A Mixed Value and Policy Iteration Method for Stochastic Control with Universally Measurable Policies

DOI10.1287/moor.2014.0704zbMath1329.90157arXiv1308.3814OpenAlexW2126905415MaRDI QIDQ3465941

Huizhen Yu, Dimitri P. Bertsekas

Publication date: 29 January 2016

Published in: Mathematics of Operations Research (Search for Journal in Brave)

Full work available at URL: https://arxiv.org/abs/1308.3814

zbMATH Keywords

convergencemeasurabilitypolicy iterationvalue iterationdiscrete-time stochastic controlBorel spaces Markov decision processtotal cost criteria

Mathematics Subject Classification ID

Dynamic programming (90C39) Optimal stochastic control (93E20) Markov and semi-Markov decision processes (90C40)

Related Items (5)

Regular Policies in Abstract Dynamic Programming ⋮ Average Cost Optimality Inequality for Markov Decision Processes with Borel Spaces and Universally Measurable Policies ⋮ On the Minimum Pair Approach for Average Cost Markov Decision Processes with Countable Discrete Action Spaces and Strictly Unbounded Costs ⋮ On Convergence of Value Iteration for a Class of Total Cost Markov Decision Processes ⋮ Robust shortest path planning and semicontractive dynamic programming

Cites Work

• Unnamed Item
• Unnamed Item
• Unnamed Item
• Q-learning and policy iteration algorithms for stochastic shortest path problems
• The optimal reward operator in dynamic programming
• The optimal reward operator in special classes of dynamic programming problems
• Borel-programmable functions
• Value iteration and optimization of multiclass queueing networks
• Q-Learning and Enhanced Policy Iteration in Discounted Dynamic Programming
• Average Cost Markov Decision Processes with Weakly Continuous Transition Probabilities
• A simple condition for regularity in negative programming
• Stability and characterisation conditions in negative programming
• A simple proof of Whittle's bridging condition in dynamic programming
• The Optimal Reward Operator in Negative Dynamic Programming
• Conditions for optimality in dynamic programming and for the limit of n-stage optimal policies to be optimal
• On the Optimality of Structured Policies in Countable Stage Decision Processes. II: Positive and Negative Problems
• Monotone Mappings with Application in Dynamic Programming
• Borel-approachable functions
• Alternative Theoretical Frameworks for Finite Horizon Discrete-Time Stochastic Optimal Control
• Universally Measurable Policies in Dynamic Programming
• Markovian Decision Processes with Compact Action Spaces
• Discrete-Time Controlled Markov Processes with Average Cost Criterion: A Survey
• Memoryless Strategies in Finite-Stage Dynamic Programming
• Discounted Dynamic Programming
• Negative Dynamic Programming
• On Finding Optimal Policies in Discrete Dynamic Programming with No Discounting
• A Borel Set Not Containing a Graph
• Discrete Dynamic Programming with a Small Interest Rate
• Discrete Dynamic Programming with Sensitive Discount Optimality Criteria
• Infinite time reachability of state-space regions by using feedback control
• Non-Existence of Everywhere Proper Conditional Distributions
• Handbook of Markov decision processes. Methods and applications

This page was built for publication: A Mixed Value and Policy Iteration Method for Stochastic Control with Universally Measurable Policies