Guardado en:
Detalles Bibliográficos
Autores principales: Hagemann, Paul, Schütte, Janina, Sommer, David, Eigel, Martin, Steidl, Gabriele
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:https://arxiv.org/abs/2412.07637
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866915057197121536
author Hagemann, Paul
Schütte, Janina
Sommer, David
Eigel, Martin
Steidl, Gabriele
author_facet Hagemann, Paul
Schütte, Janina
Sommer, David
Eigel, Martin
Steidl, Gabriele
contents Our method proposes the efficient generation of samples from an unnormalized Boltzmann density by solving the underlying continuity equation in the low-rank tensor train (TT) format. It is based on the annealing path commonly used in MCMC literature, which is given by the linear interpolation in the space of energies. Inspired by Sequential Monte Carlo, we alternate between deterministic time steps from the TT representation of the flow field and stochastic steps, which include Langevin and resampling steps. These adjust the relative weights of the different modes of the target distribution and anneal to the correct path distribution. We showcase the efficiency of our method on multiple numerical examples.
format Preprint
id arxiv_https___arxiv_org_abs_2412_07637
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Sampling from Boltzmann densities with physics informed low-rank formats
Hagemann, Paul
Schütte, Janina
Sommer, David
Eigel, Martin
Steidl, Gabriele
Machine Learning
Optimization and Control
Our method proposes the efficient generation of samples from an unnormalized Boltzmann density by solving the underlying continuity equation in the low-rank tensor train (TT) format. It is based on the annealing path commonly used in MCMC literature, which is given by the linear interpolation in the space of energies. Inspired by Sequential Monte Carlo, we alternate between deterministic time steps from the TT representation of the flow field and stochastic steps, which include Langevin and resampling steps. These adjust the relative weights of the different modes of the target distribution and anneal to the correct path distribution. We showcase the efficiency of our method on multiple numerical examples.
title Sampling from Boltzmann densities with physics informed low-rank formats
topic Machine Learning
Optimization and Control
url https://arxiv.org/abs/2412.07637