Fast and Efficient Parallel Sampling Using Higher Order Langevin Dynamics

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Mahajan, Jaideep, Zhang, Kaihong, Liang, Feng, Liu, Jingbo
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911659786764288
author Mahajan, Jaideep
Zhang, Kaihong
Liang, Feng
Liu, Jingbo
author_facet Mahajan, Jaideep
Zhang, Kaihong
Liang, Feng
Liu, Jingbo
contents We study parallel sampling from high-dimensional strongly log-concave distributions. Langevin-based samplers converge rapidly in continuous time, but their discretizations are typically sequential and often require polynomially many steps in the dimension $d$, the target accuracy $\varepsilon^{-1}$, or both. Picard-based parallel sampling methods reduce this sequential depth to polylogarithmic scale by solving for many time-discretization points in parallel; however, existing guarantees often require a polynomial number of processors, leading to substantial memory and gradient-evaluation costs in high dimensions. We show that higher-order Langevin structure can reduce this parallel resource burden while preserving polylogarithmic sequential depth. Our method combines arbitrary-order Langevin dynamics with blockwise Lagrange polynomial interpolation. This sharper discretization reduces the number of parallel points required to achieve a target accuracy. Our results cover both higher-order smooth potentials and ridge-separable potentials, including models such as Bayesian logistic regression and two-layer neural networks, and improve upon the space complexity of the current literature on parallel log-concave sampling.
format Preprint
id arxiv_https___arxiv_org_abs_2510_18242
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Fast and Efficient Parallel Sampling Using Higher Order Langevin Dynamics
Mahajan, Jaideep
Zhang, Kaihong
Liang, Feng
Liu, Jingbo
Statistics Theory
Methodology
Machine Learning
We study parallel sampling from high-dimensional strongly log-concave distributions. Langevin-based samplers converge rapidly in continuous time, but their discretizations are typically sequential and often require polynomially many steps in the dimension $d$, the target accuracy $\varepsilon^{-1}$, or both. Picard-based parallel sampling methods reduce this sequential depth to polylogarithmic scale by solving for many time-discretization points in parallel; however, existing guarantees often require a polynomial number of processors, leading to substantial memory and gradient-evaluation costs in high dimensions. We show that higher-order Langevin structure can reduce this parallel resource burden while preserving polylogarithmic sequential depth. Our method combines arbitrary-order Langevin dynamics with blockwise Lagrange polynomial interpolation. This sharper discretization reduces the number of parallel points required to achieve a target accuracy. Our results cover both higher-order smooth potentials and ridge-separable potentials, including models such as Bayesian logistic regression and two-layer neural networks, and improve upon the space complexity of the current literature on parallel log-concave sampling.
title Fast and Efficient Parallel Sampling Using Higher Order Langevin Dynamics
topic Statistics Theory
Methodology
Machine Learning
url https://arxiv.org/abs/2510.18242