feat: complete q1

inference.py

@@ -18,7 +18,7 @@ from typing import List, Dict, Tuple
 import busters
 import game
 import bayesNet as bn
-from bayesNet import normalize
+from bayesNet import constructEmptyBayesNet, normalize
 import hunters
 from util import manhattanDistance, raiseNotDefined
 from factorOperations import joinFactorsByVariableWithCallTracking, joinFactors
@@ -61,7 +61,26 @@ def constructBayesNet(gameState: hunters.GameState):
     variableDomainsDict = {}
 
     "*** YOUR CODE HERE ***"
-    raiseNotDefined()
+    # construct variables & edges
+    variables = [PAC, GHOST0, GHOST1, OBS0, OBS1]
+    edges = [(GHOST0, OBS0), (PAC, OBS0), (PAC, OBS1), (GHOST1, OBS1)]      # reference: explantion figure from UCB CS188
+                                                                            # https://inst.eecs.berkeley.edu/~cs188/sp24/assets/projects/busters_bayes_net_simplified.png
+    # agent's possible position for variable domain
+    possible_positions = []
+    for x in range(X_RANGE):
+        for y in range(Y_RANGE):
+            possible_positions.append((x, y))
+    variableDomainsDict[PAC] = possible_positions.copy()
+    variableDomainsDict[GHOST0] = possible_positions.copy()
+    variableDomainsDict[GHOST1] = possible_positions.copy()
+
+    # observation's variable domain
+    # UCB CS188's problem explantion:
+    #     - Observations here are non-negative, equal to Manhattan distances of Pacman to ghosts ± noise
+    max_observation = manhattanDistance((0, 0), (X_RANGE-1, Y_RANGE-1)) + MAX_NOISE
+    variableDomainsDict[OBS0] = list(range(max_observation+1))
+    variableDomainsDict[OBS1] = list(range(max_observation+1))
+
     "*** END YOUR CODE HERE ***"
 
     net = bn.constructEmptyBayesNet(variables, edges, variableDomainsDict)