from Env_DQN import Environment
import torch
# from DQN import DQN
# from Dueling_DQN import DQN
# from DDQN import DQN
# from n_stepDQN import DQN
from Advance_DQN import DQN
from matplotlib import pyplot as plt
import numpy as np
plt.rcParams['font.sans-serif'] = ['SimHei']
plt.rcParams['axes.unicode_minus'] = False
env = Environment()
flag = False
current = 10
price_recode = list()
x_index = list()
model_reward = 0
for ij in range(0, 38):
current += 10
print(current)
MODEL_NAME = 'model5/advanceDQN/' + str(
env.name) + '_AE_Advance_dqn_' + str(current) + '.pt'
# MODEL_NAME = 'model5/advanceDQN/'+str(env.name)+'_Advance_dqn_'+str(current)+'.pt'
# MODEL_NAME = 'model3/dqn/advance_dqn/AE_Advance_dqn2_'+str(current)+'.pt'
# MODEL_NAME = 'model3/dqn/advance_dqn/Advance_dqn2_'+str(current)+'.pt'
# MODEL_NAME = 'model2/dqn/advance_dqn/Use_AE_n-step_double_dueling_dqn_'+str(current)+'.pt'
# MODEL_NAME = 'model2/dqn/advance_dqn/AE_n_stepDQN_'+str(current)+'.pt'
# MODEL_NAME = 'model2/dqn/advance_dqn/AE_n_stepDQN_'+str(current)+'.pt'
# MODEL_NAME = 'model2/dqn/advance_dqn/dueling_dqn_'+str(current)+'.pt'
# MODEL_NAME = 'model2/dqn/advance_dqn/AE_dueling_dqn_'+str(current)+'.pt'
import torch.nn as nn
import torch.nn.functional as F
import numpy as np
from Env_DQN import Environment
from matplotlib import pyplot as plt
from Net import Net
plt.rcParams['font.sans-serif'] = ['SimHei']
plt.rcParams['axes.unicode_minus'] = False
BATCH_SIZE = 128 # 每一批的训练量
LR = 0.01 # 学习率
TAU = 0.001
EPSILON = 0.9 # 贪婪策略指数，Q-learning的一个指数，用于指示是探索还是利用。
GAMMA = 1 # reward discount
MEMORY_CAPACITY = 10000
env = Environment()
Hidden_num = 128
N_ACTIONS = env.action_dim
N_STATES = env.state_dim+2
# 创建Q-learning的模型
class DQN(object):
def __init__(self):
# 两张网是一样的，不过就是target_net是每100次更新一次，eval_net每次都更新
self.eval_net, self.target_net = Net(N_STATES, N_ACTIONS, Hidden_num), Net(N_STATES, N_ACTIONS, Hidden_num)
self.learn_step_counter = 0 # 如果次数到了，更新target_net
self.memory_counter = 0 # for storing memory
self.memory = np.zeros((MEMORY_CAPACITY, N_STATES * 2 + 2)) # 初始化记忆
self.optimizer = torch.optim.Adam(self.eval_net.parameters(), lr=LR)
self.loss_func = nn.MSELoss()