void Learn(Stat s, ActIdx a, Stat next_s, float reward)
{
QActIdx qa = MaxQActIdx(next_s);
q_table[s.Idx(), a.Idx()] +=
alpha * (reward + gamma * qa.q - q_table[s.Idx(), a.Idx()]);
}
ActIdx Policy(Stat s)
{
// FIXME
if (Random.value < 0.2)
{
return(new ActIdx(Random.Range(0, ActIdx.max_idx)));
}
QActIdx qa = MaxQActIdx(s);
return(qa.a_idx);
}
