nash.learning package

Submodules

nashpy.learning.fictitious_play module

Code to carry out fictitious learning

nashpy.learning.fictitious_play.fictitious_play(A: ndarray[Any, dtype[_ScalarType_co]], B: ndarray[Any, dtype[_ScalarType_co]], iterations: int, play_counts: Any | None = None) → Generator

Implement fictitious play

Parameters:

• A (array) – The row player payoff matrix.

• B (array) – The column player payoff matrix.

• iterations (int) – The number of iterations of the algorithm.

• play_counts (Optional) – The play counts.

Yields:

Generator – The play counts.

nashpy.learning.fictitious_play.get_best_response_to_play_count(A: ndarray[Any, dtype[_ScalarType_co]], play_count: ndarray[Any, dtype[_ScalarType_co]]) → int

Returns the best response to a belief based on the playing distribution of the opponent

Parameters:

• A (array) – The utility matrix.

• play_count (array) – The play counts.

Returns:

The action that corresponds to the best response.

Return type:

int

nashpy.learning.fictitious_play.update_play_count(play_count: ndarray[Any, dtype[_ScalarType_co]], play: int) → ndarray[Any, dtype[_ScalarType_co]]

Update a belief vector with a given play

Parameters:

• play_count (array) – The play counts.

• play (int) – The given play.

Returns:

The updated play counts.

Return type:

array