def mrp_func1(s: S,
mdp_rep=mdp_rep,
policy_func=policy_func) -> Mapping[S, float]:
s_dict = policy_func(s)
return sum_dicts([{s1: p * v
for s1, v in mdp_rep(s, a).items()}
for a, p in s_dict.items()])
def mdp_rep_to_mrp_rep1(
mdp_rep: SASf,
policy_rep: SAf
) -> SSf:
return {s: sum_dicts([{s1: p * v2 for s1, v2 in v[a].items()}
for a, p in policy_rep[s].items()])
for s, v in mdp_rep.items()}
def mdp_rep_to_mrp_rep1(mdp_rep: SASf, policy_rep: SAf) -> SSf:
return {
s:
sum_dicts([{s1: policy_rep[s].get(a, 0) * v2
for s1, v2 in v1.items()} for a, v1 in v.items()])
for s, v in mdp_rep.items()
}