Id: arxiv/2607.04466 Url: https://arxiv.org/abs/2607.04466 Tags: math.ST, math.PR, stat.CO Score: 0 Title: On the mixing properties of some preconditioned multiproposal Markov Chain Monte Carlo algorithms Domain: math.ST Source: arxiv Authors: Giulia Carigi, Nathan E. Glatt-Holtz, Cecilia F. Mondaini, Guillermina Senn Reviews: 0 Subject: math.ST Category: research Comments: 0 Disputed: false Division: math Mentions: 0 Discussed: 0 Published: 2026-07-05 Page: https://delphi.science/r/arxiv/2607.04466 We study two recently discovered "dimension-free" Monte Carlo sampling algorithms, the multiproposal and multiple-try preconditioned Crank-Nicolson methods (mpCN and MTpCN). These methods were designed to address certain non-parametric (i.e. infinite-dimensional) sampling problems, defined relative to a Gaussian reference measure, by combining proposal and acceptance mechanisms that take non-trivial advantage of parallel computing architectures. We provide the first rigorous analysis of both algorithms, establishing exponential convergence to the target measure through the weak Harris…