Linear programming formulation of MDPs in countable state space: The multichain case.
Arie HordijkJean B. LasserrePublished in: Math. Methods Oper. Res. (1994)
Keyphrases
- markov decision processes
- state space
- linear programming
- dynamic programming
- optimal policy
- average cost
- markov chain
- markov decision problems
- policy iteration
- average reward
- reinforcement learning
- finite state
- markov decision process
- heuristic search
- factored mdps
- partially observable
- linear program
- quadratic programming
- action space
- decision theoretic planning
- factored markov decision processes
- reinforcement learning algorithms
- dynamical systems
- state variables
- integer programming
- planning problems
- finite horizon
- decision problems
- lp relaxation
- action sets
- reward function
- infinite horizon
- particle filter
- np hard
- primal dual
- objective function