Generalizing Dijkstra's Algorithm and Gaussian Elimination for Solving MDPs
Abstract
The authors study the problem of computing the optimal value function for a Markov Decision Process (MDP) with positive costs. Computing this function quickly and accurately is a basic step in many schemes for deciding how to act in stochastic environments. There are efficient algorithms that compute value functions for special types of MDPs. For deterministic MDPs with S states and A actions, Dijkstra's algorithm runs in time O(AS log S). And, in single-action MDPs (Markov chains), standard linear-algebraic algorithms find the value function in time O(S sup 3), or faster by taking advantage of sparsity or good conditioning. Algorithms for solving general MDPs can take much longer: the authors are not aware of any speed guarantees better than those for comparably sized linear programs. They present a family of algorithms that reduce to Dijkstra's algorithm when applied to deterministic MDPs, and to standard techniques for solving linear equations when applied to Markov chains. More importantly, they demonstrate experimentally that these algorithms perform well when applied to MDPs that "almost" have the required special structure.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 2005
- Accession Number
- ADA456811
Entities
People
- Geoffrey J. Gordon
- H. B. Mcmahan
Organizations
- Carnegie Mellon University