Sampling via Stochastic Interpolants by Langevin-based Velocity and Initialization Estimation in Flow ODEs
Duan, Jiao, Steidl, Wald, Yang, Zhang · math.NA,cs.LG,cs.NA,math.PR,stat.ML · 2026-08-07 · 原文 · 证据态 未学
We propose a novel method for sampling from unnormalized Boltzmann densities based on a probability flow ordinary differential equation (ODE) derived from linear stochastic interpolants. The key innovation of our approach is the use of a sequence of Langevin samplers to enable efficient simulation of the flow. Specifically, these Langevin samplers are employed (i) to generate samples from the interpolant distribution at intermediate times and (ii) to construct, starting from these intermediate times, a robust estimator of the velocity field governing the probability flow ODE. Theoretically, we provide convergence guarantees for both Langevin components, and establish a non-asymptotic convergence rate for the probability flow ODE. Extensive numerical experiments demonstrate the efficiency of the proposed method on challenging multimodal distributions across a range of dimensions, as well as its effectiveness in Bayesian inference tasks.
讲义
讲义·推断 依据「原文」自动生成的结构化摘要（推断），非原文表述；以原文为准。
1. 人话版
We propose a novel method for sampling from unnormalized Boltzmann densities based on a probability flow ordinary differential equation (ODE) derived from linear stochastic interpolants.
The key innovation of our approach is the use of a sequence of Langevin samplers to enable efficient simulation of the flow.
2. 领域脉络
本文类目：math.NA、cs.LG、cs.NA、math.PR、stat.ML，属于其所在研究脉络的最新进展。
3. 机制拆解
Theoretically, we provide convergence guarantees for both Langevin components, and establish a non-asymptotic convergence rate for the probability flow ODE.
4. 证据与数字
摘要未给出量化结果——留意原文的实验与数据。
5. 反例与边界
Specifically, these Langevin samplers are employed (i) to generate samples from the interpolant distribution at intermediate times and (ii) to construct, starting from these intermediate times, a robust estimator of the velocity field governing the probability flow ODE.
Extensive numerical experiments demonstrate the efficiency of the proposed method on challenging multimodal distributions across a range of dimensions, as well as its effectiveness in Bayesian inference tasks.
6. 跨领域连接与意外收获
横跨 5 个类目（math.NA、cs.LG、cs.NA、math.PR），关注其在你兴趣板块间的迁移面。
7. 可复用方法
把本文机制与你手头项目对照，找一个两周内能验证的最小实验。
8. 术语表
精读时把不熟的术语记入此处，作为下次回忆的锚点。
主动回忆
先合上内容自己复述，再点「显示」核对，然后如实自评。评分即时进 FSRS 排程。
讲义·推断 依据「原文」自动生成的结构化摘要（推断），非原文表述；以原文为准。
1. 人话版
We propose a novel method for sampling from unnormalized Boltzmann densities based on a probability flow ordinary differential equation (ODE) derived from linear stochastic interpolants.
The key innovation of our approach is the use of a sequence of Langevin samplers to enable efficient simulation of the flow.
2. 领域脉络
本文类目：math.NA、cs.LG、cs.NA、math.PR、stat.ML，属于其所在研究脉络的最新进展。
3. 机制拆解
Theoretically, we provide convergence guarantees for both Langevin components, and establish a non-asymptotic convergence rate for the probability flow ODE.
4. 证据与数字
摘要未给出量化结果——留意原文的实验与数据。
5. 反例与边界
Specifically, these Langevin samplers are employed (i) to generate samples from the interpolant distribution at intermediate times and (ii) to construct, starting from these intermediate times, a robust estimator of the velocity field governing the probability flow ODE.
Extensive numerical experiments demonstrate the efficiency of the proposed method on challenging multimodal distributions across a range of dimensions, as well as its effectiveness in Bayesian inference tasks.
6. 跨领域连接与意外收获
横跨 5 个类目（math.NA、cs.LG、cs.NA、math.PR），关注其在你兴趣板块间的迁移面。
7. 可复用方法
把本文机制与你手头项目对照，找一个两周内能验证的最小实验。
8. 术语表
精读时把不熟的术语记入此处，作为下次回忆的锚点。