def accept(ini_solution, ini_solution_fmed, new_solution, t, data):
pre = ini_solution_fmed
post = opt_fmed(new_solution, data)
p = np.exp((-(post - pre) / t))
if post < pre:
return new_solution, post
elif np.random.random() < p:
return new_solution, post
else:
return ini_solution, pre
def local_best_first_search(solution, solution_fmed, data, time_limit):
"""
Búsqueda local del primer mejor.
"""
t_end = time.time() + time_limit
best_solution = solution
best_fmed = solution_fmed
while time.time() < t_end:
target_fmed = best_fmed
for neighbour in generate_neighbours(best_solution):
neighbour_fmed = opt_fmed(neighbour, data)
if neighbour_fmed < best_fmed:
best_solution, best_fmed = neighbour, neighbour_fmed
if target_fmed == best_fmed:
break
return best_solution, best_fmed
def local_best_search(solution, solution_fmed, data, time_limit):
"""
Búsqueda local del mejor.
"""
best_fmed = solution_fmed
best_solution = solution
t_end = time.time() + time_limit
while time.time() < t_end:
for neighbour in generate_neighbours(best_solution):
if time.time() < t_end:
neighbour_fmed = opt_fmed(neighbour, data)
if neighbour_fmed <= best_fmed:
best_solution, best_fmed = neighbour, neighbour_fmed
else:
return best_solution, best_fmed
return best_solution, best_fmed
def simulated_annealing(tries, t_ini_factor, alpha, solution, time_, data):
# Bucle de 60 segundos
t_end = time.time() + time_ - 0.1
solution_fmed = opt_fmed(solution, data)
#t = solution_fmed * t_ini_factor
t = t_ini_factor
while time.time() < t_end:
for i in range(tries): # Tries before getting t down
vecino = get_neighbour(solution)
solution, solution_fmed = accept(solution, solution_fmed, vecino,
t, data)
step = t * alpha / 100 # Baja la temperatura un alpha%
if t - step > 0.1:
t -= step
return solution, solution_fmed
Reordena una matriz de costes (f) para comparar mi solución con las de los casos de prueba.
"""
ordered_mat = np.zeros_like(cost_matrix)
i = 0
for step in sol:
ordered_mat[step] = cost_matrix[i]
i += 1
return ordered_mat
# timeit.timeit('read_file("ProblemasFlowShopPermutacional/Doc11.txt")', setup="from __main__ import read_file", number=100000)
sol_doc1_mat = np.loadtxt("Tests/SolDoc1_Mat.txt", usecols=range(5), dtype=int)
sol_doc1_vec = np.loadtxt("Tests/SolDoc1_Vec.txt", usecols=range(11), dtype=int)
sol_doc2_mat = np.loadtxt("Tests/SolDoc2_Mat.txt", usecols=range(4), dtype=int)
sol_doc2_vec = np.loadtxt("Tests/SolDoc2_Vec.txt", usecols=range(13), dtype=int)
sol_doc1_vec = [a-1 for a in sol_doc1_vec]
sol_doc2_vec = [a-1 for a in sol_doc2_vec]
doc1 = read_file("Datasets/Doc1.txt")
doc2 = read_file("Datasets/Doc2.txt")
print(reorder(opt_fmed(sol_doc1_vec, doc1)[1], sol_doc1_vec) == sol_doc1_mat)
print(reorder(opt_fmed(sol_doc2_vec, doc2)[1], sol_doc2_vec) == sol_doc2_mat)
print(reorder(opt_fmed(sol_doc1_vec, doc1)[1], sol_doc1_vec))
print(sol_doc1_mat)
print(reorder(opt_fmed(sol_doc2_vec, doc2)[1], sol_doc2_vec))
print(sol_doc2_mat)