new def of features

cognitive_game_env.py

@@ -307,9 +307,9 @@ class CognitiveGame:
     :param trajs: generated by the expert
     :return: Nx3 feature map
     """
-    max_attempt = self.n_solution*self.n_attempt
-    max_time = self.timeout*self.n_solution*self.n_attempt
-    N = self.n_solution * self.n_attempt * self.n_user_action
+    max_attempt = self.task_length*self.n_max_attempt
+    max_time = self.timeout*self.task_length*self.n_max_attempt
+    N = (self.task_length+1) * self.n_max_attempt * len(self.user_action)
 
 
     feat = np.ones([N, 4])
@@ -318,11 +318,17 @@ class CognitiveGame:
     for traj in trajs:
       for s1, a, s2 in traj._t:
         ix, iy, iz = self.state_index_to_point(s1)
-        feat[s1, 0] = 0
-        feat[s1, 1] = 0#max_attempt*ix
-        feat[s1, 2] = ix
-        feat[s1, 3] = iy
-
+        # feat[s1, 0] = 0
+        # feat[s1, 1] = 0
+        feat[s1, 0] = (ix)
+        feat[s1, 1] = (iy)#*(ix+1)
+        feat[s1, 2] = iz
+        feat[s1, 3] = a
+        # for index in range(len(self.action_space)):
+        #   if index == a:
+        #     feat[s1, 3+index] = 1
+        #   else:
+        #     feat[s1, 3 + index] = 0
         # for trans in traj:
         #   if trans[0] == i:
         #     current_react_time = trans[2]