def normal_play():
players = []
players = [LudoPlayerRandom() for _ in range(3)]
epsilon = 0.05 #
discount_factor = 0.5 #
learning_rate = 0.25 #
parameters = [epsilon, discount_factor, learning_rate]
t1 = LudoPlayerQLearningAction(parameters, chosenPolicy="greedy", QtableName='QTable', RewardName='Reward')
players.append(t1)
for i, player in enumerate(players):
player.id = i # selv tildele atributter uden defineret i klassen
score = [0, 0, 0, 0]
n = 1000
start_time = time.time()
tqdm_1 = tqdm(range(n), ascii=True)
for i in tqdm_1:
tqdm_1.set_description_str(f"win rates {np.around(score/np.sum(score),decimals=2)*100}")
random.shuffle(players)
ludoGame = LudoGame(players)
winner = ludoGame.play_full_game()
score[players[winner].id] += 1
for player in players: # Saving reward for QLearning player
if type(player)==LudoPlayerQLearningAction:
player.append_reward()
for player in players:
if type(player)==LudoPlayerQLearningAction:
player.saveQTable()
player.saveReward()
duration = time.time() - start_time
print('win distribution percentage', (score/np.sum(score))*100)
print('win distribution:', score)
from pyludo import LudoGame
from pyludo.StandardLudoPlayers import LudoPlayerRandom, LudoPlayerFast, LudoPlayerAggressive, LudoPlayerDefensive
import random
import time
players = [
LudoPlayerRandom(),
LudoPlayerFast(),
LudoPlayerAggressive(),
LudoPlayerDefensive(),
]
scores = {}
for player in players:
scores[player.name] = 0
n = 1000
start_time = time.time()
for i in range(n):
random.shuffle(players)
ludoGame = LudoGame(players)
winner = ludoGame.play_full_game()
scores[players[winner].name] += 1
print('Game ', i, ' done')
duration = time.time() - start_time
print('win distribution:', scores)
print('games per second:', n / duration)
game_count = 2500
sim_chromo = np.load(
'genetic_algorithm/populations/Simple/Simple_Cellular-20-10_Blend-0.5_NStep/1000.pop.winner.npy'
)
adv_chromo = np.load(
'genetic_algorithm/populations/Advanced/Advanced_Normal-100-10_Whole_Normal-0.1/1000.pop.winner.npy'
)
full_chromo = np.load(
'genetic_algorithm/populations/Full/Full_Cellular-100-10_Whole_Normal-0.1/1000.pop.winner.npy'
)
sim_player = GAPlayerSimple(sim_chromo)
adv_player = GAPlayerAdvanced(adv_chromo)
full_player = GAPlayerFull(full_chromo)
rand_player = LudoPlayerRandom()
players = [sim_player, adv_player, full_player, rand_player]
f = open('genetic_algorithm/agent_evaluations/prediction_times.txt', 'w')
tournament_player_ids = {}
for i, player in enumerate(players):
tournament_player_ids[player] = i
print(f'{player.name} with id {i}')
f.write(f'{player.name} with id {i}\n')
pred_times_sim = []
pred_times_adv = []
pred_times_full = []
# for l in learning_rate:
# p = multiprocessing.Process(target=play_with_on_QLearning_thread, args=(10000, e, d, l))
# multiprocess.append(p)
# p.start()
# for index, process in enumerate(multiprocess):
# process.join()
# print('That took {} seconds'.format(time.time() - starttime))
####################################################################################################################################################
### SINGLE GAME TEST ###
####################################################################################################################################################
players = [LudoPlayerRandom() for _ in range(4)]
# players.append(LudoPlayerQLearning("epsilon greedy", QtableName='QTable', RewardName='Reward', epsilon=0.1, discount_factor=0.5, learning_rate=0.1))
for i, player in enumerate(players):
player.id = i # selv tildele atributter uden defineret i klassen
score = [0, 0, 0, 0]
n = 100
start_time = time.time()
for i in range(n):
random.shuffle(players)
ludoGame = LudoGame(players)
# for player in players: # Saving reward for QLearning player
# if player.id == 3: