[question]TRPO can't use LstmPolicy or MlpLstmPolicy?

[GIS-PuppetMaster]

from stable_baselines.common.policies import *
from stable_baselines import *
from stable_baselines.common.env_checker import check_env
from stable_baselines.common.vec_env.dummy_vec_env import DummyVecEnv
from TradeEnv import TradeEnv
from Util.Util import *
from Util.Callback import CustomCallback

episode = 5000
EP_LEN = 250 * 3
FILE_TAG = "TRPO"
mode = "train"
n_training_envs = 1


def post_processor(state):
    return log10plus1R(state) / 10


def make_env():
    env = TradeEnv(obs_time_size='60 day', obs_delta_frequency='1 day', sim_delta_time='1 day',
                   start_episode=0, episode_len=EP_LEN, stock_code='000938_XSHE',
                   result_path="E:/运行结果/TRPO/" + FILE_TAG + "/" + mode + "/",
                   stock_data_path='E:\PycharmProjects\DPPO\myDPPO\Data/train/',
                   poundage_rate=1.5e-3, reward_verbose=1, post_processor=post_processor,
                   max_episode_days=EP_LEN)
    env.seed(0)
    check_env(env)
    return env


env = DummyVecEnv([make_env for _ in range(n_training_envs)])
callback = CustomCallback()
model = TRPO(MlpLstmPolicy, env, verbose=1, tensorboard_log="./log/", seed=0)             
model.learn(total_timesteps=episode * EP_LEN, callback=callback)
model.save("./model")

Describe the bug
the observation of my ENV is
self.observation_space = spaces.Box(
low=np.array([[float('-inf') for _ in range(26)]
for _ in range(self.obs_time // self.obs_delta_frequency)]),
high=np.array([[float('inf') for _ in range(26)]
for _ in range(0, self.obs_time // self.obs_delta_frequency)]))
its shape is like (timestep=60, feature=26)

When I run this code, I got error like this:
ValueError: The initializer passed is not valid. It should be a callable with no arguments and the shape should not be provided or an instance of tf.keras.initializers.*' and shape` should be fully defined.
But if I use PPO2, it runs normal

Code example
see above

Traceback (most recent call last): 
Instructions for updating:
Colocations handled automatically by placer.
Traceback (most recent call last):
  File "C:\Users\zkx74\Anaconda3\lib\site-packages\stable_baselines\trpo_mpi\trpo_mpi.py", line 138, in setup_model
    None, reuse=False, **self.policy_kwargs)
  File "C:\Users\zkx74\Anaconda3\lib\site-packages\stable_baselines\common\policies.py", line 681, in __init__
    layer_norm=False, feature_extraction="mlp", **_kwargs)
  File "C:\Users\zkx74\Anaconda3\lib\site-packages\stable_baselines\common\policies.py", line 427, in __init__
    layer_norm=layer_norm)
  File "C:\Users\zkx74\Anaconda3\lib\site-packages\stable_baselines\common\tf_layers.py", line 143, in lstm
    weight_x = tf.get_variable("wx", [n_input, n_hidden * 4], initializer=ortho_init(init_scale))
  File "C:\Users\zkx74\Anaconda3\lib\site-packages\tensorflow\python\ops\variable_scope.py", line 1479, in get_variable
    aggregation=aggregation)
  File "C:\Users\zkx74\Anaconda3\lib\site-packages\tensorflow\python\ops\variable_scope.py", line 1220, in get_variable
    aggregation=aggregation)
  File "C:\Users\zkx74\Anaconda3\lib\site-packages\tensorflow\python\ops\variable_scope.py", line 547, in get_variable
    aggregation=aggregation)
  File "C:\Users\zkx74\Anaconda3\lib\site-packages\tensorflow\python\ops\variable_scope.py", line 499, in _true_getter
    aggregation=aggregation)
  File "C:\Users\zkx74\Anaconda3\lib\site-packages\tensorflow\python\ops\variable_scope.py", line 890, in _get_single_variable
    raise ValueError("The initializer passed is not valid. It should "
ValueError: The initializer passed is not valid. It should be a callable with no arguments and the shape should not be provided or an instance of `tf.keras.initializers.*' and `shape` should be fully defined.

System Info
Describe the characteristic of your environment:

• Describe how the library was installed: pip in anaconda env
• GPU models and configuration: default tf-gpu
• Python version 3.7
• Tensorflow version 1.13.2
• Versions of any other relevant libraries

Additional context
Add any other context about the problem here.

[Miffyli]

TRPO does not support recurrent architectures, as shown in the implementation grid.

Duplicate of #140.