'end_time': first_experiment_end_time }) # Schedule second experiment second_experiment_start_time = '185530' # format: %H%M%S e.g. 2:30pm is 143000 second_experiment_end_time = '191000' # format: %H%M%S e.g. 4:00pm is 160000 second_experiment_start_delay = (datetime.strptime( date.today().strftime("%Y%m%d") + '-' + second_experiment_start_time, '%Y%m%d-%H%M%S') - open_time).total_seconds() if second_experiment_start_delay < 0: logging.error( 'Second Experiment starts earlier than the end of First Experiment!') scheduler.enter(second_experiment_start_delay, 1, interact_with_learning_agent, kwargs={ 'agent': LAS_agent_community, 'env': envLAS, 'end_time': second_experiment_end_time }) if __name__ == '__main__': # Run two experiments logging.info('Run scheduler...') scheduler.run() logging.info('Scheduler done!') envLAS.destroy()
reward = 0 done = False start_time = time.time() try: for i in range(max_episodes): observation = env.reset() ep_reward = 0 for j in range(max_episode_len): if render_env == True: env.render() # Added exploration noise action = LASAgent.perceive_and_act(observation, reward, done) observation, reward, done, info = env.step(action[0]) ep_reward += reward if done or j == (max_episode_len - 1): print('| Reward: {:d} | Episode: {:d} '.format( int(ep_reward), i)) episod_reward_memory.append(ep_reward) plot_cumulative_reward(episod_reward_memory) break #time.sleep(0.5) print("Time elapsed:{}".format(time.time() - start_time)) except KeyboardInterrupt: sess.close() env.destroy() print("Shut Down.")