Scholarlink

<p>Flow-based generative models have been employed for Boltzmann sampling tasks, but their application to high-dimensional systems is hindered by the significant computational cost of obtaining the Jacobian of the flow. We introduce a flow perturbation method that bypasses this bottleneck by injecting stochastic perturbations into the flow, delivering orders-of-magnitude speed-ups. Unlike the Hutchinson estimator, our approach is inherently unbiased in Boltzmann sampling. Notably, this method significantly accelerates Boltzmann sampling of a Chignolin mutant with all atomic Cartesian coordinates explicitly represented, while delivering more accurate results than the Hutchinson estimator.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Flow perturbation to accelerate Boltzmann sampling

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年7月17日

查看全文
  • Xin Peng,
  • Ang Gao

摘要

Flow-based generative models have been employed for Boltzmann sampling tasks, but their application to high-dimensional systems is hindered by the significant computational cost of obtaining the Jacobian of the flow. We introduce a flow perturbation method that bypasses this bottleneck by injecting stochastic perturbations into the flow, delivering orders-of-magnitude speed-ups. Unlike the Hutchinson estimator, our approach is inherently unbiased in Boltzmann sampling. Notably, this method significantly accelerates Boltzmann sampling of a Chignolin mutant with all atomic Cartesian coordinates explicitly represented, while delivering more accurate results than the Hutchinson estimator.