def calc_session_fitness_df(session, session_data):
'''Calculate the session fitness df'''
session_fitness_data = {}
for aeb in session_data:
aeb_df = session_data[aeb]
util.downcast_float32(aeb_df)
body = session.aeb_space.body_space.data[aeb]
aeb_fitness_sr = calc_aeb_fitness_sr(aeb_df, body.env.name)
aeb_fitness_df = pd.DataFrame([aeb_fitness_sr], index=[session.index])
aeb_fitness_df = aeb_fitness_df.reindex(FITNESS_COLS[:3], axis=1)
session_fitness_data[aeb] = aeb_fitness_df
# form multiindex df, then take mean across all bodies
session_fitness_df = pd.concat(session_fitness_data, axis=1)
mean_fitness_df = calc_mean_fitness(session_fitness_df)
session_fitness = calc_fitness(mean_fitness_df)
logger.info(f'Session mean fitness: {session_fitness}\n{mean_fitness_df}')
return session_fitness_df
def tick_clocks(self, session):
'''Tick all the clock in body_space, and check its own done_space to see if clock should be reset to next episode'''
from slm_lab.experiment import analysis
# TODO simplify below
env_dones = []
body_end_sessions = []
for env in self.env_space.envs:
done = env.done or env.clock.get('t') > env.max_timestep
env_dones.append(done)
if done:
epi = env.clock.get('epi')
save_this_epi = 'save_epi_frequency' in env.env_spec and (
epi % env.env_spec['save_epi_frequency']) == 0
for body in env.nanflat_body_e:
self.body_done_log(body)
if epi > 0 and save_this_epi:
body.agent.algorithm.save(epi=epi)
env.clock.tick('epi')
else:
env.clock.tick('t')
env_end_session = env.clock.get('epi') > env.max_episode
body_end_sessions.append(env_end_session)
env_early_stops = []
if any(env_dones) and self.clock.get('epi') > analysis.MA_WINDOW:
session_mdp_data, session_data = analysis.get_session_data(session)
for aeb in session_data:
aeb_df = session_data[aeb]
util.downcast_float32(aeb_df)
body = self.body_space.data[aeb]
env_epi = body.env.clock.get('epi')
if env_epi > max(analysis.MA_WINDOW, body.env.max_episode / 2):
aeb_fitness_sr = analysis.calc_aeb_fitness_sr(
aeb_df, body.env.name)
strength = aeb_fitness_sr['strength']
# TODO properly trigger early stop
# env_early_stop = strength < analysis.NOISE_WINDOW
env_early_stop = False
else:
env_early_stop = False
env_early_stops.append(env_early_stop)
else:
env_early_stops.append(False)
end_session = all(body_end_sessions) or all(env_early_stops)
return end_session
def calc_session_fitness_df(session, session_data):
'''Calculate the session fitness df'''
session_fitness_data = {}
for aeb in session_data:
aeb_df = session_data[aeb]
util.downcast_float32(aeb_df)
body = session.aeb_space.body_space.data[aeb]
aeb_fitness_sr = calc_aeb_fitness_sr(aeb_df, body.env.name)
aeb_fitness_df = pd.DataFrame([aeb_fitness_sr], index=[session.index])
aeb_fitness_df = aeb_fitness_df.reindex(FITNESS_COLS[:3], axis=1)
session_fitness_data[aeb] = aeb_fitness_df
# form multiindex df, then take mean across all bodies
session_fitness_df = pd.concat(session_fitness_data, axis=1)
mean_fitness_df = session_fitness_df.mean(axis=1, level=3)
session_fitness = calc_fitness(mean_fitness_df)
logger.info(f'Session mean fitness: {session_fitness}\n{mean_fitness_df}')
return session_fitness_df
def tick_clocks(self, session):
'''Tick all the clock in body_space, and check its own done_space to see if clock should be reset to next episode'''
from slm_lab.experiment import analysis
# TODO simplify below
env_dones = []
body_end_sessions = []
for env in self.env_space.envs:
done = env.done or env.clock.get('t') > env.max_timestep
env_dones.append(done)
if done:
epi = env.clock.get('epi')
save_this_epi = 'save_epi_frequency' in env.env_spec and (epi % env.env_spec['save_epi_frequency']) == 0
for body in env.nanflat_body_e:
self.body_done_log(body)
if epi > 0 and save_this_epi:
body.agent.algorithm.save(epi=epi)
env.clock.tick('epi')
else:
env.clock.tick('t')
env_end_session = env.clock.get('epi') > env.max_episode
body_end_sessions.append(env_end_session)
env_early_stops = []
if any(env_dones) and self.clock.get('epi') > analysis.MA_WINDOW:
session_mdp_data, session_data = analysis.get_session_data(session)
for aeb in session_data:
aeb_df = session_data[aeb]
util.downcast_float32(aeb_df)
body = self.body_space.data[aeb]
env_epi = body.env.clock.get('epi')
if env_epi > max(analysis.MA_WINDOW, body.env.max_episode / 2):
aeb_fitness_sr = analysis.calc_aeb_fitness_sr(aeb_df, body.env.name)
strength = aeb_fitness_sr['strength']
# TODO properly trigger early stop
# env_early_stop = strength < analysis.NOISE_WINDOW
env_early_stop = False
else:
env_early_stop = False
env_early_stops.append(env_early_stop)
else:
env_early_stops.append(False)
end_session = all(body_end_sessions) or all(env_early_stops)
return end_session
def tick_clocks(self, session):
'''Tick all the clock in body_space, and check its own done_space to see if clock should be reset to next episode'''
from slm_lab.experiment import analysis
done_space = self.data_spaces['done']
env_dones = []
body_end_sessions = []
for env in self.env_space.envs:
clock = env.clock
done = env.done or clock.get('t') > env.max_timestep
env_dones.append(done)
if done:
done_space.data[:, env.e, :] = 1.
done_space.swap_data[env.e, :, :] = 1.
msg = f'Done: trial {self.info_space.get("trial")} session {self.info_space.get("session")} env {env.e} epi {clock.get("epi")}, t {clock.get("t")}'
logger.info(msg)
clock.tick('epi')
else:
clock.tick('t')
env_end_session = clock.get('epi') > env.max_episode
body_end_sessions.append(env_end_session)
env_early_stops = []
if any(env_dones) and self.clock.get('epi') > analysis.MA_WINDOW:
session_mdp_data, session_data = analysis.get_session_data(session)
for aeb in session_data:
aeb_df = session_data[aeb]
util.downcast_float32(aeb_df)
body = self.body_space.data[aeb]
env_epi = body.env.clock.get('epi')
if env_epi > max(analysis.MA_WINDOW, body.env.max_episode / 2):
aeb_fitness_sr = analysis.calc_aeb_fitness_sr(
aeb_df, body.env.name)
strength = aeb_fitness_sr['strength']
env_early_stop = strength < analysis.NOISE_WINDOW
else:
env_early_stop = False
env_early_stops.append(env_early_stop)
else:
env_early_stops.append(False)
end_session = all(body_end_sessions) or all(env_early_stops)
return end_session
