Issues in the multiple try Metropolis mixing

DOI10.1007/S00180-016-0643-9MaRDI QIDQ2358918zbMATH OpenOpenAlexDBLPWikidataFDO

Authors Francisco Louzada, Luca Martino

Publication date 27 June 2017

Published in Computational Statistics (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1508.04253

zbMATH Keywords

MCMC methodsmulti-point Metropolis algorithmmultiple try Metropolis algorithmMTM with variable number of tries

Mathematics Subject Classification ID

Computational methods in Markov chains (60J22) Computational methods for problems pertaining to statistics (62-08) Bayesian inference (62F15) Monte Carlo methods (65C05) Numerical analysis or methods applied to Markov chains (65C40)

Abstract: The multiple Try Metropolis (MTM) algorithm is an advanced MCMC technique based on drawing and testing several candidates at each iteration of the algorithm. One of them is selected according to certain weights and then it is tested according to a suitable acceptance probability. Clearly, since the computational cost increases as the employed number of tries grows, one expects that the performance of an MTM scheme improves as the number of tries increases, as well. However, there are scenarios where the increase of number of tries does not produce a corresponding enhancement of the performance. In this work, we describe these scenarios and then we introduce possible solutions for solving these issues.

Recommendations

• On the flexibility of the design of multiple try Metropolis schemes
• An adaptive multiple-try Metropolis algorithm
• Adaptive Component-Wise Multiple-Try Metropolis Sampling
• A multiple-try Metropolis-Hastings algorithm with tailored proposals
• Multipoint Metropolis method with application to hybrid Monte Carlo
• Mixing of MCMC algorithms
• An Extension of the Metropolis Algorithm
• A general limitation on Monte Carlo algorithms of the Metropolis type

Cites work

• scientific article; zbMATH DE number 1082719 (Why is no real title available?)
• scientific article; zbMATH DE number 2117879 (Why is no real title available?)
• A multi-point Metropolis scheme with generic weight functions
• Adaptive multiple importance sampling
• Advanced Markov chain Monte Carlo methods. Learning from past samples.
• An Adaptive Population Importance Sampler: Learning From Uncertainty
• Equation of state calculations by fast computing machines
• Generalized multiple importance sampling
• Interacting multiple try algorithms with different proposal distributions
• Layered adaptive importance sampling
• Markov chain Monte Carlo methods with applications to signal processing.
• Monte Carlo sampling methods using Markov chains and their applications
• Multipoint Metropolis method with application to hybrid Monte Carlo
• On the flexibility of the design of multiple try Metropolis schemes
• Safe and Effective Importance Sampling
• Scaling analysis of multiple-try MCMC methods
• The Multiple-Try Method and Local Optimization in Metropolis Sampling

Cited in

(6)

• Plateau proposal distributions for adaptive component-wise multiple-try metropolis
• On the flexibility of the design of multiple try Metropolis schemes
• A multiple-try Metropolis-Hastings algorithm with tailored proposals
• Convergence rate of multiple-try Metropolis independent sampler
• Scaling analysis of multiple-try MCMC methods
• Interacting multiple try algorithms with different proposal distributions

This page was built for publication: Issues in the multiple try Metropolis mixing

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2358918)