def get_act_value_func(self, pol_func: PolicyType) -> QFType:
return self.get_qv_func_fa(
lambda s, pol_func=pol_func: get_pdf_from_samples(
pol_func(s)(len(self.state_action_func(s)) *
RLFuncApproxBase.NUM_SAMPLES_PER_ACTION)
)
)
def get_value_func(self, polf: PolicyType) -> Callable[[S], float]:
pol = Policy({
s: get_pdf_from_samples(
polf(s)(len(v) * TabularBase.NUM_SAMPLES_PER_ACTION))
for s, v in self.get_state_action_dict().items()
})
# noinspection PyShadowingNames
def vf(state: S, pol=pol) -> float:
return self.get_value_func_dict(pol)[state]
return vf
def get_value_func(self, pol_func: PolicyType) -> VFType:
return self.get_value_func_fa(lambda s: get_pdf_from_samples(
pol_func(s)
(len(self.state_action_func(s)) * ADP.NUM_SAMPLES_PER_ACTION)))