eval_policy.py

"""
        This file helps the testing with evaluating the policy
"""


#--------------------------------- Rollout -----------------------------------#
# running the simulation to evaluate the policy
def rollout(policy, env, render):

    while True:

        # reset the environment for new episode
        obs, _ = env.reset()
        done = False

        # initialize the length of the episode
        t = 0
        ep_len = 0

        # initialize the summation of the reward in the episode
        ep_rew = 0

        while not done:
            t += 1         # calculate the length of the episode

            # render the environment
            if render:
                env.render()

            # The action generated by the policy
            action = policy(obs).detach().numpy()

            # observation, reward and done from simulation
            obs, rew, done, _, _ = env.step(action)

            # the summation of the reward in the episode
            ep_rew += rew

        # calculate the length of the episode
        ep_len = t

        # return the length of total reward of the episode:
        yield ep_len, ep_rew


#------------------------------ logger summary -------------------------------#
def _log_summary(ep_len, ep_ret, ep_num):

    print(flush = True)
    print(f"-------------------- episode #{ep_num+1} --------------------", flush=True)
    print(f"Total cost:    {str(ep_ret)}", flush=True)
    print(f"----------------------------------------------------", flush=True)
    print(flush=True)


#---------------------------- Policy evaluation ------------------------------#
def eval_policy(policy, env, render= False):

    for ep_num, (ep_len, ep_ret) in enumerate(rollout(policy, env, render)):
        _log_summary(ep_len=ep_len, ep_ret=ep_ret, ep_num=ep_num)


#-----------------------------------------------------------------------------#