Free-support Wasserstein Resampling in Sequential Monte Carlo
Abstract
In modern sequential Monte Carlo (SMC) applications, such as diffusion model samplers, propagating a particle is expensive, so only a few particles can be afforded. In this regime, each resampling step decides which regions of the state space survive, so a better resampling scheme is worth extra computation. However, classical resampling methods only replicate existing particles and suffer from sample impoverishment, especially when the particle size is small relative to the dimension. To create new particles, we relax the unbiasedness of resampling and control the mean squared error via a Wasserstein criterion. Specifically, we propose Wasserstein-Dirichlet resampling (WDR), which constructs an equal-weight measure that optimally approximates the weighted particle distribution by solving a free-support Wasserstein barycenter problem. We then randomize the resulting optimal coupling by Dirichlet mixing. Theoretically, we provide non-asymptotic mean squared error bounds that quantify the bias-variance trade-off and show that the rate depends on the intrinsic dimension. A truncated variant preserves high-weight particles while applying WDR to the remaining mass, targeting mild weight degeneracy. Experiments on state-space models and conditional image generation demonstrate that WDR consistently improves estimation accuracy, stability, and sample diversity. The improvements persist at matched wall-clock cost when the propagation cost dominates.
Related
- Differentiable Systematic Resampling for Variational Sequential Monte Carlo
- Sequince: a modular sequential Monte Carlo toolkit
- Anchored Reversible Jump Sequential Monte Carlo
- Parameter estimation in Conditional Sequential Monte Carlo algorithms through Particle Learning
- Steering Diffusion Models to Rare Events with Sequential Monte Carlo
BibTeX
@misc{li2026freesupport,
title = {{Free-support Wasserstein Resampling in Sequential Monte Carlo}},
author = {Mengyu Li and Cheng Meng and Jun S. Liu},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.17924},
eprint = {2610.17924},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.17924},
}
RIS
TY - UNPB
TI - Free-support Wasserstein Resampling in Sequential Monte Carlo
AU - Mengyu Li
AU - Cheng Meng
AU - Jun S. Liu
PY - 2026
DA - 2026/10/13
PB - arXiv
DO - 10.48550/arxiv.2610.17924
UR - https://arxiv.org/abs/2610.17924
AB - In modern sequential Monte Carlo (SMC) applications, such as diffusion model samplers, propagating a particle is expensive, so only a few particles can be afforded. In this regime, each resampling step decides which regions of the state space survive, so a better resampling scheme is worth extra computation. However, classical resampling methods only replicate existing particles and suffer from sample impoverishment, especially when the particle size is small relative to the dimension. To create new particles, we relax the unbiasedness of resampling and control the mean squared error via a Wasserstein criterion. Specifically, we propose Wasserstein-Dirichlet resampling (WDR), which constructs an equal-weight measure that optimally approximates the weighted particle distribution by solving a free-support Wasserstein barycenter problem. We then randomize the resulting optimal coupling by Dirichlet mixing. Theoretically, we provide non-asymptotic mean squared error bounds that quantify the bias-variance trade-off and show that the rate depends on the intrinsic dimension. A truncated variant preserves high-weight particles while applying WDR to the remaining mass, targeting mild weight degeneracy. Experiments on state-space models and conditional image generation demonstrate that WDR consistently improves estimation accuracy, stability, and sample diversity. The improvements persist at matched wall-clock cost when the propagation cost dominates.
ER -