SOURCE-LINKED INTELLIGENCE

Penalized Nonreversible Langevin for Constrained Sampling

arXiv · AI, language, vision and robotics · article · Sep 21, 2026 · UTC

We propose penalized nonreversible Langevin algorithms for sampling from $π(x)\propto e^{-f(x)}\mathbf 1_{\mathcal C}(x)$, where $\mathcal C\subset\mathbb R^d$ is a compact convex set. The algorithms combine a squared distance penalty with constant or compatible state dependent skew symmetric perturbations that preserve the penalized Gibbs distribution. For smooth, possibly nonconvex $f$, we derive nonasymptotic total variation bounds for the full gradient algorithm under a log Sobolev inequality. When unbiased stochastic gradients are available, we establish $2$-Wasserstein bounds under globa

Read original source ↗ Open in workspace

recordType

paper

region

Global