def test_creation_multienv(self):
nb_env = 2
with warnings.catch_warnings():
warnings.filterwarnings("ignore")
with make("rte_case5_example", test=True) as env:
multi_envs = MultiEnvironment(env=env, nb_env=nb_env)
obss, rewards, dones, infos = multi_envs.step(
[env.action_space() for _ in range(nb_env)])
for ob in obss:
assert isinstance(ob, CompleteObservation)
obss = multi_envs.reset()
for ob in obss:
assert isinstance(ob, CompleteObservation)
multi_envs.close()
def make_multi_env(env_init, nb_env):
"""
This function creates a multi environment compatible with what is expected in the baselines. In particular, it
adds the observation_space, the action_space and the reward_range attribute.
The way this function works is explained in the getting_started of grid2op.
Attributes
-----------
env_init: :class:`grid2op.Environment.Environment`
The environment to duplicates
nb_env: ``int``
The number of environment on with which you want to interact at the same time
Returns
-------
res: :class:`grid2op.Environment.MultiEnvironment` or :class:`grid2op.Environment.Environment`
A copy of the initial environment (if nb_env = 1) or a MultiEnvironment based on the initial environment
if nb_env >= 2.
"""
res = None
nb_env = int(nb_env)
if nb_env <= 0:
raise RuntimeError(
"Impossible to create a negative number of environments")
if nb_env == 1:
warnings.warn(
"You asked to create 1 environment. We didn't use the MultiEnvironment for that. We instead "
"created a copy of your initial environment.")
res = Environment(**env_init.get_kwargs())
else:
res = MultiEnvironment(nb_env=nb_env, env=env_init)
res.observation_space = env_init.observation_space
res.action_space = env_init.action_space
res.reward_range = env_init.reward_range
return res
def test_creation_multienv(self):
nb_env = 1
with make("case5_example") as env:
multi_envs = MultiEnvironment(env=env, nb_env=nb_env)
multi_envs.close()
return res
# Use custom params
params = Parameters()
# Create grid2op game environement
env_init = None
env = make(args.env_name,
param=params,
reward_class=MyReward,
backend=backend)
if args.nb_env > 1:
env_init = env
from grid2op.Environment import MultiEnvironment
env = MultiEnvironment(int(args.nb_env), env)
# TODO hack i'll fix in 0.9.0
env.action_space = env_init.action_space
env.observation_space = env_init.observation_space
env.fast_forward_chronics = lambda x: None
env.chronics_handler = env_init.chronics_handler
env.current_obs = env_init.current_obs
nm_ = args.name if args.name is not None else DEFAULT_NAME
try:
train(env,
name=nm_,
iterations=args.num_train_steps,
save_path=args.save_path,
load_path=args.load_path,
logs_dir=args.logs_dir,
def test_seeding(self):
nb_env = 2
with warnings.catch_warnings():
warnings.filterwarnings("ignore")
with make("rte_case5_example", test=True) as env:
env.seed(2)
multi_envs1 = MultiEnvironment(env=env, nb_env=nb_env)
seeds_1 = multi_envs1.get_seeds()
multi_envs1.close()
multi_envs2 = MultiEnvironment(env=env, nb_env=nb_env)
seeds_2 = multi_envs2.get_seeds()
multi_envs2.close()
env.seed(2)
multi_envs3 = MultiEnvironment(env=env, nb_env=nb_env)
seeds_3 = multi_envs3.get_seeds()
multi_envs3.close()
assert np.all(seeds_1 == seeds_3)
assert np.any(seeds_1 != seeds_2)
def test_creation_multienv(self):
nb_env = 2
with warnings.catch_warnings():
warnings.filterwarnings("ignore")
with make("rte_case5_example", test=True) as env:
multi_envs = MultiEnvironment(env=env, nb_env=nb_env)
obss, rewards, dones, infos = multi_envs.step(
[env.action_space() for _ in range(nb_env)])
for ob in obss:
assert isinstance(ob, CompleteObservation)
obss = multi_envs.reset()
for ob in obss:
assert isinstance(ob, CompleteObservation)
# test some actions will not throw errors
multi_envs.set_ff(7 * 288)
multi_envs.set_chunk_size(128)
obss = multi_envs.reset()
seeds = multi_envs.get_seeds()
multi_envs.close()