def main():
parser = argparse.ArgumentParser(description="Compute the partition of a "
"graph using the Spectral Partition Algorithm.")
parser.add_argument('--nodes-file', '-f', help='the file containing the nodes',
default='demo_nodes.txt')
parser.add_argument('--output-file', '-o', help='the filename of the'
' communities PNG graph to be written')
args = parser.parse_args()
# part_1, part_2, nodes, edges, nodes_data, node_map = partition_graph(args.nodes_file)
# create_subgraph(part_1, edges, nodes_data, node_map, 'part_1.stg')
# create_subgraph(part_2, edges, nodes_data, node_map, 'part_2.stg')
add_deadline()
tasks = parse_tasks('deadline.stg')
population_size = 100
population = create_population(tasks, population_size)
fitness_history = [grade(tasks, population), ]
for i in range(500):
population = evolve(tasks, population)
score, missed = grade(tasks, population)
fitness_history.append(score)
print('iteration {} population: {} score: {} missed: {}'.format(i + 1, population_size, score, missed))
def oneway(*blocks, showid=False, unfold=True, stretch=2, output=False,
ga=False, **gakwargs) -> Union[str, Tuple[str, Gene]]:
global _blocks, _unfold
_str = StringIO()
_stretch, _unfold = stretch, unfold
_blocks = blocks
_len = len(blocks)
if _len > 0:
_sequence = tuple(range(_len))
if ga:
gakw_local = dict(evaluate=intersect, size=_len, elite=0.2,
population=100, generation=300, showprogress=True)
gakw_local.update(**gakwargs)
_sequence = evolve(_sequence, **gakw_local)
_inters = intersect(_sequence, output=True)
print(f'{_inters[0]} intersection(s)', file=_str)
print(f'< {_inters[-1][0]} | {_inters[-1][1]} >', file=_str)
_blocks = tuple([_blocks[i] for i in _sequence])
_map = draw(_blocks, *_inters[1:], showid=showid, stretch=2)
for row in _map:
print('\n', end='', file=_str)
for p in row:
print(p, end='', file=_str)
print(file=_str)
if output:
return _str.getvalue(), _sequence
else:
return _str.getvalue()
def test_evolve(self):
print(self._str('evolve'))
from genetics import evolve
def f(nlist):
s = 0
for n in nlist:
s = s * 10 + n
return s
print(evolve(evaluate=f, size=15, generation=20, showprogress=False))
def evolve_further():
print('Loading population from',tempfile,'...')
model = new_actor()
population = []
try:
# population = load_actors(savefile, model, criterion='best', num=population_size)
population = undump_genomes(tempfile, model)
os.remove(tempfile)
except:
print('Could not load from',tempfile,'.')
population = list(population) # enforce type of population
# Add members to population as necessary
population += [new_actor() for _ in range(population_size - len(population))]
print('Evolving population...')
population = evolve(population, env, savefile=savefile, dumpfile=tempfile, **evolve_args)
dump_genomes(tempfile, population)
from actor import GeneticNNActor
from genetics import evolve, render_from_file
HIDDEN_LAYERS = [10, 6]
SAVEFILE = 'MsPacman_NN_10_6.txt'
try:
env = gym.make('MsPacman-ram-v0')
population = [
GeneticNNActor(env.observation_space,
env.action_space,
hidden_layers=HIDDEN_LAYERS) for _ in range(100)
]
model = population[0]
evolve(population,
env,
generations=1000,
simulation_reps=5,
p_mutation=0.05,
mutation_scale=0.25,
max_steps=100000,
render_gens=None,
savefile=SAVEFILE,
savenum=3,
allow_parallel=True)
except KeyboardInterrupt:
print('Interrupted...')
pass
finally:
print('Top 3 Actors found:')
render_from_file(SAVEFILE, model, env, num=3)
import gym
from actor import GeneticPerceptronActor
from execution import simulate
from genetics import evolve
env = gym.make('Acrobot-v1')
population = [
GeneticPerceptronActor(env.observation_space, env.action_space)
for _ in range(100)
]
evolve(population,
env,
generations=101,
simulation_reps=25,
max_steps=250,
render_gens=5)
if __name__ == '__main__':
import gym
from actor import GeneticPerceptronActor
from execution import simulate
from genetics import evolve
env = gym.make('CartPole-v1')
population = [
GeneticPerceptronActor(env.observation_space, env.action_space)
for _ in range(100)
]
evolve(population,
env,
generations=21,
simulation_reps=25,
max_steps=20000,
render_gens=1,
savefile='CartPole_Actors.txt',
savenum=1)