Linear and dynamic programs for risk-sensitive cost minimization

TL;DR

This paper develops linear and dynamic programming approaches for infinite horizon risk-sensitive control, focusing on minimizing the long-term growth rate of cumulative costs in uncertain environments.

Contribution

It introduces a novel formulation that transforms risk-sensitive control problems into equivalent linear and dynamic programs for better computational handling.

Findings

Derived linear programs for risk-sensitive control.

Formulated dynamic programming equations for the problem.

Established equivalence between control problem and optimization programs.

Abstract

We derive equivalent linear and dynamic programs for infinite horizon risk-sensitive control for minimization of the asymptotic growth rate of the cumulative cost.