def create_agent_66_experiment_runner(num_dcs, num_customers, dcs_per_customer,
demand_mean, demand_var, num_commodities,
orders_per_day, num_steps):
physical_network = PhysicalNetwork(
num_dcs,
num_customers,
dcs_per_customer,
demand_mean,
demand_var,
num_commodities,
)
order_generator = ActualOrderGenerator(physical_network, orders_per_day)
generator = DirichletInventoryGenerator(physical_network)
environment_parameters = EnvironmentParameters(physical_network,
order_generator, generator,
num_steps)
env = ShippingFacilityEnvironment(environment_parameters)
agent = AgentHighest(env)
return ExperimentRunner(order_generator,
generator,
agent,
env,
experiment_name="randomvalid_validation")
def run_with_params(num_dcs, num_customers, dcs_per_customer, demand_mean,
demand_var, num_commodities, orders_per_day, num_steps):
physical_network = PhysicalNetwork(
num_dcs,
num_customers,
dcs_per_customer,
demand_mean,
demand_var,
num_commodities,
)
# order_generator = NaiveOrderGenerator(num_dcs, num_customers, orders_per_day)
order_generator = ActualOrderGenerator(physical_network, orders_per_day)
generator = DirichletInventoryGenerator(physical_network)
environment_parameters = EnvironmentParameters(physical_network,
order_generator, generator,
num_steps)
env = ShippingFacilityEnvironment(environment_parameters)
agent = RandomAgent(env)
obs = env.reset()
reward = 0
done = False
print("=========== starting episode loop ===========")
print("Initial environment: ")
env.render()
actions = []
episode_rewards = []
#demands_per_k = np.zeros((num_commodities,num_steps))
#inventory_at_t = np.zeros((num_commodities,num_steps)) #todo llenar estos eventualmente
while not done:
action = agent.act(obs, reward, done)
# print(f"Agent is taking action: {action}")
# the agent observes the first state and chooses an action
# environment steps with the agent's action and returns new state and reward
obs, reward, done, info = env.step(action)
# print(f"Got reward {reward} done {done}")
# Render the current state of the environment
env.render()
actions.append(action)
episode_rewards.append(reward)
if done:
print("===========Environment says we are DONE ===========")
return actions, episode_rewards