API tools faq
paste
Guest User

tictactoe

a guest
Apr 15th, 2024
125
0
Never
Add comment
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
Python 8.05 KB | None | 0 0
raw download clone embed print report
  1. import math, random, pickle, numpy as np
  2.  
  3.  
  4.  
  5. class boardFunctions:
  6.     def __init__(self, boardState):
  7.         # Current state of the board
  8.         self.boardState = boardState
  9.  
  10.     # Return 1/-1 if player has won, return 0 if draw, return 2 if game ongoing
  11.     def checkWin(self, board) :
  12.         winIndices =  [[0,3,6], [1,4,7], [2,5,8], [0,1,2], [3,4,5], [6,7,8], [0,4,8], [2,4,6]]
  13.         for element in winIndices :
  14.             winCheckSum = (board[element[0]] + board[element[1]] + board[element[2]]) / 3
  15.             if abs(winCheckSum) == 1 :
  16.                 return winCheckSum
  17.         if 0 in board :
  18.             return 2
  19.         else :
  20.             return 0
  21.    
  22.     # Prints a  certain boardstate
  23.     def print(self, board) :
  24.         boardDisp = [' '] * 9
  25.         for i in range(len(board)):
  26.             if board[i] == 1 :
  27.                 boardDisp[i] = 'X'
  28.             elif board[i] == -1 :
  29.                 boardDisp[i] = 'O'
  30.             i += 1
  31.         print('-----')
  32.         print(boardDisp[0], boardDisp[1], boardDisp[2])
  33.         print(boardDisp[3], boardDisp[4], boardDisp[5])
  34.         print(boardDisp[6], boardDisp[7], boardDisp[8])
  35.         print('-----')
  36.    
  37.     # Helper func for imputting indices as a row and column
  38.     def rowColumn(self, row, column) :
  39.         i = (3 * row) + column
  40.         return i
  41.  
  42.     # Prompts the user to imput a move and updates the specified board
  43.     def playerInput(self, board, player) :
  44.         column = int(input('Column? '))
  45.         row = int(input('Row? '))
  46.         space = self.rowColumn(row, column)
  47.         if board[space] == 0 :
  48.             board[space] = player
  49.         else:
  50.             if self.checkWin(board) == 0 :
  51.                 print('Square already filled. Try again')
  52.                 self.playerInput(player)
  53.             else :
  54.                 return
  55.    
  56.     # Plays a move on the specified board in a random empty square
  57.     def randomMove(self, board, player) :
  58.         board[random.choice([idx for idx, val in enumerate(board) if 0 == val])] = player
  59. bfunc = boardFunctions([0] * 9)
  60.  
  61. class minimax :
  62.     def __init__(self) :
  63.         self.minimaxSeed = 4
  64.    
  65.     def moveSearch(self, board, player, depth) :
  66.         winState = bfunc.checkWin(bfunc.boardState)
  67.         if abs(winState) == 1 :
  68.             if depth == 0 :
  69.                 return
  70.             else :
  71.                 return winState * player * -1
  72.         elif winState == 0:
  73.             if depth == 0 :
  74.                 return
  75.             else :
  76.                 return 0
  77.         elif winState == 2 :
  78.             moveList = []
  79.             i = 0
  80.             while i < 9 :
  81.                 if board[i] == 0 :
  82.                     moveList.append(i)
  83.                 i += 1
  84.             j = 0
  85.             valueList = [0] * len(moveList)
  86.             while j < len(moveList) :
  87.                 board[moveList[j]] = player
  88.                 valueList[j] = self.moveSearch(board, -1 * player, depth + 1)
  89.                 board[moveList[j]] = 0
  90.                 j += 1
  91.             if depth == 0 :
  92.                 maxValue = max(valueList)
  93.                 maxIndices = [idx for idx, val in enumerate(valueList) if maxValue == val]
  94.                 boardStateInt = sum([(2**j)*abs(bfunc.boardState[j]) + 1 + self.minimaxSeed for j in range(len(valueList))])
  95.                 boardStateSeed = int(((100*abs(math.cos(boardStateInt)))%1)*100) % len(maxIndices)
  96.                 return moveList[maxIndices[boardStateSeed]]
  97.             else :
  98.                 return max(valueList) * -1
  99.    
  100.     def minimaxMove(self, board, player) :
  101.         board[self.moveSearch(board, player, 0)] = player
  102.    
  103.     def minimaxGame(self, startingPlayer) :
  104.         i = 0
  105.         while bfunc.checkWin(bfunc.boardState) == 2 :
  106.             self.minimaxMove(bfunc.boardState, startingPlayer * ((-1) ** i))
  107.             bfunc.print(bfunc.boardState)
  108.             i += 1
  109. mmx = minimax()
  110.  
  111. class pickler:
  112.     def __init__(self, path) :
  113.         self.picklePath = path
  114.    
  115.     def encode(self, data, fileName):
  116.         with open(self.picklePath + '/' + fileName + '.pickle', 'wb') as f:
  117.             pickle.dump(data, f)
  118.    
  119.     def decode(self, fileName) :
  120.         with open(self.picklePath + '/' + fileName + '.pickle', 'rb') as f :
  121.             return pickle.load(f)
  122. pkl = pickler('C:/Users/riley/Documents/PythonProjects/Neural/pickles/')
  123.  
  124. class neuralNetwork:
  125.     def __init__(self) :
  126.         pass # eventually will put all neural network related functions inside this class
  127.  
  128. def sigmoid(x) :
  129.     return 1/(1 + np.exp(-x))  
  130.  
  131. def derivativeSigmoid(x) :
  132.     return sigmoid(x) * (1 - sigmoid(x))
  133.  
  134. def inverseSigmoid(x) :
  135.     return np.log(x / (1 - x))
  136.  
  137. def sigmoidArray(array) :
  138.     arr = np.copy(array)
  139.     arr = [[sigmoid(j) for j in i] for i in arr]
  140.     arr = np.reshape(arr,array.shape)
  141.     return arr
  142.  
  143. layerSizes = [10,20,9]
  144. neuralPlayer = 1
  145.  
  146. weightMatrices = pkl.decode('weights')
  147. biasMatrices = pkl.decode('biases')
  148.  
  149.  
  150.  
  151. def neuralValuePass(board, player) :
  152.     layers = [np.concatenate(([player], board)).reshape((layerSizes[0],1))]
  153.     for i in range(len(weightMatrices)) :
  154.         layers.append(sigmoidArray(np.matmul(weightMatrices[i], layers[i]) + biasMatrices[i]))
  155.     return layers
  156.  
  157. def neuralMoveSelect(board, player) :
  158.     moveSelection = np.argmax(neuralValuePass(board,player)[-1])
  159.     return moveSelection
  160.  
  161. def neuralMove(board, player) :
  162.     move = neuralMoveSelect(board, player)
  163.     if board[move] == 0 :
  164.         board[move] = player
  165.     else :
  166.         return
  167.  
  168. def cost(board, player) :
  169.     nnwValues = neuralValuePass(board, player)[-1].reshape(9)
  170.     mmxValues = np.zeros(9)
  171.     mmxValues[mmx.moveSearch(board, player, 0)] = 1
  172.     costValues = [(mmxValues[i] - nnwValues[i]) ** 2 for i in range(9)]
  173.     return sum(costValues)
  174.  
  175. layers = [i.reshape((i.shape)[0]) for i in neuralValuePass(bfunc.boardState, 1)]
  176.  
  177. mmxValues = np.zeros(9)
  178. mmxValues[mmx.moveSearch(bfunc.boardState, neuralPlayer, 0)] = 1
  179.  
  180. def layerGradients(layers) :
  181.     layerGradients = [0] * len(weightMatrices)
  182.     layerGradients[-1] = [2 * (layers[-1][i] - mmxValues[i]) for i in range(len(layers[-1]))]
  183.     layerGradients[-2] = [sum([(weightMatrices[-1][j][i]) * (derivativeSigmoid(inverseSigmoid(layers[-1][j]))) * (2 * (layers[-1][j] - mmxValues[j])) for j in range(len(layers[-1]))]) for i in range(len(layers[-2]))]
  184.     return layerGradients
  185.  
  186. def weightGradients(weights) :
  187.     weightGradients = [0] * len(weights)
  188.     weightGradients[-1] = np.copy(weights[-1])
  189.     weightGradients[-1] = [[(layers[-2][j]) * (derivativeSigmoid(inverseSigmoid(layers[-1][i]))) * (2 * (layers[-1][i] - mmxValues[i])) for j in range(len(weightGradients[-1][i]))] for i in range(len(weightGradients[-1]))]
  190.     weightGradients[-1] = np.reshape(weightGradients[-1],weights[-1].shape)
  191.  
  192.     weightGradients[-2] = np.copy(weights[-2])
  193.     weightGradients[-2] = [[(layers[-3][j]) * (derivativeSigmoid(inverseSigmoid(layers[-2][i]))) * (layerGradients(layers)[-2][j]) for j in range(len(weightGradients[-2][i]))] for i in range(len(weightGradients[-2]))]
  194.     weightGradients[-2] = np.reshape(weightGradients[-2],weights[-2].shape)
  195.     return weightGradients
  196.  
  197. def biasGradients(biases) :
  198.     biasGradients = [0] * len(biases)
  199.     biasGradients[-1] = np.copy(biases[-1])
  200.     biasGradients[-1] = [[(1) * (derivativeSigmoid(inverseSigmoid(layers[-1][i]))) * (2 * (layers[-1][i] - mmxValues[i])) for j in range(len(biasGradients[-1][i]))] for i in range(len(biasGradients[-1]))]
  201.     biasGradients[-1] = np.reshape(biasGradients[-1],biases[-1].shape)
  202.  
  203.     biasGradients[-2] = np.copy(biases[-2])
  204.     biasGradients[-2] = [[(1) * (derivativeSigmoid(inverseSigmoid(layers[-2][i]))) * (layerGradients(layers)[-2][j]) for j in range(len(biasGradients[-2][i]))] for i in range(len(biasGradients[-2]))]
  205.     biasGradients[-2] = np.reshape(biasGradients[-2],biases[-2].shape)
  206.     return biasGradients
  207.  
  208. layerGradients(layers)
  209. weightGradients(weightMatrices)
  210. biasGradients(biasMatrices)
  211.  
  212.  
  213.  
  214. '''
  215. To-Do List:
  216. - Weight/bias shape check
  217. - Gradient Descent
  218. - Training
  219. - Major clean up/refactor
  220. - GUI???'''
  221.  
Advertisement
Add Comment
Please, Sign In to add comment
Public Pastes
We use cookies for various purposes including analytics. By continuing to use Pastebin, you agree to our use of cookies as described in the Cookies Policy.  OK, I Understand
Not a member of Pastebin yet?
Sign Up, it unlocks many cool features!