File size: 9,458 Bytes
eb34cec |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 |
import logging
import math
import numpy as np
# from no_one.NoOnePlayes import HumanPlayer
EPS = 1e-8
log = logging.getLogger(__name__)
class MCTS():
"""
This class handles the MCTS tree.
"""
def __init__(self, game, nnet, args):
self.game = game
self.nnet = nnet
self.args = args
self.Qsa = {} # stores Q values for s,a (as defined in the paper)
self.Nsa = {} # stores #times edge s,a was visited
self.Ns = {} # stores #times board s was visited
self.Ps = {} # stores initial policy (returned by neural net)
self.Es = {} # stores game.getGameEnded ended for board s
self.Vs = {} # stores game.getValidMoves for board s
def getActionProb(self, canonicalBoard, temp=1):
"""
This function performs numMCTSSims simulations of MCTS starting from
canonicalBoard.
Returns:
probs: a policy vector where the probability of the ith action is
proportional to Nsa[(s,a)]**(1./temp)
"""
for i in range(self.args.numMCTSSims):
# self.search(canonicalBoard)
self.game.reset_steps()
self.search(canonicalBoard)
s = self.game.stringRepresentation(canonicalBoard)
counts = [self.Nsa[(s, a)] if (s, a) in self.Nsa else 0 for a in range(self.game.getActionSize())]
if temp == 0:
bestAs = np.array(np.argwhere(counts == np.max(counts))).flatten()
bestA = np.random.choice(bestAs)
probs = [0] * len(counts)
probs[bestA] = 1
return probs
counts = [x ** (1. / temp) for x in counts]
counts_sum = float(sum(counts))
if counts_sum == 0:
print(len(counts))
probs = [x / counts_sum for x in counts]
return probs
def search_iterate(self, canonicalBoard):
stack = [(0, (canonicalBoard,))] # Stack of (state, depth, parent_index)
results = [] # To store the results of leaf or terminal nodes
while stack:
st, sv = stack.pop()
if st == 0:
result, ns = self.search_iterate_st0(sv[0])
if result is not None:
results.append(result)
if ns is not None:
stack.append((1, (ns[1], ns[2])))
stack.append((0, (ns[0],)))
elif st == 1:
v = results.pop()
v = self.search_iterate_update(v, sv[0], sv[1])
results.append(v)
else:
raise ValueError("Invalid state")
return results.pop()
def search_iterate_st0(self, canonicalBoard):
s = self.game.stringRepresentation(canonicalBoard)
if s not in self.Es:
self.Es[s] = self.game.getGameEnded(canonicalBoard, 1)
if self.Es[s] != 0:
result = -self.Es[s]
return result, None
if s not in self.Ps:
# leaf node
self.Ps[s], v = self.nnet.predict(canonicalBoard)
valids = self.game.getValidMoves(canonicalBoard, 1)
self.Ps[s] = self.Ps[s] * valids
sum_Ps_s = np.sum(self.Ps[s])
if sum_Ps_s > 0:
self.Ps[s] /= sum_Ps_s
else:
self.Ps[s] = self.Ps[s] + valids
self.Ps[s] /= np.sum(self.Ps[s])
self.Vs[s] = valids
self.Ns[s] = 0
return -v, None
valids = self.Vs[s]
cur_best = -float('inf')
best_act = -1
for a in range(self.game.getActionSize()):
if valids[a]:
if (s, a) in self.Qsa:
u = self.Qsa[(s, a)] + self.args.cpuct * self.Ps[s][a] * math.sqrt(self.Ns[s]) / (1 + self.Nsa[(s, a)])
else:
u = self.args.cpuct * self.Ps[s][a] * math.sqrt(self.Ns[s] + EPS)
if u > cur_best:
cur_best = u
best_act = a
next_s, next_player = self.game.getNextState(canonicalBoard, 1, best_act)
next_s = self.game.getCanonicalForm(next_s, next_player)
return None, (next_s, s, best_act)
# index = len(results) # Current index in results
# stack.append((next_s, depth + 1, index))
# # Backpropagate results
# for v, parent_index in reversed(results):
# if parent_index is not None:
# parent_v, _ = results[parent_index]
# results[parent_index] = ((parent_v * self.Ns[s] + v) / (self.Ns[s] + 1), _)
# self.Ns[s] += 1
# # Update Qsa and Nsa based on backpropagation
# for i, (v, parent_index) in enumerate(results):
# if parent_index is not None: # Ignore root
# _, action = stack[i] # Assuming we also pushed actions to stack
# if (s, action) in self.Qsa:
# self.Qsa[(s, action)] = (self.Nsa[(s, action)] * self.Qsa[(s, action)] + v) / (self.Nsa[(s, action)] + 1)
# self.Nsa[(s, action)] += 1
# else:
# self.Qsa[(s, action)] = v
# self.Nsa[(s, action)] = 1
# return -results[0][0] # Return the negated value of the root node
def search_iterate_update(self, v, s, a):
if (s, a) in self.Qsa:
self.Qsa[(s, a)] = (self.Nsa[(s, a)] * self.Qsa[(s, a)] + v) / (self.Nsa[(s, a)] + 1)
self.Nsa[(s, a)] += 1
else:
self.Qsa[(s, a)] = v
self.Nsa[(s, a)] = 1
self.Ns[s] += 1
return -v
def search(self, canonicalBoard, depth=0):
"""
This function performs one iteration of MCTS. It is recursively called
till a leaf node is found. The action chosen at each node is one that
has the maximum upper confidence bound as in the paper.
Once a leaf node is found, the neural network is called to return an
initial policy P and a value v for the state. This value is propagated
up the search path. In case the leaf node is a terminal state, the
outcome is propagated up the search path. The values of Ns, Nsa, Qsa are
updated.
NOTE: the return values are the negative of the value of the current
state. This is done since v is in [-1,1] and if v is the value of a
state for the current player, then its value is -v for the other player.
Returns:
v: the negative of the value of the current canonicalBoard
"""
s = self.game.stringRepresentation(canonicalBoard)
if s not in self.Es:
self.Es[s] = self.game.getGameEnded(canonicalBoard, 1)
if self.Es[s] != 0:
# terminal node
return -self.Es[s]
if s not in self.Ps:
# leaf node
self.Ps[s], v = self.nnet.predict(canonicalBoard)
valids = self.game.getValidMoves(canonicalBoard, 1)
self.Ps[s] = self.Ps[s] * valids # masking invalid moves
sum_Ps_s = np.sum(self.Ps[s])
if sum_Ps_s > 0:
self.Ps[s] /= sum_Ps_s # renormalize
else:
# if all valid moves were masked make all valid moves equally probable
# NB! All valid moves may be masked if either your NNet architecture is insufficient or you've get overfitting or something else.
# If you have got dozens or hundreds of these messages you should pay attention to your NNet and/or training process.
log.error("All valid moves were masked, doing a workaround.")
self.Ps[s] = self.Ps[s] + valids
self.Ps[s] /= np.sum(self.Ps[s])
self.Vs[s] = valids
self.Ns[s] = 0
return -v
valids = self.Vs[s]
cur_best = -float('inf')
best_act = -1
# pick the action with the highest upper confidence bound
for a in range(self.game.getActionSize()):
if valids[a]:
if (s, a) in self.Qsa:
u = self.Qsa[(s, a)] + self.args.cpuct * self.Ps[s][a] * math.sqrt(self.Ns[s]) / (
1 + self.Nsa[(s, a)])
else:
u = self.args.cpuct * self.Ps[s][a] * math.sqrt(self.Ns[s] + EPS) # Q = 0 ?
if u > cur_best:
cur_best = u
best_act = a
a = best_act
if depth > 100:
candidates = self.game.getValidMoves(canonicalBoard, 1)
a = np.random.choice([i for i in range(len(candidates)) if candidates[i] == 1])
# self.game.display(canonicalBoard)
# human_player = HumanPlayer(self.game)
# a = human_player.play(canonicalBoard)
depth = 80
next_s, next_player = self.game.getNextState(canonicalBoard, 1, a)
next_s = self.game.getCanonicalForm(next_s, next_player)
# print("*", end="")
# self.game.display(next_s)
v = self.search(next_s, depth=depth + 1)
if (s, a) in self.Qsa:
self.Qsa[(s, a)] = (self.Nsa[(s, a)] * self.Qsa[(s, a)] + v) / (self.Nsa[(s, a)] + 1)
self.Nsa[(s, a)] += 1
else:
self.Qsa[(s, a)] = v
self.Nsa[(s, a)] = 1
self.Ns[s] += 1
return -v
|