def exact(filename, max_seconds=None):
original_problem_graph = TSPProblemParser().parseDirected(filename)
solution, cost = tsp_mip_cutting_planes.TSP().load_problem(
original_problem_graph).build().solve(max_seconds)
# print(f'solution = {solution}, cost = {cost}')
return solution, cost
def local_search_augmented(filename,
batch_size=8,
n_iters=30,
max_attemps=4,
max_seconds=None):
original_problem_graph = TSPProblemParser().parseDirected(filename)
solution, cost = LocalSearchAugmented().solve(original_problem_graph,
batch_size, n_iters,
max_attemps, max_seconds)
print(f'solution = {solution}, cost = {cost}')
return solution, cost
def reduced_edges_aproximation(filename, max_seconds=None):
original_problem_graph = TSPProblemParser().parse(filename)
solution_mst, cost = MST().solve(original_problem_graph)
solution_nn, cost = NN().solve_complete(original_problem_graph)
reduced_graph = nx.operators.compose(
get_reduced_graph(solution_mst, original_problem_graph, True),
get_reduced_graph(solution_nn, original_problem_graph, True))
graph = convert_to_digraph(original_problem_graph)
solution, cost = tsp_mip_cutting_planes.TSP().load_problem(
reduced_graph).build().solve(max_seconds)
# print(f'solution = {solution}, cost = {cost}')
return solution, cost
def planar_aproximation_2(filename, max_seconds=None):
original_problem_graph = TSPProblemParser().parse(filename)
planar_graph = PMFG().get_reduced_graph(original_problem_graph)
nn_algo = NN()
solution_nn, cost = nn_algo.solve_complete(original_problem_graph)
all_nn_graph = get_reduced_graph(solution_nn, original_problem_graph, True)
for s in nn_algo.solutions:
all_nn_graph = nx.operators.compose(
all_nn_graph, get_reduced_graph(s, original_problem_graph, True))
planar_graph = convert_to_digraph(planar_graph)
graph = convert_to_digraph(original_problem_graph)
reduced_graph = nx.operators.compose(planar_graph, all_nn_graph)
solution, cost = tsp_mip_cutting_planes.TSP().load_problem(
reduced_graph).build().solve(max_seconds)
# print(f'solution = {solution}, cost = {cost}')
return solution, cost
solution = []
cost = 0
mst = self.get_mst(problem_graph)
odd_graph = self.build_graph_with_odd_vertex(problem_graph, mst)
matching_edges = nx.algorithms.matching.max_weight_matching(odd_graph)
mst_extended = self.join_mst_and_matching_edges(
problem_graph, mst, matching_edges)
walk = list(nx.eulerian_circuit(mst_extended))
solution = self.fast_hamiltonian_circuit_from_euler_circuit(walk)
for i in range(len(solution)):
u = solution[i]
v = solution[(i + 1) % len(solution)]
weight = problem_graph[u][v]['weight']
cost += weight
return solution, cost
if __name__ == '__main__':
filename = 'examples/ulysses16.json'
original_problem_graph = TSPProblemParser().parse(filename)
solution, cost = Christofides().solve(original_problem_graph)
print(cost)
def nn_aproximation(filename):
original_problem_graph = TSPProblemParser().parse(filename)
solution, cost = NN().solve_complete(original_problem_graph)
# print(f'solution = {solution}, cost = {cost}')
return solution, cost
def christofides_aproximation(filename):
original_problem_graph = TSPProblemParser().parse(filename)
solution, cost = Christofides().solve(original_problem_graph)
# print(f'solution = {solution}, cost = {cost}')
return solution, cost