from agents.minimax.minimax import RandomPlayer
data_saver = Save_data()
minimax = RandomPlayer(mark='X')
q_player = DqnAgent(mark='O')
q_table = Save_data().load_data(path='q_player_QS.p')
if q_table is not None:
q_player.Q = q_table
print("The qlearn agent has loaded the qtable")
EPISODES = 100_000
sys.setrecursionlimit(10000)
game = NtxuvaGame(master=tk.Tk(), player1=minimax, player2=q_player)
statistical_data = {}
statistical_data_minimax = {}
minimax_win = 0
q_player_win = 0
for episode in range(EPISODES):
print(f"EPISODE --- {episode} --- ")
try:
winner = game.play()
# statistics for q_player
statistical_data[episode] = [
f"bad-moves: {game.player2.bad_moves}",
# Important classes
from game.ntxuva_ttt_game import NtxuvaGame
from extra_tools.save_data import Save_data
from timeit import default_timer as timer
from agents.qplayerandothers.players import QPlayer
from agents.dqn.dqn import DqnAgent
start = timer()
EPISODES = 10
data_saver = Save_data(filename='dqn_vs_qlearn')
game = NtxuvaGame(player1=DqnAgent(mark='X', rating=1000), player2=QPlayer(mark='O', rating=1000))
for episode in range(EPISODES):
print(f" -- {episode} -- ")
winner = game.play()
if winner is not None:
winner.wins = winner.wins + 1
game.reset()
data_saver.save(players={1: game.player1,
2: game.player2})
end = timer()
# Defining episodes
EPISODES = 100_000
# Initializing the players and trained data
# The Random player is initialized here
minimax_q = MinimaxQPlayer(mark='O', numActionsA=16, numActionsB=16)
dqn_agent = DqnAgent(mark='X')
Q = minimax_q.Q
V = minimax_q.V
pi = minimax_q.pi
# Initializing the game environment
game = NtxuvaGame(player1=dqn_agent, player2=minimax_q)
sys.setrecursionlimit(10000)
FILENAME = "dqn_vs_minimaxq.p"
# Initializing the statistical data variables
statistical_data = {}
dqn_win = 0
minimax_q_win = 0
# Inidializing the statistical data Object named data_saver
data_saver = Save_data()
start = timer()
# Starting the trainment from episode 0 to the last Episode
episode = 0
from extra_tools.save_data import Save_data
from timeit import default_timer as timer
# Defining episodes
EPISODES = 5_000
# Initializing the players and trainded data
# The Random player is initialized here
minimax_q = MinimaxQPlayer(mark='O', numActionsA=16, numActionsB=16)
minimax = RandomPlayer(mark='X', prof=1)
Q = minimax_q.Q
V = minimax_q.V
pi = minimax_q.pi
# Initializing the game environment
game = NtxuvaGame(player1=minimax, player2=minimax_q)
sys.setrecursionlimit(10000)
FILENAME = "minimax_vs_minimaxq"
# Initializing the statistical data variables
statistical_data = {}
minimax_win = 0
minimax_q_win = 0
# Inidializing the statistical data Object named data_saver
data_saver = Save_data()
start = timer()
# Starting the trainment from episode 0 to the last Episode
episode = 0
import tkinter as tk
import sys
import pickle as pickle # cPickle is for Python 2.x only; in Python 3, simply "import pickle" and the accelerated version will be used automatically if available
from game.ntxuva_ttt_game import NtxuvaGame, DqnAgent
tk.wantobjects = False
root = tk.Tk()
epsilon = 0.9
player1 = DqnAgent(mark="X", epsilon=epsilon)
player2 = DqnAgent(mark="O", epsilon=epsilon)
game = NtxuvaGame(root, player1, player2)
sys.setrecursionlimit(10000)
N_episodes = 45000
for episodes in range(N_episodes):
print('Episode ', episodes)
try:
game.play()
except RecursionError:
Q = game.Q
filename = "Q_table_ntxuva_dictionary_%s.p" % episodes
pickle.dump(Q, open(filename, "wb"))
except MemoryError:
Q = game.Q
filename = "Q_table_ntxuva_dictionary_%s.p" % episodes
pickle.dump(Q, open(filename, "wb"))
game.reset()
# Important classes
from timeit import default_timer as timer
from game.ntxuva_ttt_game import NtxuvaGame
from extra_tools.save_data import Save_data
from extra_tools.elo_rating import EloRater
from agents.qplayerandothers.players import QPlayer, RandomPlayer
start = timer()
EPISODES = 100_000
game = NtxuvaGame(player1=RandomPlayer(mark='X'), player2=QPlayer(mark='O'))
data_saver = Save_data(player1=game.player1,
player2=game.player2,
filename=f'qlearn_vs_random_{EPISODES}')
p1_rating = data_saver.get_elo(game.player1.player_name)
p2_rating = data_saver.get_elo(game.player2.player_name)
elo = EloRater(game.player1, game.player2, K=1)
for episode in range(EPISODES):
print(f" -- {episode} -- ")
winner = game.play()
if winner is not None:
winner.wins = winner.wins + 1