Questions tagged [markov-decision-process]

For questions related to Markov decision processes (MDPs), which model decision making in time-varying and usually stochastic environments.

Filter by
Sorted by
Tagged with
9
votes
1answer
84 views

Infinite horizon versus finite horizon MDP

When can we approximate a finite horizon MDP with infinite horizon? Can we use infinite horizon stochastic shortest path problem on a directed acyclic graph?
8
votes
0answers
95 views

For the solution of complex gaming are we to leave behind approximation and fixed algorithms in favor of agents?

For an evolutionary extensive form game is the use of multi-agent systems a necessity or can more traditional approximate Nash equilibria and mean-field interactions suffice? Few games can be ...
9
votes
1answer
86 views

Open source Markov decision process solvers

Which sources are out there that provide reliable and fast MDP solvers? I prefer that the library is callable from C++ but other languages are fine too. There is indeed an abundance of repositories ...
7
votes
1answer
62 views

Convergence rate value iteration

Value iteration is one of the most commonly used methods to solve Markov decision processes. Its convergence rate obviously depends on the number of states and actions. However, the convergence rate ...