Stable Optimal Control and Semicontractive Dynamic Programming

TL;DR

This paper introduces a new framework for analyzing stable policies in infinite horizon deterministic optimal control, extending classical results to cases where policies may not stabilize the system, and characterizes solutions to Bellman's equation.

Contribution

It proposes a unifying notion of stable feedback policies based on cost perturbation, and characterizes the solutions and convergence properties of modified dynamic programming algorithms.

Findings

$ ilde J$ and $ar J$ are solutions to Bellman's equation.

The smallest and largest solutions are $ ilde J$ and $J^+$.

Modified value and policy iteration algorithms have specific convergence regions.

Abstract

We consider discrete-time infinite horizon deterministic optimal control problems with nonnegative cost per stage, and a destination that is cost-free and absorbing. The classical linear-quadratic regulator problem is a special case. Our assumptions are very general, and allow the possibility that the optimal policy may not be stabilizing the system, e.g., may not reach the destination either asymptotically or in a finite number of steps. We introduce a new unifying notion of stable feedback policy, based on perturbation of the cost per stage, which in addition to implying convergence of the generated states to the destination, quantifies the speed of convergence. We consider the properties of two distinct cost functions: $\jstar$, the overall optimal, and $\hat J$, the restricted optimal over just the stable policies. Different classes of stable policies (with different speeds of convergence) may yield different values of $\hat J$. We show that for any class of stable policies, $\hat J$ is a solution of Bellman's equation, and we characterize the smallest and the largest solutions: they are $\jstar$, and $J^+$, the restricted optimal cost function over the class of (finitely) terminating policies. We also characterize the regions of convergence of various modified versions of value and policy iteration algorithms, as substitutes for the standard algorithms, which may not work in general.