Zero temperature versus long time for diffusions driven by metric projections onto convex sets

For a closed subset $C$ of Euclidean space, we study the diffusion whose drift pulls the state towards its nearest point in $C$, with noise intensity $T$. We let $T$ tend to zero and time tend to infinity. Convexity of $C$ is equivalent to a pathwise bound of the distance from $C$ by the norm of the process of Ornstein and Uhlenbeck. For arbitrary closed sets the bound holds in law. For compact convex $C$, we describe the invariant law by the curvature measures and the intrinsic volumes of $C$. The limit points of the law of the state form the orbit of reflected Brownian motion in $C$ started from the projection of the initial law. So the two limits commute only when this projection is the uniform law on $C$. Unless $C$ is a point, the mixing time is of order $1/T$ up to a logarithm. For the control problem the two limits of the value function commute for every compact convex set and every bounded Lipschitz terminal cost. A companion paper treats finite disjoint unions of convex sets.

Authors

Publication Details

Journal
arXiv (Cornell University)
Published
2026-10-26
DOI
https://doi.org/10.48550/arxiv.2610.35145
Primary Topic
Stochastic processes and financial applications
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
preprint

Zero temperature versus long time for diffusions driven by metric projections onto convex sets

Narayana Darapaneni
arXiv (Cornell University)
Stochastic processes and financial applications
preprint

Zero temperature versus long time for diffusions driven by metric projections onto convex sets

Narayana Darapaneni
preprint en

Abstract

For a closed subset $C$ of Euclidean space, we study the diffusion whose drift pulls the state towards its nearest point in $C$, with noise intensity $T$. We let $T$ tend to zero and time tend to infinity. Convexity of $C$ is equivalent to a pathwise bound of the distance from $C$ by the norm of the process of Ornstein and Uhlenbeck. For arbitrary closed sets the bound holds in law. For compact convex $C$, we describe the invariant law by the curvature measures and the intrinsic volumes of $C$. The limit points of the law of the state form the orbit of reflected Brownian motion in $C$ started from the projection of the initial law. So the two limits commute only when this projection is the uniform law on $C$. Unless $C$ is a point, the mixing time is of order $1/T$ up to a logarithm. For the control problem the two limits of the value function commute for every compact convex set and every bounded Lipschitz terminal cost. A companion paper treats finite disjoint unions of convex sets.

arXiv (Cornell University)
Stochastic processes and financial applications
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon