Abstract:We study ergodic optimization problems for one-dimensional diffusions with resetting boundaries, namely diffusions that are instantaneously reset according to prescribed distributions upon reaching designated boundary points. For controlled diffusions with resetting boundaries, we consider optimization of the ergodic cost over a broad class of admissible real-time controls. We derive a Hamilton--Jacobi--Bellman equation with non-local boundary conditions and establish a verification theorem for optimal stationary Markov controls. In the case of one resetting and one reflecting boundary, we also prove an existence result under weak integrability assumptions.
We then study renewal diffusions. Using their regenerative structure, we derive explicit formulas for the ergodic cost. These formulas are applied to optimization problems over families of drift coefficients, yielding explicit characterizations of optimal drift parameters in several examples. Finally, we study an infinite-penalty regime in which boundary hitting is prohibited and characterize the corresponding optimal ergodic cost.
From: Sergey Nikitin [view email]
[v1]
Mon, 22 Jun 2026 08:10:14 UTC (27 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。