The value iteration algorithm is not strongly polynomial for discounted dynamic programming.
Eugene A. FeinbergJefferson HuangPublished in: Oper. Res. Lett. (2014)
Keyphrases
- dynamic programming
- strongly polynomial
- markov decision processes
- linear programming
- cost function
- knapsack problem
- optimal solution
- worst case
- infinite horizon
- np hard
- special case
- computational complexity
- objective function
- learning algorithm
- state space
- optimization algorithm
- ant colony optimization
- linear program
- benchmark problems
- reinforcement learning
- genetic algorithm
- optimization problems
- search space
- feasible solution