def __init__(self, instance, minibuckets, dimensions, search_method):
self.instance = instance
self.minibuckets = minibuckets
self.dimensions = dimensions
self.search_method = search_method
if self.dimensions == 1:
self.path = os.path.join(MONO_DIR, instance)
else:
self.path = os.path.join(BI_DIR, instance)
self.graph, self.original_graph = read_graph(self.path)
self.order = get_variables_order(self.graph)
self.original_order = get_variables_order(self.original_graph)
self.heuristic_solver = MiniBucket(self.order,
self.original_order,
self.minibuckets,
vertex_cover_cost,
debug=False)
self.heuristic_solver.build_buckets()
if search_method == "bb":
self.search_solver = BranchAndBound(self.heuristic_solver,
len(self.graph))
elif search_method == "nsga2":
self.search_solver = NSGA2(self.order, self.heuristic_solver)
def objective(trial):
test_graphs = [read_graph(
"data/random/1000_2500_{}".format(i)).adj for i in range(5)]
layer_num = trial.suggest_int('layer_num', 2, 20)
feature = trial.suggest_int('feature', 5, 10)
beta = trial.suggest_uniform('beta', 0.99, 0.997)
gnn = GIN3(layer_num=layer_num, feature=feature)
gnn.to(device)
trainer = MCTSTrainer(gnn, test_graphs, "optuna_tmp_100_tran2_{}_{}_{}".format(
layer_num, feature, beta))
Timer.start('all')
ans = []
for i in range(500):
print("epoch: ", i)
graph = generate_random_graph(100, 250).adj
trainer.train2(graph, 10 * beta ** i)
trainer.test()
Timer.end('all')
score = 0
coef = 1
for all_rewards in reversed(trainer.test_result):
coef *= 0.9
for rewards in all_rewards:
score += 10 * coef * np.max(rewards)
score += coef * np.mean(rewards)
Timer.print()
trainer.save_model()
trainer.save_test_result()
return -score
def test_graph():
flights = [("ORD", "SEA"), ("ORD", "LAX"), ('ORD', 'DFW'), ('ORD', 'PIT'),
('SEA', 'LAX'), ('LAX', 'DFW'), ('ATL', 'PIT'), ('ATL', 'RDU'),
('RDU', 'PHL'), ('PIT', 'PHL'), ('PHL', 'PVD')]
G = {}
for (x,y) in flights: graph.make_link(G,x,y)
marvel_G = graph.read_graph("resources/marvel_graph.tsv")
assert graph.clustering_coefficient(G) == 2.0/9.0
assert len(marvel_G) == 19255
assert graph.path(marvel_G, 'A', 'ZZZAX') == ['A', 'W2 159', 'WOLVERINE/LOGAN ', 'W2 41', 'SUMMERS, NATHAN CHRI', 'C2 59', 'ZZZAX']
assert 5.11 > graph.centrality(marvel_G, 'A') > 5.1
def genetic():
start_time = time.perf_counter()
# input_file = 'inputs/test.txt'
# input_file = 'inputs/test2.txt'
# input_file = 'inputs/test3.txt'
# input_file = 'inputs/slide.txt'
# input_file = 'inputs/graph2.txt'
# input_file = 'inputs/medium_graph.txt'
# input_file = 'inputs/big_graph.txt'
# input_file = 'instances/mono-objective/n10_ep0.2_d1'
# input_file = 'instances/bi-objective/n30_ep0.2_d2'
# input_file = 'instances/bi-objective/n30_ep0.5_d2'
# input_file = 'instances/bi-objective/n30_ep0.8_d2'
# input_file = 'instances/bi-objective/n40_ep0.2_d2'
input_file = 'instances/bi-objective/n40_ep0.5_d2'
# input_file = 'instances/bi-objective/n40_ep0.8_d2'
# input_file = 'instances/bi-objective/n50_ep0.5_d2'
# input_file = 'instances/bi-objective/n60_ep0.2_d2'
# input_file = 'instances/bi-objective/n100_ep0.2_d2'
# input_file = 'instances/bi-objective/n100_ep0.5_d2'
# input_file = 'instances/bi-objective/n100_ep0.8_d2'
graph, original_graph = read_graph(input_file)
input_name = os.path.basename(input_file) + '_{}'.format(
MAX_MINIBUCKET_VARIABLES)
order = get_variables_order(graph)
original_order = get_variables_order(original_graph)
solver = MiniBucket(order,
original_order,
MAX_MINIBUCKET_VARIABLES,
vertex_cover_cost,
debug=False)
solver.build_buckets()
print('Finished building minibuckets ({} max vars) in {:.3f}s\n'.format(
MAX_MINIBUCKET_VARIABLES,
time.perf_counter() - start_time))
try:
ga = NSGA2(order, solver)
ga.run()
except KeyboardInterrupt:
pass
print('Finished GA: {:.3f}s'.format(time.perf_counter() - start_time))
def main():
start_time = time.perf_counter()
# input_file = 'inputs/test.txt'
input_file = 'inputs/test2.txt'
# input_file = 'inputs/test3.txt'
# input_file = 'instances/mono-objective/n10_ep0.2_d1'
# input_file = 'instances/bi-objective/n30_ep0.2_d2'
# input_file = 'instances/bi-objective/n50_ep0.5_d2'
# input_file = 'instances/bi-objective/n100_ep0.2_d2'
# input_file = 'instances/bi-objective/n100_ep0.5_d2'
# input_file = 'instances/bi-objective/n100_ep0.8_d2'
# input_file = 'inputs/slide.txt'
# input_file = 'inputs/graph2.txt'
# input_file = 'inputs/medium_graph.txt'
# input_file = 'inputs/big_graph.txt'
graph, original_graph = read_graph(input_file)
input_name = os.path.basename(input_file) + '_{}'.format(
MAX_MINIBUCKET_VARIABLES)
print('Graph:')
print(graph)
print()
# order = get_variables_order(graph, heuristic=None)
# order = get_variables_order(graph, heuristic='custom', custom_order='ADBECF')
order = get_variables_order(graph)
original_order = get_variables_order(original_graph)
print('Order:')
print(order)
print()
solver = MiniBucket(order,
original_order,
MAX_MINIBUCKET_VARIABLES,
vertex_cover_cost,
debug=False)
solver.build_buckets()
print('Finished building minibuckets ({} max vars) in {:.3f}s\n'.format(
MAX_MINIBUCKET_VARIABLES,
time.perf_counter() - start_time))
assignment = [1, 0, 1, 1]
cost, best_next_assignment = solver.compute_cost(assignment)
new_next = solver.get_best_next(assignment)
print(cost, best_next_assignment, new_next)
def test_graph():
flights = [("ORD", "SEA"), ("ORD", "LAX"), ('ORD', 'DFW'), ('ORD', 'PIT'),
('SEA', 'LAX'), ('LAX', 'DFW'), ('ATL', 'PIT'), ('ATL', 'RDU'),
('RDU', 'PHL'), ('PIT', 'PHL'), ('PHL', 'PVD')]
G = {}
for (x, y) in flights:
graph.make_link(G, x, y)
marvel_G = graph.read_graph("resources/marvel_graph.tsv")
assert graph.clustering_coefficient(G) == 2.0 / 9.0
assert len(marvel_G) == 19255
assert graph.path(marvel_G, 'A', 'ZZZAX') == [
'A', 'W2 159', 'WOLVERINE/LOGAN ', 'W2 41', 'SUMMERS, NATHAN CHRI',
'C2 59', 'ZZZAX'
]
assert 5.11 > graph.centrality(marvel_G, 'A') > 5.1
def train(idx):
np.random.seed()
torch.manual_seed(idx)
test_graphs = [
read_graph("data/random/{}_{}".format(test_graph, i)).adj
for i in range(5)
]
gnn = GIN3(layer_num=layer_num, feature=feature)
gnn.to(device)
trainer = MCTSTrainer(gnn, test_graphs, "{}_{}th".format(file_prefix, idx))
Timer.start('all')
for i in range(epoch):
print("epoch: ", i)
graph = generate_random_graph(node, edge).adj
Timer.start('test')
trainer.test()
Timer.end('test')
Timer.start('train')
tmp = 0.01**(1 / epoch)
# 10 * tmp^epoch ~= 0.1
if train_method == "train1":
trainer.train1(graph, 10 * tmp**i, iter_p=iter_p)
elif train_method == "train2":
trainer.train2(graph, 10 * tmp**i, iter_p=iter_p)
else:
print("no such method")
assert False
Timer.end('train')
Timer.start('test')
trainer.test()
Timer.end('test')
Timer.end('all')
Timer.print()
Counter.print()
trainer.save_model()
trainer.save_test_result()
from config import device
import numpy as np
import torch
from utils.graph import read_graph, generate_random_graph
from mcts.mcts import MCTS
from mcts.mcts_trainer import MCTSTrainer
from gin.gin import GIN3
from utils.timer import Timer
from utils.counter import Counter
if __name__ == "__main__":
test_graphs = [
read_graph("data/random/100_250_{}".format(i)).adj for i in range(5)
]
gnn = GIN3(layer_num=6)
gnn.to(device)
trainer = MCTSTrainer(gnn, test_graphs, "train2_p2_0th")
Timer.start('all')
for i in range(100):
print("epoch: ", i)
graph = generate_random_graph(100, 250).adj
Timer.start('train')
trainer.train2(graph, 10 * 0.96**i, iter_p=2)
Timer.end('train')
Timer.start('test')
trainer.test()
Timer.end('test')
Timer.start('all')
result = mcts.search_for_exp(graph, time_limit=10 * 60, min_iter_num=100)
print("graph: {}, result: {}".format(name, result))
print("max: ", max(result))
Timer.end('all')
Timer.print()
return max(result)
if __name__ == "__main__":
gnns = best_gins()
filename = "random/10_25_0"
graph = read_graph("data/" + filename).adj
print(filename)
results = {}
pool = Pool()
results = pool.map(use_model, [(gnn, filename, graph) for gnn in gnns])
pool.close()
pool.join()
print(results)
print("file name: {} final max: {}".format(filename, max(results)))
# for gnn in gnns:
# use_model(gnn, filename, graph)