(fix) Bug in sample() method of FiniteMDP related to done flag

rlberry-py · Dec 9, 2021 · 25a12f8 · 25a12f8
1 parent 091e156
commit 25a12f8
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/rlberry/envs/finite/finite_mdp.py b/rlberry/envs/finite/finite_mdp.py
@@ -121,7 +121,7 @@ def sample(self, state, action):
         prob = self.P[state, action, :]
         next_state = self.rng.choice(self._states, p=prob)
         reward = self.reward_fn(state, action, next_state)
-        done = self.is_terminal(self.state)
+        done = self.is_terminal(state)
         info = {}
         return next_state, reward, done, info