New sufficient conditions for average optimality in continuous-time Markov decision processes.
Liuer YeXianping GuoPublished in: Math. Methods Oper. Res. (2010)
Keyphrases
- stationary policies
- sufficient conditions
- markov decision processes
- average cost
- optimal policy
- action sets
- state space
- finite state
- optimal control
- dynamic programming
- reinforcement learning
- transition matrices
- exponential stability
- average reward
- fixed point
- long run
- linear program
- infinite horizon
- partially observable
- lot sizing
- factored mdps
- markov chain
- decision theoretic planning
- asymptotic stability
- markov decision problems
- policy iteration
- initial state
- planning under uncertainty
- action space
- markov decision process
- discounted reward
- model based reinforcement learning