Robustness of optimal control for controlled regime-switching diffusions with incorrect models

TL;DR

This paper demonstrates that optimal control policies for regime-switching diffusions remain robust under model misspecification, with value functions and policies converging as models approximate the true system.

Contribution

It extends robustness analysis from diffusion processes to hybrid systems with continuous and discrete dynamics, covering various cost formulations.

Findings

Value functions and policies converge under model approximation.

Robustness holds across multiple cost criteria.

Analysis relies on regularity of weakly coupled HJB systems.

Abstract

This paper investigates the robustness of stochastic optimal control for controlled regime switching diffusions. We consider systems driven by both continuous fluctuations and discrete regime changes, allowing for model misspecification in both the diffusion and switching components. Within a unified framework, we study four classical cost formulations finite horizon, infinite-horizon discounted and ergodic costs, and the exit time cost, and establish continuity of value functions and robustness of optimal controls. Specifically, we show that as a sequence of approximating regime switching models converges to the true model, the associated value functions and optimal policies converge as well, ensuring vanishing performance loss. The analysis relies on the regularity of the solution to the associated weakly coupled HJB systems, and their stochastic representation. The results extend the robustness framework developed for diffusion processes to a significantly broader class of hybrid systems with interacting continuous and discrete dynamics.