def run(count=10, **kwargs):
force.init()
for i in range(count):
kwargs['xh'] = i
_run(**kwargs)
env = SingleCartPoleEnv().unwrapped
RL = DeepQNetwork(n_actions=env.action_space.n,
n_features=env.observation_space.shape[0])
force.force_generator = force.ForceGenerator(0.0, 0.0, 0.0, 1.01)
def run(count=10, **kwargs):
force.init()
for i in range(count):
kwargs = {
'mode': 'reset',
'xh': i,
'maxepochcount': 1500,
'complexunit': 100.0
}
env = SingleCartPoleEnv().unwrapped
net = PolicyGradients()
while 1:
result = net.run(**kwargs)
if kwargs['mode'] == 'noreset':
break
if not result:
break
env = SingleCartPoleEnv().unwrapped
net = PolicyGradients()
force.force_generator = force.ForceGenerator(0.0, 0.0, 0.0, 1.01)
if not changed or newcomplex is None or newcomplex == complexes_list[-1]:
sess.close()
return False
break # 复杂度已经达到最大,结束
print('新的环境复杂度=%.3f,k=%.2f,w=%.2f,f=%.2f,sigma=%.2f' % (newcomplex, k, w, f, sigma))
if mode == 'reset':
sess.close()
return True
episode_number = 0
sess.close()
if __name__ == '__main__':
force.init()
for i in range(10):
kwargs = {'mode': 'reset', 'xh':i,'maxepochcount' : 1500,'complexunit':100.0}
env = SingleCartPoleEnv().unwrapped
net = PolicyGradients()
while 1:
result = net.run(**kwargs)
if kwargs['mode'] == 'noreset':
break
if not result:
break
env = SingleCartPoleEnv().unwrapped
net = PolicyGradients()
force.force_generator = force.ForceGenerator(0.0, 0.0, 0.0, 1.01)