A Note on the Existence of Optimal Policies in Total Reward Dynamic Programs with Compact Action Sets.
Rolando Cavazos-CadenaEugene A. FeinbergRaúl Montes-de-OcaPublished in: Math. Oper. Res. (2000)
Keyphrases
- stationary policies
- action sets
- total reward
- optimal policy
- markov decision processes
- finite state
- state space
- markov decision process
- average cost
- dynamic programming
- lot sizing
- reinforcement learning
- linear program
- sufficient conditions
- average reward
- infinite horizon
- long run
- markov decision problems
- decision problems
- policy iteration
- initial state
- multistage
- reinforcement learning algorithms
- state dependent
- action selection
- reward function
- dynamic environments
- action space
- inventory level
- decision processes
- search space
- markov chain
- dynamical systems
- state variables