English

Partition function approach to non-Gaussian likelihoods: macrocanonical partitions and replicating Markov-chains

Cosmology and Nongalactic Astrophysics 2024-10-31 v3 Instrumentation and Methods for Astrophysics Data Analysis, Statistics and Probability

Abstract

Monte-Carlo techniques are standard numerical tools for exploring non-Gaussian and multivariate likelihoods. Many variants of the original Metropolis-Hastings algorithm have been proposed to increase the sampling efficiency. Motivated by Ensemble Monte Carlo we allow the number of Markov chains to vary by exchanging particles with a reservoir, controlled by a parameter analogous to a chemical potential μ\mu, which effectively establishes a random process that samples microstates from a macrocanonical instead of a canonical ensemble. In this paper, we develop the theory of macrocanonical sampling for statistical inference on the basis of Bayesian macrocanonical partition functions, thereby bringing to light the relations between information-theoretical quantities and thermodynamic properties. Furthermore, we propose an algorithm for macrocanonical sampling, Avalanche Sampling\texttt{Avalanche Sampling}, and apply it to various toy problems as well as the likelihood on the cosmological parameters Ωm\Omega_m and ww on the basis of data from the supernova distance redshift relation.

Keywords

Cite

@article{arxiv.2311.16218,
  title  = {Partition function approach to non-Gaussian likelihoods: macrocanonical partitions and replicating Markov-chains},
  author = {Maximilian Philipp Herzog and Heinrich von Campe and Rebecca Maria Kuntz and Lennart Röver and Björn Malte Schäfer},
  journal= {arXiv preprint arXiv:2311.16218},
  year   = {2024}
}

Comments

13 pages, 9 figures

R2 v1 2026-06-28T13:33:16.785Z