def reset(self, Date, path):
self.env = envirment.Env(Date, path)
self.a = np.zeros((1,self.a_dim))
for i in range(self.a_dim):
if np.random.uniform()<0.2:
self.a[0,i] = 1
self.a = np.ravel(self.a)
self.a_last = np.ravel(np.zeros((1,self.a_dim)))
self.s = np.ravel(np.zeros((1,self.s_dim)))
self.s_ = np.ravel(np.zeros((1,self.s_dim)))
self.s = self.env.update(self.a, self.a_last, 0, 0)
self.s_ = self.s
self.a_last = self.a
self.s = np.transpose(self.s)
self.s_ = np.transpose(self.s_)
os.makedirs(path+'/car_L1_reward_figure')
os.makedirs(path+'/AdaptSpeed')
os.makedirs(path+'/ConvergenceRate')
os.makedirs(path+'/log')
os.makedirs(path+'/result_images')
os.makedirs(path+'/SystemPerformance')
Date = 1001
'''
main program
'''
# dqn = dqn.DQN()
Agent = control_group.GreedyPolicy.Greedy(a_dim)
ddpg = PDDPG.PDDPG(a_dim, s_dim)
# ddpg = ddpg.DDPG(a_dim, s_dim)
env = envirment.Env(Date, path)
# retarder = Queue()
sys_per = output.SystemPerformance(a_dim)
line_fig = output.LineFigures(a_dim)
conv_rate = output.ConvergenceRate(MAX_EPISODES, SLOTNUM, a_dim, s_dim)
adapt_speed = output.AdaptSpeed(SLOTNUM, a_dim)
EPSILON = 0.95
a = np.zeros((1,a_dim))
for i in range(a_dim):
if np.random.uniform()<0.2:
a[0,i] = 1
a = np.ravel(a)