def test_scheduler_tick_multiple_events():
executed_events = set()
scheduler = Scheduler()
scheduler.schedule(TrackedEvent("event one", executed_events), 1)
scheduler.schedule(TrackedEvent("event two", executed_events), 2)
scheduler.schedule(TrackedEvent("event three", executed_events), 2)
scheduler.schedule(TrackedEvent("event four", executed_events), 3)
scheduler.tick()
assert len(scheduler.priority_queue._queue) == 3
assert executed_events == {"event one"}
scheduler.tick()
assert len(scheduler.priority_queue._queue) == 2
assert executed_events == {"event one", "event two"}
scheduler.tick()
assert len(scheduler.priority_queue._queue) == 1
assert executed_events == {"event one", "event two", "event three"}
scheduler.tick()
assert len(scheduler.priority_queue._queue) == 0
assert executed_events == {
"event one", "event two", "event three", "event four"
}
def test_scheduler_unschedule():
event = Event()
scheduler = Scheduler()
scheduler.schedule(event, 6)
scheduler.schedule(Event(), 5)
scheduler.schedule(Event(), 7)
scheduler.unschedule(event)
for priority, scheduled_event in scheduler.priority_queue._queue:
assert scheduled_event is not event
import time
from scheduling.Scheduler import *
start = time.time()
scheduler = Scheduler()
scheduler.run()
scheduler.print_history()
scheduler.visualize_history()
end = time.time()
print("Elapsed time: %f s" % (end - start))
# # example for using visualize_history_file()
history = read_json_file("scheduling_history.json")
visualize_history_file(history)
#calculate group average response time from history file
group_response_time = get_group_avg_response_time(history)
print(group_response_time)
#Elapsed time: 381.918144 s
#[10846.0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
type=str,
default='../adv_agent/obs_rms.pkl')
parser.add_argument('--x_method', type=str, default='grad')
parser.add_argument('--surrogate_model',
type=str,
default="../agent-zoo/agent/YouShallNotPass_agent.pkl")
parser.add_argument('--mimic_model_path',
type=str,
default="../agent-zoo/agent/mimic_model.h5")
args = parser.parse_args()
adv_agent_path = args.adv_agent_path
adv_agent_norm_path = args.adv_agent_norm_path
scheduler = Scheduler(
annealer_dict={'lr': ConstantAnnealer(learning_rate)})
env_name = env_list[args.env]
# define the env_path
env_path = args.surrogate_model
mimic_model_path = args.mimic_model_path
env = gym.make(env_name)
venv = SubprocVecEnv([
lambda: make_adv_multi2single_env(env_name, adv_agent_path,
adv_agent_norm_path, False)
for i in range(n_cpu)
])
venv = Monitor(venv, 0)
rew_shape_venv = apply_reward_wrapper(single_env=venv,
def test_scheduler_tick():
scheduler = Scheduler()
scheduler.schedule(TrackedEvent("event one", set()), 1)
latest_event = scheduler.tick()
assert latest_event.id == "event one"
def test_scheduler_schedule():
scheduler = Scheduler()
scheduler.schedule(Event(), 5)
assert len(scheduler.priority_queue._queue) == 1
last_checkpoint = step
if step - log_interval > last_log:
log_callback(logger, locals, globals)
last_log = step
return True
model.learn(total_timesteps=total_timesteps,
log_interval=1,
callback=callback,
seed=seed)
if __name__ == "__main__":
scheduler = Scheduler(annealer_dict={'lr':
ConstantAnnealer(LR)}) # useless
env_name = GAME_ENV
# multi to single, apply normalization to victim agent's observation, reward, and diff reward.
venv = SubprocVecEnv([
lambda: make_zoo_multi2single_env(env_name,
VIC_AGT_ID,
REW_SHAPE_PARAMS,
scheduler,
reverse=REVERSE,
total_step=TRAINING_ITER)
for i in range(N_GAME)
])
# test
if REVERSE:
def setUp(self):
self.s = Scheduler()
def setUp(self):
self.pcb1 = PCB(3, 5, 10)
self.pcb2 = PCB(2, 5, 10)
self.pcb3 = PCB(1, 5, 10)
self.scheduler = Scheduler()
