fictplay

class quantecon.game_theory.fictplay.FictitiousPlay(data, gain=None)[source]

Bases: object

Class representing a fictitious play model.

Parameters:
dataNormalFormGame, or array_like

The game played in a fictitious play model. data must be either NormalFormGame or an array. See NormalFormGame.

gainscalar(float), optional(default=None)

The gain of fictitous play model. If gain is None, the model becomes a decreasing gain model. If gain is a scalar, the model becomes a constant gain model.

Attributes:
gNomalFormGame

The game played in the model.

Nscalar(int)

The number of players in the model.

playerstuple(Player)

Tuple of the Player instances in the model.

nums_actionstuple(int)

Tuple of the number of actions, one for each player.

Methods

play([actions, num_reps, t_init, out])

Return a new action profile which is updated by playing the game num_reps times.

time_series(ts_length[, init_actions, t_init])

Return a tuple of arrays representing a time series of mixed action profiles.

play(actions=None, num_reps=1, t_init=0, out=None, **options)[source]

Return a new action profile which is updated by playing the game num_reps times.

Parameters:
actionstuple(array_like(float)), optional(default=None)

The action profile in the initial period. If None, selected randomly.

num_repsscalar(int), optional(default=1)

The number of iterations.

t_initscalar(int), optional(default=0)

The period when the game starts.

outtuple(array_like(float)), optional(default=None)

Alternative output tuple of arrays in which to place the result. Must be of the same shape as the expected output.

**optionsKeyword arguments passed to the best response method and

other methods.

Returns:
tuple(ndarray(float, ndim=1))

The mixed action profile after iteration.

time_series(ts_length, init_actions=None, t_init=0, **options)[source]

Return a tuple of arrays representing a time series of mixed action profiles.

Parameters:
ts_lengthscalar(int)

The number of iterations.

init_actionstuple(int), optional(default=None)

The action profile in the initial period. If None, selected randomly.

t_initscalar(int), optional(default=0)

The period when the game starts.

**optionsKeyword arguments passed to the best response method and

other methods.

Returns:
tuple(ndarray(float, ndim=2))

Tuple of arrays representing time series of mixed action profile.

class quantecon.game_theory.fictplay.StochasticFictitiousPlay(data, distribution, gain=None)[source]

Bases: FictitiousPlay

Class representing a stochastic fictitious play model.

Parameters:
dataNormalFormGame or array_like

The game played in the stochastic fictitious play model.

distributionscipy.stats object

The distribution of payoff shocks, which is a scipy.stats object.

gainscalar(scalar), optional(default=None)

The gain of fictitious play model. If gain is None, the model becomes a decreasing gain model. If gain is a scalar, the model becomes a constant gain model.

Attributes:
See attributes of `FictitousPlay`.

Methods

play([actions, num_reps, t_init, out])

Return a new action profile which is updated by playing the game num_reps times.

time_series(ts_length[, init_actions, t_init])

Return a tuple of arrays representing a time series of mixed action profiles.