cs.LGJan 29, 2026

Flow Perturbation++: Multi-Step Unbiased Jacobian Estimation for High-Dimensional Boltzmann Sampling

Authors: Xin Peng, Ang Gao

Organizations: State Key Laboratory of Information Photonics and Optical Communications, Beijing University of Posts and Telecommunications, Beijing 100876, China · School of Physical Science and Technology, Beijing University of Posts and Telecommunications, Beijing 100876, China

Abstract

The scalability of continuous normalizing flows (CNFs) for unbiased Boltzmann sampling remains limited in high-dimensional systems due to the cost of Jacobian-determinant evaluation, which requires DD backpropagation passes through the flow layers. Existing stochastic Jacobian estimators such as the Hutchinson trace estimator reduce computation but introduce bias, while the recently proposed Flow Perturbation method is unbiased yet suffers from high variance. We present Flow Perturbation++, a variance-reduced extension of Flow Perturbation that discretizes the probability-flow ODE and performs unbiased stepwise Jacobian estimation at each integration step. This multi-step construction retains the unbiasedness of Flow Perturbation while achieves substantially lower estimator variance. Integrated into a Sequential Monte Carlo framework, Flow Perturbation++ achieves significantly improved equilibrium sampling on a 1000D Gaussian Mixture Model and the all-atom Chignolin protein compared with Hutchinson-based and single-step Flow Perturbation baselines.

Figures & tables

Appendix figures & tables16 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. Bridging Stochastic Flow Maps and Boltzmann Generators with Normalizing Flows

    Sep 25, 2026Louis Grenioux, RuiKang OuYang, Luhuan WuBoltzmann SamplingMolecular Generation

  2. Jeffreys Flow: Robust Boltzmann Generators for Rare Event Sampling via Parallel Tempering Distillation

    Apr 7, 2026Guang Lin, Christian Moya, Di Qi +1Boltzmann Sampling

  3. Neural Sampling with Reweighted Normalizing Flows via the Wasserstein--Fisher--Rao JKO Scheme

    Oct 7, 2026Chenguang Duan, Johannes Hertrich, Gabriele SteidlSampling from Unnormalized DistributionsBoltzmann Sampling