Merge pull request #2 from nrecoder/Tanvi

RunGame good to go Contains olivers investigations

modeling/farmgame.py

@@ -140,6 +140,43 @@ def get_cost(self, action: Action) -> float:
                 n_steps += 3 # three because moving 1,2 away
         return n_steps * self.stepcost
 
+    def all_objects_picked_up(self, player_color: str) -> bool:
+        player_items = [
+            item for item in self.items if item.color == player_color and item.type == ActionType.veggie
+        ]
+        return all(item.status == "backpack" for item in player_items)
+
+    def get_steps(self, action: Action) -> int:
+        # no cost if no actions were available to the player
+        if action.type == ActionType.none:
+            return 0
+
+        # fixed cost if voluntarily passing
+        if action.type == ActionType.pillow or action.type == ActionType.timeout:
+            return self.pillowcost
+
+        # whose turn is it?
+        currentplayer = self.whose_turn()
+
+        # player moves to location. decrease energy by steps taken.
+        n_steps = utils.getManhattanDistance(currentplayer["loc"], action.loc)
+        # if going to the farm the player might have to step around the farm wall
+        if action.type == ActionType.box and currentplayer["loc"]["y"] <= self.farmbox.loc["y"]:
+            n_steps += 2
+        # if the player, other player and target are in the same row or column, and the other
+        # player is in the way, then add 2 sidesteps to go around them.
+        elif utils.is_in_line(currentplayer["loc"], self.other_player()["loc"], action.loc):
+            # Add 2 sidesteps if the other player is in the way. Somehow getPath doesn't cover this.
+            n_steps += 2
+
+        # decrease energy for move out of the way
+        if action.type == ActionType.box:
+            statuses = Counter(item.status for item in self.items)
+            # check if this move will end the game. If it does then the payer does not move away.
+            if "farm" in statuses or statuses["backpack"] > len(currentplayer["backpack"]["contents"]):
+                n_steps += 3 # three because moving 1,2 away
+        return n_steps
+
     def take_action(self, action: Action, inplace=True) -> Farm:
         if inplace:
             new_state = self

modeling/mcts.py

@@ -220,7 +220,9 @@ def run_simulation(self):
 
             # everyone gets punished for slowness regardless of reward policy
             # punish these bots for taking too long thoooooo
-            self.rewards[(player,q)]-= simstate.trial # TEMPORAL COST - THIS PENALTY MAY NEED TO BE TUNED
+            penalty_factor = 0.0  # Scale factor
+            self.rewards[(player, q)] -= penalty_factor * simstate.trial
+            # self.rewards[(player,q)]-= simstate.trial # TEMPORAL COST - THIS PENALTY MAY NEED TO BE TUNED
 
     # here, we use string representation of hash rather than int that hash() itself gives
     def hash_and_store(self, s):