Sticky PDMP samplers for sparse and local inference problems

DOI10.1007/S11222-022-10180-5zbMATH Open1499.62011arXiv2103.08478OpenAlexW4310153751MaRDI QIDQ2104011FDOQ2104011

Authors: Joris Bierkens, Sebastiano Grazzi, Moritz Schauer, Frank van der Meulen 

Publication date: 9 December 2022

Published in: Statistics and Computing (Search for Journal in Brave)

Abstract: We construct a new class of efficient Monte Carlo methods based on continuous-time piecewise deterministic Markov processes (PDMPs) suitable for inference in high dimensional sparse models, i.e. models for which there is prior knowledge that many coordinates are likely to be exactly $0$. This is achieved with the fairly simple idea of endowing existing PDMP samplers with 'sticky' coordinate axes, coordinate planes etc. Upon hitting those subspaces, an event is triggered during which the process sticks to the subspace, this way spending some time in a sub-model. This results in non-reversible jumps between different (sub-)models. While we show that PDMP samplers in general can be made sticky, we mainly focus on the Zig-Zag sampler. Compared to the Gibbs sampler for variable selection, we heuristically derive favourable dependence of the Sticky Zig-Zag sampler on dimension and data size. The computational efficiency of the Sticky Zig-Zag sampler is further established through numerical experiments where both the sample size and the dimension of the parameter space are large.

Full work available at URL: https://arxiv.org/abs/2103.08478

Recommendations

• Automatic zig-zag sampling in practice
• Infinite dimensional piecewise deterministic Markov processes
• Coordinate sampler: a non-reversible Gibbs-like MCMC sampler
• The zig-zag process and super-efficient sampling for Bayesian analysis of big data
• Limit theorems for the zig-zag process

zbMATH Keywords

Monte Carlohigh-dimensional problemspiecewise deterministic Markov processBayesian variable selectionspike-and-slabbig-datanon-reversible jump

Mathematics Subject Classification ID

Computational methods for problems pertaining to statistics (62-08) Bayesian inference (62F15) Continuous-time Markov processes on general state spaces (60J25)

Cites Work

• MCMC using Hamiltonian dynamics
• Bayesian variable selection regression for genome-wide association studies and other large-scale problems
• Reversible jump Markov chain Monte Carlo computation and Bayesian model determination
• Sparsity and Smoothness Via the Fused Lasso
• Bayesian Inference for Logistic Models Using Pólya–Gamma Latent Variables
• Bayesian Variable Selection in Linear Regression
• Spike and slab variable selection: frequentist and Bayesian strategies
• Title not available (Why is that?)
• The zig-zag process and super-efficient sampling for Bayesian analysis of big data
• The Bouncy Particle Sampler: A Non-Reversible Rejection-Free Markov Chain Monte Carlo Method
• Title not available (Why is that?)
• Stability of Markovian processes II: continuous-time processes and sampled chains
• Title not available (Why is that?)
• Continuous time Markov processes. An introduction.
• MCMC methods for functions: modifying old algorithms to make them faster
• A piecewise deterministic Monte Carlo method for diffusion bridges
• Scalable Importance Tempering and Bayesian Variable Selection
• Peskun-Tierney ordering for Markovian Monte Carlo: beyond the reversible scenario
• Ergodicity of the zigzag process
• Adaptive random neighbourhood informed Markov chain Monte Carlo for high-dimensional Bayesian variable selection
• Bayesian estimation of sparse precision matrices in the presence of Gaussian measurement error

Cited In (2)

• Automatic zig-zag sampling in practice
• Reversible Jump PDMP Samplers for Variable Selection

Uses Software

• pi-MASS
• Boids
• ccpdmp

This page was built for publication: Sticky PDMP samplers for sparse and local inference problems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2104011)