Updated for Python3 and fixed a bug

README.md

@@ -123,3 +123,8 @@ Ultimately, the results obtained here demonstrate the usefulness of convolutiona
 [3] Gerald Tesauro. Temporal difference learning and td-gammon. Communications of the ACM, 38(3):58–68, 1995.
 
 [4] Martin Riedmiller. Neural fitted q iteration–first experiences with a data efficient neural reinforcement learning method. In Machine Learning: ECML 2005, pages 317–328. Springer, 2005.
+
+
+## 9. installation
+
+[python2 pygame installer](http://www.pygame.org/ftp/pygame-1.9.2pre-py2.7-macosx10.7.mpkg.zip)

Raw Game Code/mario.py

@@ -0,0 +1,4 @@
+from Mario_Level_1 import mario_level_1
+
+if __name__ == '__main__':
+    mario_level_1.run_game()

Raw Game Code/mario.py~

@@ -0,0 +1,4 @@
+from Mario-Level-1 import mario_level_1
+
+if __name__ == '__main__':
+    mario_level_1.run_game()

Raw Game Code/pong.py

@@ -23,7 +23,7 @@
 bar2 = bar.convert()
 bar2.fill((255,255,255))
 circ_sur = pygame.Surface((15,15))
-circ = pygame.draw.circle(circ_sur,(255,255,255),(15/2,15/2),15/2)
+circ = pygame.draw.circle(circ_sur,(255,255,255),(15//2,15//2),15//2)
 circle = circ_sur.convert()
 circle.set_colorkey((0,0,0))
 
@@ -73,7 +73,7 @@
 
     # movement of circle
     time_passed = clock.tick(30)
-    time_sec = time_passed / 1000.0
+    time_sec = time_passed // 1000.0
 
     circle_x += speed_x * time_sec
     circle_y += speed_y * time_sec

Raw Game Code/testing/test.py

@@ -0,0 +1,3 @@
+
+def hi():
+	print "hello"

Raw Game Code/tetris.py

@@ -17,7 +17,7 @@
 MOVESIDEWAYSFREQ = 0.15
 MOVEDOWNFREQ = 0.1
 
-XMARGIN = int((WINDOWWIDTH - BOARDWIDTH * BOXSIZE) / 2)
+XMARGIN = int((WINDOWWIDTH - BOARDWIDTH * BOXSIZE) // 2)
 TOPMARGIN = WINDOWHEIGHT - (BOARDHEIGHT * BOXSIZE) - 5
 
 #               R    G    B
@@ -326,17 +326,17 @@ def showTextScreen(text):
     # center of the screen until a key is pressed.
     # Draw the text drop shadow
     titleSurf, titleRect = makeTextObjs(text, BIGFONT, TEXTSHADOWCOLOR)
-    titleRect.center = (int(WINDOWWIDTH / 2), int(WINDOWHEIGHT / 2))
+    titleRect.center = (int(WINDOWWIDTH // 2), int(WINDOWHEIGHT // 2))
     DISPLAYSURF.blit(titleSurf, titleRect)
 
     # Draw the text
     titleSurf, titleRect = makeTextObjs(text, BIGFONT, TEXTCOLOR)
-    titleRect.center = (int(WINDOWWIDTH / 2) - 3, int(WINDOWHEIGHT / 2) - 3)
+    titleRect.center = (int(WINDOWWIDTH // 2) - 3, int(WINDOWHEIGHT // 2) - 3)
     DISPLAYSURF.blit(titleSurf, titleRect)
 
     # Draw the additional "Press a key to play." text.
     pressKeySurf, pressKeyRect = makeTextObjs('Press a key to play.', BASICFONT, TEXTCOLOR)
-    pressKeyRect.center = (int(WINDOWWIDTH / 2), int(WINDOWHEIGHT / 2) + 100)
+    pressKeyRect.center = (int(WINDOWWIDTH // 2), int(WINDOWHEIGHT // 2) + 100)
     DISPLAYSURF.blit(pressKeySurf, pressKeyRect)
 
     while checkForKeyPress() == None:
@@ -356,7 +356,7 @@ def checkForQuit():
 def calculateLevelAndFallFreq(score):
     # Based on the score, return the level the player is on and
     # how many seconds pass until a falling piece falls one space.
-    level = int(score / 10) + 1
+    level = int(score // 10) + 1
     fallFreq = 0.27 - (level * 0.02)
     return level, fallFreq
 
@@ -365,7 +365,7 @@ def getNewPiece():
     shape = random.choice(list(PIECES.keys()))
     newPiece = {'shape': shape,
                 'rotation': random.randint(0, len(PIECES[shape]) - 1),
-                'x': int(BOARDWIDTH / 2) - int(TEMPLATEWIDTH / 2),
+                'x': int(BOARDWIDTH // 2) - int(TEMPLATEWIDTH // 2),
                 'y': -2, # start it above the board (i.e. less than 0)
                 'color': random.randint(0, len(COLORS)-1)}
     return newPiece

Wrapped Game Code/pong_fun.py

@@ -25,7 +25,7 @@
 bar2 = bar.convert()
 bar2.fill((255,255,255))
 circ_sur = pygame.Surface((15,15))
-circ = pygame.draw.circle(circ_sur,(255,255,255),(15/2,15/2),15/2)
+circ = pygame.draw.circle(circ_sur,(255,255,255),(15//2,15//2),15//2)
 circle = circ_sur.convert()
 circle.set_colorkey((0,0,0))
 font = pygame.font.SysFont("calibri",40)

Wrapped Game Code/tetris_fun.py

@@ -16,7 +16,7 @@
 MOVESIDEWAYSFREQ = 0.15
 MOVEDOWNFREQ = 0.1
 
-XMARGIN = int((WINDOWWIDTH - BOARDWIDTH * BOXSIZE) / 2)
+XMARGIN = int((WINDOWWIDTH - BOARDWIDTH * BOXSIZE) // 2)
 TOPMARGIN = WINDOWHEIGHT - (BOARDHEIGHT * BOXSIZE) - 5
 
 #               R    G    B
@@ -396,7 +396,7 @@ def makeTextObjs(self,text, font, color):
     def calculateLevelAndFallFreq(self):
         # Based on the self.score, return the self.level the player is on and
         # how many seconds pass until a falling piece falls one space.
-        self.level = min(int(self.lines / 10) + 1, 10)
+        self.level = min(int(self.lines // 10) + 1, 10)
         self.fallFreq = 0.27 - (self.level * 0.02)
         return self.level, self.fallFreq
 
@@ -405,7 +405,7 @@ def getNewPiece(self):
         shape = random.choice(list(PIECES.keys()))
         newPiece = {'shape': shape,
                     'rotation': random.randint(0, len(PIECES[shape]) - 1),
-                    'x': int(BOARDWIDTH / 2) - int(TEMPLATEWIDTH / 2),
+                    'x': int(BOARDWIDTH // 2) - int(TEMPLATEWIDTH // 2),
                     'y': 0, # start it above the self.board (i.e. less than 0)
                     'color': random.randint(0, len(COLORS)-1)}
         return newPiece

deep_q_network.py

@@ -107,10 +107,11 @@ def trainNetwork(s, readout, h_fc1, sess):
     sess.run(tf.initialize_all_variables())
     checkpoint = tf.train.get_checkpoint_state("saved_networks")
     if checkpoint and checkpoint.model_checkpoint_path:
+
         saver.restore(sess, checkpoint.model_checkpoint_path)
-        print "Successfully loaded:", checkpoint.model_checkpoint_path
+        print("Successfully loaded:", checkpoint.model_checkpoint_path)
     else:
-        print "Could not find old network weights"
+        print("Could not find old network weights")
 
     epsilon = INITIAL_EPSILON
     t = 0
@@ -185,7 +186,7 @@ def trainNetwork(s, readout, h_fc1, sess):
             state = "explore"
         else:
             state = "train"
-        print "TIMESTEP", t, "/ STATE", state, "/ LINES", game_state.total_lines, "/ EPSILON", epsilon, "/ ACTION", action_index, "/ REWARD", r_t, "/ Q_MAX %e" % np.max(readout_t)
+        print("TIMESTEP", t, "/ STATE", state, "/ LINES", game_state.total_lines, "/ EPSILON", epsilon, "/ ACTION", action_index, "/ REWARD", r_t, "/ Q_MAX %e" % np.max(readout_t))
 
         # write info to files
         '''

saved_networks/checkpoint

@@ -1,6 +1,6 @@
-model_checkpoint_path: "saved_networks/tetris-dqn-31660000"
-all_model_checkpoint_paths: "saved_networks/tetris-dqn-31620000"
-all_model_checkpoint_paths: "saved_networks/tetris-dqn-31630000"
-all_model_checkpoint_paths: "saved_networks/tetris-dqn-31640000"
-all_model_checkpoint_paths: "saved_networks/tetris-dqn-31650000"
-all_model_checkpoint_paths: "saved_networks/tetris-dqn-31660000"
+model_checkpoint_path: "tetris-dqn-31660000"
+all_model_checkpoint_paths: "tetris-dqn-31620000"
+all_model_checkpoint_paths: "tetris-dqn-31630000"
+all_model_checkpoint_paths: "tetris-dqn-31640000"
+all_model_checkpoint_paths: "tetris-dqn-31650000"
+all_model_checkpoint_paths: "tetris-dqn-31660000"