Linear dependence of stationary distributions in ergodic Markov decision processes.
Ronald OrtnerPublished in: Oper. Res. Lett. (2007)
Keyphrases
- markov decision processes
- stationary distribution
- markov chain
- state space
- finite state
- initial state
- optimal policy
- random walk
- transition matrices
- average cost
- queueing networks
- policy iteration
- reinforcement learning
- queue length
- transition probabilities
- dynamic programming
- sufficient conditions
- steady state
- decision theoretic planning
- reachability analysis
- markov decision process
- infinite horizon
- model based reinforcement learning
- state dependent
- learning algorithm
- action sets
- action space
- average reward
- planning under uncertainty
- partially observable
- service times
- reward function
- markov decision problems
- long run
- generative model
- real time dynamic programming
- stochastic shortest path