why self.Q = self.parent.V + self.r

[yyht]

in the mcts.py line 86, why the Q of node is initialzied via self.Q = self.parent.V + self.r

since Q(s,a)=E_{p(s', r|s,a)}[r(s,a)+weight * V(s')]
In LLM-setting, p(s',r|s,a) = 1 given s and a, Q(s,a)=r(s,a)+weight * V(s')
In this case, node.Q = reward + node.V ? which is used in the def _back_propagate in line 272