def make_agent(env, submission_dir):
"""
This function will be used by codalab to create your agent. It should accept exactly an environment and a path
to your submission directory and return a valid agent.
"""
agent = RandomAgent(env.action_space)
return agent
def test_2random(self):
with warnings.catch_warnings():
warnings.filterwarnings("ignore")
env = grid2op.make("rte_case5_example", test=True)
env2 = grid2op.make("rte_case14_realistic", test=True)
agent = RandomAgent(env.action_space)
agent2 = RandomAgent(env2.action_space)
# test i can reset the env
obs = env.reset()
obs2 = env2.reset()
# test the agent can act
act = agent.act(obs, 0., False)
act2 = agent2.act(obs2, 0., False)
# test the env can step
_ = env.step(act)
_ = env2.step(act2)
def test_random(self):
with warnings.catch_warnings():
warnings.filterwarnings("ignore")
with grid2op.make("rte_case5_example", test=True) as env:
obs = env.reset()
my_agent = RandomAgent(env.action_space)
my_agent.seed(0)
nb_test = 100
res = np.zeros(nb_test, dtype=np.int)
res2 = np.zeros(nb_test, dtype=np.int)
res3 = np.zeros(nb_test, dtype=np.int)
for i in range(nb_test):
res[i] = my_agent.my_act(obs, 0., False)
my_agent.seed(0)
for i in range(nb_test):
res2[i] = my_agent.my_act(obs, 0., False)
my_agent.seed(1)
for i in range(nb_test):
res3[i] = my_agent.my_act(obs, 0., False)
# the same seeds should produce the same sequence
assert np.all(res == res2)
# different seeds should produce different sequence
assert np.any(res != res3)
def __init__(self, action_space):
RandomAgent.__init__(self, action_space)
self.i = 0
def __init__(self, *args, **kwargs):
RandomAgent.__init__(self, *args, **kwargs)
self.seeds = []
this_epi_scores = json.load(f)
score_this_ep, nb_ts_survived, total_ts_tmp = \
self._compute_episode_score(ep_id,
meta=this_epi_meta,
other_rewards=this_epi_scores,
dn_metadata=meta_data_dn,
no_ov_metadata=no_ov_metadata)
all_scores.append(score_this_ep)
ts_survived.append(nb_ts_survived)
total_ts.append(total_ts_tmp)
if need_delete:
dir_tmp.cleanup()
return all_scores, ts_survived, total_ts
if __name__ == "__main__":
import grid2op
from lightsim2grid import LightSimBackend
from grid2op.Agent import RandomAgent, DoNothingAgent
env = grid2op.make("l2rpn_case14_sandbox", backend=LightSimBackend())
nb_scenario = 16
my_score = ScoreL2RPN2020(env,
nb_scenario=nb_scenario,
env_seeds=[0 for _ in range(nb_scenario)],
agent_seeds=[0 for _ in range(nb_scenario)])
my_agent = RandomAgent(env.action_space)
my_agent = DoNothingAgent(env.action_space)
print(my_score.get(my_agent))
stats_dn.clear_episode_data()
# you can also change the parameters
param = Parameters()
param.NO_OVERFLOW_DISCONNECTION = True
stats_no_overflow = EpisodeStatistics(env, name_stats="no_overflow")
stats_no_overflow.compute(
nb_scenario=nb_scenario,
parameters=param,
pbar=True,
scores_func=L2RPNSandBoxScore
) # this will take a while to compute in most cases
stats_no_overflow.clear_episode_data()
# or use a different agent
my_agent = RandomAgent(
env.action_space) # use any grid2op agent you want here
stats_custom_agent = EpisodeStatistics(env, name_stats="custom_agent")
stats_custom_agent.compute(
nb_scenario=nb_scenario,
agent=my_agent,
pbar=True,
scores_func=L2RPNSandBoxScore
) # this will take a while to compute in most cases
stats_custom_agent.clear_episode_data()
# and then you can retrieve the statistics
rho_dn, ids = stats_dn.get("rho")
rho_dn_all, ids = stats_no_overflow.get("rho")
rho_custom_agent, ids = stats_custom_agent.get("rho")