Real-world uses

Created for this library

1. 1.

   A trading research team checks the Markov property assumption holds well enough on its problem before applying standard RL algorithms.

2. 2.

   A research team augments its state with additional history when the raw state violates the Markov property too strongly.

3. 3.

   An RL engineer ensures the state representation captures enough history to make the Markov property reasonable for the routing problem.