import numpy as np
from .policy import ParametricPolicy
[docs]class DeterministicPolicy(ParametricPolicy):
"""
Simple parametric policy representing a deterministic policy. As
deterministic policies are degenerate probability functions where all
the probability mass is on the deterministic action,they are not
differentiable, even if the mean value approximator is differentiable.
"""
[docs] def __init__(self, mu):
"""
Constructor.
Args:
mu (Regressor): the regressor representing the action to select
in each state.
"""
self._approximator = mu
self._add_save_attr(_approximator='mushroom')
[docs] def get_regressor(self):
"""
Getter.
Returns:
the regressor that is used to map state to actions.
"""
return self._approximator
[docs] def __call__(self, state, action):
policy_action = self._approximator.predict(state)
return 1. if np.array_equal(action, policy_action) else 0.
[docs] def draw_action(self, state):
return self._approximator.predict(state)
[docs] def set_weights(self, weights):
self._approximator.set_weights(weights)
[docs] def get_weights(self):
return self._approximator.get_weights()
@property
def weights_size(self):
return self._approximator.weights_size