 9 October 2023

Published in: Journal of Computational and Graphical Statistics (Search for Journal in Brave)

Abstract: Bayesian inference for Continuous-Time Markov Chains (CTMCs) on countably infinite spaces is notoriously difficult because evaluating the likelihood exactly is intractable. One way to address this challenge is to first build a non-negative and unbiased estimate of the likelihood -- involving the matrix exponential of finite truncations of the true rate matrix -- and then to use the estimates in a pseudo-marginal inference method. In this work, we show that we can dramatically increase the efficiency of this approach by avoiding the computation of exact matrix exponentials. In particular, we develop a general methodology for constructing an unbiased, non-negative estimate of the likelihood using doubly-monotone matrix exponential approximations. We further develop a novel approximation in this family -- the skeletoid -- as well as theory regarding its approximation error and how that relates to the variance of the estimates used in pseudo-marginal inference. Experimental results show that our approach yields more efficient posterior inference for a wide variety of CTMCs.

Full work available at URL: https://arxiv.org/abs/2105.13566

zbMATH Keywords

Markov chain Monte Carlocontinuous-time Markov chainpseudo-marginal methodsstochastic processesmatrix exponentialreaction networks

Mathematics Subject Classification ID

Statistics (62-XX)

Cites Work

• Title not available (Why is that?)
• Efficient parameter estimation for DNA kinetics modeled as continuous-time Markov chains
• The pseudo-marginal approach for efficient Monte Carlo computations
• Auxiliary Variable Methods for Markov Chain Monte Carlo with Applications
• Nineteen Dubious Ways to Compute the Exponential of a Matrix, Twenty-Five Years Later
• Concentration inequalities. A nonasymptotic theory of independence
• Exact and Computationally Efficient Likelihood-Based Estimation for Discretely Observed Diffusion Processes (with Discussion)
• Markov chain Monte Carlo for exact inference for diffusions
• Closed-form likelihood expansions for multivariate diffusions
• Continuous-time Markov chains. An applications-oriented approach
• An Exact Gibbs Sampler for the Markov-Modulated Poisson Process
• Bayesian analysis of elapsed times in continuous‐time Markov chains
• Title not available (Why is that?)
• Computing the action of the matrix exponential, with an application to exponential integrators
• Maximum-likelihood estimation for diffusion processes via closed-form density expansions
• On Russian roulette estimates for Bayesian inference with doubly-intractable likelihoods
• A robust discrete state approximation to the optimal nonlinear filter for a diffusiont
• Unbiased estimation with square root convergence for SDE models
• A general method for debiasing a Monte Carlo estimator
• Markov models for accumulating mutations
• The linear birth-death process: an inferential retrospective
• Stochastic simulation of the kinetics of multiple interacting nucleic acid strands
• On nonnegative unbiased estimators
• Exact Inference for Continuous Time Markov Chain Models
• Continuous-time approximations for the nonlinear filtering problem
• Unbiased Bayesian inference for population Markov jump processes via random truncations
• Large-sample asymptotics of the pseudo-marginal method
• Algorithm 955: Approximation of the inverse Poisson cumulative distribution function
• Direct statistical inference for finite Markov jump processes via the matrix exponential
• Efficient Parameter Sampling for Markov Jump Processes

This page was built for publication: Pseudo-Marginal Inference for CTMCs on Infinite Spaces via Monotonic Likelihood Approximations

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6094080)