Existence of Optimal Policies for Semi-Markov Decision Processes Using Duality for Infinite Linear Programming.
Diego KlabjanDaniel AdelmanPublished in: SIAM J. Control. Optim. (2006)
Keyphrases
- semi markov decision processes
- linear programming
- optimal policy
- average reward
- dynamic programming
- markov decision processes
- linear program
- state space
- decision problems
- policy iteration
- multistage
- stationary policies
- finite horizon
- reinforcement learning
- long run
- primal dual
- markov decision problems
- optimal solution
- objective function
- markov decision process
- infinite horizon
- finite state
- np hard
- average reward reinforcement learning
- nonlinear programming
- column generation
- sufficient conditions
- decision processes
- lost sales
- dynamic programming algorithms