Abstract
Assume interest is in sampling from a probability distribution μ defined on (𝖹,𝒵). We develop a framework to construct sampling algorithms taking full advantage of numerical integrators of ODEs, say ψ:𝖹→𝖹 for one integration step, to explore μ efficiently and robustly. The popular Hybrid/Hamiltonian Monte Carlo (HMC) algorithm [Duane, 1987], [Neal, 2011] and its derivatives are example of such a use of numerical integrators. However we show how the potential of integrators can be exploited beyond current ideas and HMC sampling in order to take into account aspects of the geometry of the target distribution. A key idea is the notion of integrator snippet, a fragment of the orbit of an ODE numerical integrator ψ, and its associate probability distribution μ¯, which takes the form of a mixture of distributions derived from μ and ψ. Exploiting properties of mixtures we show how samples from μ¯ can be used to estimate expectations with respect to μ. We focus here primarily on Sequential Monte Carlo (SMC) algorithms, but the approach can be used in the context of Markov chain Monte Carlo algorithms as discussed at the end of the manuscript. We illustrate performance of these new algorithms through numerical experimentation and provide preliminary theoretical results supporting observed performance.
Original language  English 

DOIs  
Publication status  Published  20 Apr 2024 
Fingerprint
Dive into the research topics of 'Monte Carlo sampling with integrator snippets'. Together they form a unique fingerprint.Projects
Student theses

Approximate Manifold Sampling: Robust Bayesian Inference for Machine Learning
Author: Camara Escudero, M., 18 Jun 2024Supervisor: Andrieu, C. (Supervisor), Beaumont, M. (Supervisor) & Johnson, O. (Supervisor)
Student thesis: Doctoral Thesis › Doctor of Philosophy (PhD)
File