build_opti#
- statrl.settings.markovdecisionprocess.discrete_nostructure.agents._Oracle.build_opti(name, env, nS, nA)[source]#
Build the oracle for an environment.
- Parameters:
name (str) – Environment name. Currently ignored: the per-map hand-coded oracles are commented out, so every environment gets the generic solver.
env (DiscreteMDP) – The environment to solve.
nS (int) – Numbers of states and actions.
nA (int) – Numbers of states and actions.
- Returns:
An oracle whose policy was computed by value iteration.
- Return type: