simulation with differect available actions function

DecMCTS.py

@@ -90,6 +90,7 @@ class Tree:
             avail_actions_func,
             state_store_func,
             sim_selection_func,
+            sim_avail_actions_func,
             comm_n,
             robot_id,
             c_p=1):
@@ -98,6 +99,7 @@ class Tree:
         self.graph = nx.DiGraph()
         self.reward = reward_func
         self.available_actions = avail_actions_func
+        self.sim_available_actions = sim_avail_actions_func
         self.state_store = state_store_func
         self.sim_selection_func = sim_selection_func
         self.c_p = c_p
@@ -243,15 +245,16 @@ class Tree:
         best_rollout = None
         for i in range(nsims):
             temp_state = self.graph.node[start_node]["state"]
+            state[self.id] = temp_state
 
             d = 0 # depth
             while d<depth: # also breaks at no available options
                 d += 1
 
                 # Get the available actions
-                options = self.available_actions(
+                options = self.sim_available_actions(
                             self.data,
-                            temp_state,
+                            state,
                             self.id
                         )
 
@@ -265,9 +268,9 @@ class Tree:
 
                 # add that to the actions of the current robot
                 temp_state = self.state_store(self.data, temp_state, sim_action, self.id)
+                state[self.id] = temp_state
 
             # calculate the reward at the end of simulation
-            state[self.id] = temp_state
             rew = self.reward(self.data, state) \
                 - self.reward(self.data, self._null_state(state))