def __init__(self, lifespan=None, test=False): """ Initialize the world. Base it on the grid_1D world. """ Grid_1D_World.__init__(self, lifespan) self.name = 'grid_1D_delay' self.name_long = 'one dimensional grid world with delay' print '--delayed' self.MAX_DELAY = 4 self.future_reward = [0.] * self.MAX_DELAY self.world_visualize_period = 1e6
def __init__(self, lifespan=None): """ Initialize the world. Base it on the grid_1D world. """ Grid_1D_World.__init__(self, lifespan) self.name = 'grid_1D_delay' self.name_long = 'one dimensional grid world with delay' print '--delayed' self.display_state = False self.MAX_DELAY = 4 self.future_reward = [0.] * self.MAX_DELAY
def __init__(self, lifespan=None): """ Initialize the world. Base it on the grid_1D world. Parameters ---------- lifespan : int The number of time steps to continue the world. """ Grid_1D_World.__init__(self, lifespan) self.name = 'grid_1D_delay' self.name_long = 'one dimensional grid world with delay' print '--delayed' self.max_delay = 9 self.future_reward = [0.] * self.max_delay self.world_visualize_period = 1e6
""" agent_name = '_'.join((world.name, 'agent')) agent = Agent(world.num_sensors, world.num_actions, agent_name=agent_name, show=show) if restore: agent = agent.restore() actions = np.zeros((world.num_actions, 1)) # Repeat the loop through the duration of the existence of the world while (world.is_alive()): sensors, reward = world.step(actions) world.visualize(agent) actions = agent.step(sensors, reward) return agent.report_performance() def profile(): print 'profiling BECCA\'s performance...' cProfile.run('test(World(lifespan=profiling_lifespan), restore=True)', 'tester_profile') p = pstats.Stats('tester_profile') p.strip_dirs().sort_stats('time', 'cumulative').print_stats(30) if __name__ == '__main__': if profile_flag: profile() else: test(World(lifespan=testing_lifespan), restore=True)