Chess-AI-Python/chess_ai.py
2023-03-04 18:42:17 +01:00

414 lines
13 KiB
Python

from chess_tools import get_possible_moves, move_piece, init_board, get_time
import numpy as np
import time
import random
inf = float("inf")
score_table = {"Q" : 9, "R" : 5, "B" : 3.3, "N" : 3.2, "P" : 1, "K" : 10**5 ,"q" : -9, "r" : -5, "b" : -3, "n" : -3, "p" : -1, "k" : -10**5, '':0, '.':0}
count = 0
class Heuristics:
# The tables denote the points scored for the position of the chess pieces on the board.
# Source : https://www.chessprogramming.org/Simplified_Evaluation_Function
PAWN_TABLE = [
[ 0, 0, 0, 0, 0, 0, 0, 0],
[ 5, 10, 10,-20,-20, 10, 10, 5],
[ 5, -5,-10, 0, 0,-10, -5, 5],
[ 0, 0, 0, 20, 20, 0, 0, 0],
[ 5, 5, 10, 25, 25, 10, 5, 5],
[10, 10, 20, 30, 30, 20, 10, 10],
[50, 50, 50, 50, 50, 50, 50, 50],
[ 0, 0, 0, 0, 0, 0, 0, 0]
]
KNIGHT_TABLE = [
[-50, -40, -30, -30, -30, -30, -40, -50],
[-40, -20, 0, 5, 5, 0, -20, -40],
[-30, 5, 10, 15, 15, 10, 5, -30],
[-30, 0, 15, 20, 20, 15, 0, -30],
[-30, 5, 15, 20, 20, 15, 0, -30],
[-30, 0, 10, 15, 15, 10, 0, -30],
[-40, -20, 0, 0, 0, 0, -20, -40],
[-50, -40, -30, -30, -30, -30, -40, -50]
]
BISHOP_TABLE = [
[-20, -10, -10, -10, -10, -10, -10, -20],
[-10, 5, 0, 0, 0, 0, 5, -10],
[-10, 10, 10, 10, 10, 10, 10, -10],
[-10, 0, 10, 10, 10, 10, 0, -10],
[-10, 5, 5, 10, 10, 5, 5, -10],
[-10, 0, 5, 10, 10, 5, 0, -10],
[-10, 0, 0, 0, 0, 0, 0, -10],
[-20, -10, -10, -10, -10, -10, -10, -20]
]
ROOK_TABLE = [
[ 0, 0, 0, 5, 5, 0, 0, 0],
[-5, 0, 0, 0, 0, 0, 0, -5],
[-5, 0, 0, 0, 0, 0, 0, -5],
[-5, 0, 0, 0, 0, 0, 0, -5],
[-5, 0, 0, 0, 0, 0, 0, -5],
[-5, 0, 0, 0, 0, 0, 0, -5],
[ 5, 10, 10, 10, 10, 10, 10, 5],
[ 0, 0, 0, 0, 0, 0, 0, 0]
]
QUEEN_TABLE = [
[-20, -10, -10, -5, -5, -10, -10, -20],
[-10, 0, 0, 0, 0, 0, 0, -10],
[-10, 0, 0, 0, 0, 0, 0, -10],
[ 0, 0, 0, 0, 0, 0, 0, -5],
[ -5, 0, 0, 0, 0, 0, 0, -5],
[-10, 0, 0, 0, 0, 0, 0, -10],
[-10, 0, 0, 0, 0, 0, 0, -10],
[-20, -10, -10, -5, -5, -10, -10, -20]
]
KING_TABLE_MG = [
[ 20, 50, 10, 0, 0, 10, 10, 20],
[ 20, 20, 0, 0, 0, 0, 20, 20],
[-10, -20, -20, -20, -20, -20, -20, -10],
[-20, -30, -30, -40, -40, -30, -30, -20],
[-30, -40, -40, -50, -50, -40, -40, -30],
[-30, -40, -40, -50, -50, -40, -40, -30],
[-30, -40, -40, -50, -50, -40, -40, -30],
[-30, -40, -40, -50, -50, -40, -40, -30]
]
KING_TABLE_EG = [
[-50, -30, -30, -30, -30, -30, -30, -50],
[-30, -30, 0, 0, 0, 0, -30, -30],
[-30, -10, 20, 30, 30, 20, -10, -30],
[-30, -10, 30, 40, 40, 30, -10, -30],
[-30, -10, 30, 40, 40, 30, -10, -30],
[-30, -10, 20, 30, 30, 20, -10, -30],
[-30, -20, -10, 0, 0, -10, -20, -30],
[-50, -40, -30, -20, -20, -30, -40, -50]
]
def get_position_score(start, board):
"""Evaluate the position of a piece on the board."""
global game_phase
piece = board[start]
if piece == 'P':
return Heuristics.PAWN_TABLE[start[0]][start[1]]
if piece == 'p':
return -Heuristics.PAWN_TABLE[7 - start[0]][start[1]]
if piece == 'N':
return Heuristics.KNIGHT_TABLE[start[0]][start[1]]
if piece == 'n':
return -Heuristics.KNIGHT_TABLE[7 - start[0]][start[1]]
if piece == 'B':
return Heuristics.BISHOP_TABLE[start[0]][start[1]]
if piece == 'b':
return -Heuristics.BISHOP_TABLE[7 - start[0]][start[1]]
if piece == 'R':
return Heuristics.ROOK_TABLE[start[0]][start[1]]
if piece == 'r':
return -Heuristics.ROOK_TABLE[7 - start[0]][start[1]]
if piece == 'Q':
return Heuristics.QUEEN_TABLE[start[0]][start[1]]
if piece == 'q':
return -Heuristics.QUEEN_TABLE[7 - start[0]][start[1]]
if piece == 'K':
if game_phase < 45:
return Heuristics.KING_TABLE_EG[start[0]][start[1]]
else:
return Heuristics.KING_TABLE_MG[start[0]][start[1]]
if piece == 'k':
if game_phase < 45:
return -Heuristics.KING_TABLE_EG[7 - start[0]][start[1]]
else:
return -Heuristics.KING_TABLE_MG[7 - start[0]][start[1]]
return 0
def get_king_safety(start, board):
"""King is safe if protected by a wall of pawns"""
x, y = start
piece = board[x, y]
if piece == "K":
king_safety = (board[max(x-1, 0), y+1] == "P") + (board[x, y+1] == "P") + (board[min(x+1, 7), y+1] == "P")
elif piece == "k":
king_safety = (board[max(x-1, 0), y-1] == "p") + (board[x, y-1] == "p") + (board[min(x+1, 7), y-1] == "p")
king_safety = -king_safety
else:
king_safety = 0
return king_safety
def get_pawn_defend(start, board):
"""A good pawn is a pawn that defend another piece"""
x, y = start
piece = board[x, y]
if piece == "P" and y+1<=7:
if x-1 >= 0:
if x+1<=7:
pawn_defend = board[x-1, y+1].isupper() + board[x+1, y+1].isupper()
else:
pawn_defend = board[x-1, y+1].isupper()
else:
pawn_defend = board[x+1, y+1].isupper()
elif piece == "p" and y-1>=0:
if x-1 >= 0:
if x+1<=7:
pawn_defend = board[x-1, y-1].islower() + board[x+1, y-1].islower()
else:
pawn_defend = board[x-1, y-1].islower()
else:
pawn_defend = board[x+1, y-1].islower()
else:
pawn_defend = 0
return pawn_defend
def get_rook_score(start, board):
"""A good rook is a rook that can move in a straight line"""
x, y = start
piece = board[x, y]
rook_score = 0
if piece == "R" or piece == "r":
for i in range(1, 8):
if x-i >= 0:
if board[x-i, y] == ".":
rook_score += 1
else:
break
else:
break
for i in range(1, 8):
if x+i <= 7:
if board[x+i, y] == ".":
rook_score += 1
else:
break
else:
break
for i in range(1, 8):
if y-i >= 0:
if board[x, y-i] == ".":
rook_score += 2
else:
break
else:
break
for i in range(1, 8):
if y+i <= 7:
if board[x, y+i] == ".":
rook_score += 2
else:
break
else:
break
return rook_score
def get_other_eval(start, board):
"""Evaluation not based on the conventional chess pieces values and on the position"""
king_safety = get_king_safety(start, board)
pawn_defend = get_pawn_defend(start, board)
rook_score = get_rook_score(start, board)
return king_safety*10 + pawn_defend*10 + rook_score*2
def leaf_eval(board):
"""Evaluate a certain board position"""
global count
count += 1
#evalute using score_table, position score and other eval
evaluate = lambda x,y: score_table[board[x, y]] * 100 + get_position_score((x, y), board) + get_other_eval((x, y), board)
return sum(evaluate(x, y) for y in range(8) for x in range(8)) + random.random()*10
game_phase = 0
def get_game_phase(board):
global game_phase
game_phase = sum(abs(score_table[board[x, y]]) for y in range(8) for x in range(8)) - 2 * score_table['K']
def minmax(node, depth, alpha=-inf, beta=inf):
is_maximizing = node.color == 'white'
if depth == 0 or node.is_leaf:
return node.score
if is_maximizing:
score = -inf
for child in node.children:
score = max(score, minmax(child, depth - 1, alpha, beta))
alpha = max(alpha, score)
if beta <= alpha:
break
return score
else:
score = inf
for child in node.children:
score = min(score, minmax(child, depth - 1, alpha, beta))
beta = min(beta, score)
if beta <= alpha:
break
return score
def evaluate_move(start, end, board):
"""Evaluate a move. Just used in "create_and_evaluate_tree" to sort the moves to optimize alpha-beta pruning"""
position_score = get_position_score(start, board)
take_score = score_table[board[end]]*100
return position_score + take_score
class Node:
__slots__ = 'board', 'move', 'parent', 'children', 'score', 'depth', 'color', 'board_score'
def __init__(self, board, move, parent, children, score, depth, color):
self.board = board
self.move = move
self.parent = parent
self.children = children
self.score = score
self.depth = depth
self.color = color
@property
def is_leaf(self):
return not self.children
@property
def is_root(self):
return self.parent is None
def evalute_tree(self):
if self.is_leaf:
self.score = leaf_eval(self.board)
else:
for child in self.children:
child.evalute_tree()
self.score = minmax(self, self.depth)
def create_and_evaluate_tree(self, depth, alpha=-inf, beta=inf):
if depth == 0:
self.score = leaf_eval(self.board)
else:
movable_pieces = ((x, y) for x in range(8) for y in range(8) if
self.board[x, y] != '.' and self.color == 'white' and self.board[x, y].isupper() or self.color == 'black' and self.board[x, y].islower())
for piece in movable_pieces:
possible_moves = get_possible_moves(piece, self.board)
possible_moves.sort(key=lambda x: abs(evaluate_move(piece, x, self.board)), reverse=True)
#possible_moves.sort(key=lambda x: evaluate_move(piece, x, self.board), reverse= self.color == "white")
for move in possible_moves:
new_board = move_piece(piece, move, self.board)
new_move = (piece, move)
new_node = Node(new_board, new_move, self, [], 0, self.depth - 1, 'white' if self.color == 'black' else 'black')
self.children.append(new_node)
if self.color == 'white':
alpha = max(alpha, new_node.create_and_evaluate_tree(depth - 1, alpha, beta))
if beta <= alpha:
break
#To stop exploring if one king is taken
if abs(self.score) > 10**4:
break
else:
beta = min(beta, new_node.create_and_evaluate_tree(depth - 1, alpha, beta))
if beta <= alpha:
break
if abs(self.score) > 10**4:
break
if not self.children:
self.score = leaf_eval(self.board)
else:
self.score = max(child.score for child in self.children if child.score is not None) if self.color == 'white' else min(child.score for child in self.children if child.score is not None)
return self.score
def get_best_move(self):
self.evalute_tree()
maximazing = self.color == 'white'
if maximazing:
maxi = max(self.children, key=lambda x: x.score)
print(f"maxi: {maxi.score}")
return maxi.move
else:
mini = min(self.children, key=lambda x: x.score)
print(f"mini: {mini.score}")
return mini.move
def pretty(self):
string = f"Node: {self.move} {self.score} {self.depth} {self.color} \n"
for child in self.children:
string += "| " + child.pretty() + "\n"
return string
def __repr__(self):
return self.pretty()
get_deeper = 0
def get_best_move(board, color, depth):
global count, get_deeper
get_game_phase(board)
print(f"End game: {game_phase}")
depth += get_deeper
print(f"curent depth: {depth}")
a = time.time()
root = Node(board, None, None, [], 0, 0, color)
root.create_and_evaluate_tree(depth)
best_move = root.get_best_move()
exec_time = time.time() - a
print(f"Tree evaluated in {time.time() - a} seconds")
print(f"Number of leafs: {count}")
if exec_time < 1.5:
get_deeper += 1
print("Getting deeper, current depth: ", depth)
elif exec_time > 20:
get_deeper += -1
print("Getting shallower, current depth: ", depth)
print(f"Mean number of children : {count**(1/depth)}")
count = 0
return best_move
if __name__ == '__main__':
board = init_board()
b = get_best_move(board, 'white', 2)
print(b)