Env creating error - 'int' object is not iterable

[guidesc]

Hi Wassname,

Thanks for open source your code. However, when I tried to run the following code to create an env.

import rl_portfolio_management.environments  # this registers them

env = gym.envs.spec('CryptoPortfolioEIIE-v0').make()
print("CryptoPortfolioEIIE has an history shape suitable for an EIIE model (see https://arxiv.org/abs/1706.10059)")
observation = env.reset()
print("shape =", observation["history"].shape)
# shape = (5, 50, 3)```

I encountered the following error: 
```---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-10-f00e24a5d49e> in <module>()
      2 import rl_portfolio_management.environments  # this registers them
      3 
----> 4 env = gym.envs.spec('CryptoPortfolioEIIE-v0').make()
      5 print("CryptoPortfolioEIIE has an history shape suitable for an EIIE model (see https://arxiv.org/abs/1706.10059)")
      6 observation = env.reset()

D:\Users\ThinkPad\Anaconda3\lib\site-packages\gym\envs\registration.py in make(self)
     84         else:
     85             cls = load(self._entry_point)
---> 86             env = cls(**self._kwargs)
     87 
     88         # Make the enviroment aware of which spec it came from.

F:\DSBA\T2\Advanced ML\Final\rl-portfolio-management-master\rl-portfolio-management-master\rl_portfolio_management\environments\portfolio.py in __init__(self, df, steps, trading_cost, time_cost, window_length, augment, output_mode, log_dir, scale, scale_extra_cols, random_reset)
    266         nb_assets = len(self.src.asset_names)
    267         self.action_space = gym.spaces.Box(
--> 268             0.0, 1.0, shape=nb_assets + 1)
    269 
    270 

D:\Users\ThinkPad\Anaconda3\lib\site-packages\gym\spaces\box.py in __init__(self, low, high, shape, dtype)
     32         self.low = low.astype(dtype)
     33         self.high = high.astype(dtype)
---> 34         gym.Space.__init__(self, shape, dtype)
     35 
     36     def sample(self):

D:\Users\ThinkPad\Anaconda3\lib\site-packages\gym\core.py in __init__(self, shape, dtype)
    200     """
    201     def __init__(self, shape=None, dtype=None):
--> 202         self.shape = None if shape is None else tuple(shape)
    203         self.dtype = None if dtype is None else np.dtype(dtype)
    204 

TypeError: 'int' object is not iterable

I also tried gym.spaces.Box(0.0, 1.0, shape=5), it has the same error. While then I tried gym.spaces.Box(0.0, 1.0, shape=(5,)), it works well.

However, I tried to update your code with self.action_space = gym.spaces.Box(0.0, 1.0, shape=(nb_assets + 1, )), it still did not work well.

Could you please help me on that?

Thx in advance.

[wassname]

It could be a version difference, could you please run conda list and compare with requirements.txt? Or you can run gym.__version__ from python and compare that.

[guidesc]

Thanks, it is a version issue. My gym version is 0.10.4. Then I tried to create a virtualenv and pip install gym==0.9.3, it works well.

Thank you. I'm really curious what has been updating in 0.10.4 version of gym.

[wassname]

No problems, glad it works now

Looks like breaking changes were probobly in 0.9.6. They probably did it because of the robotics environments release that happened around the same time (I guess).

[yutingsz]

my gym version is 0.10.5. Get the same error. Any update to the code apart from the virtualenv? Thanks