def search(knapsack: Knapsack, max_nb_eval: int) -> Tuple[float, int]:
base_solution = random_solution(knapsack)
base_score = knapsack.evaluate(base_solution)
nb_eval = 1
voisins_possibles = list(range(knapsack.items))
while True:
shuffle(voisins_possibles)
current_score = base_score
index_voisin = -1
while (index_voisin + 1 < len(voisins_possibles)
and current_score <= base_score and nb_eval <= max_nb_eval):
index_voisin += 1
index = voisins_possibles[index_voisin]
base_solution[index] = not base_solution[index]
current_score = knapsack.evaluate(base_solution)
nb_eval += 1
base_solution[index] = not base_solution[index]
if current_score > base_score:
base_score = current_score
index = voisins_possibles[index_voisin]
base_solution[index] = not base_solution[index]
elif index_voisin + 1 == len(voisins_possibles):
break # optimum local
if nb_eval >= max_nb_eval:
break
return base_score, nb_eval
def search(knapsack: Knapsack, max_evals: int) -> Tuple[float, int]:
"""Use the hill climber best improvement to find a solution."""
base_solution = random_solution(knapsack)
base_score = knapsack.evaluate(base_solution)
nb_eval = 1
while True:
best_score = -math.inf
index_solution = -1
for i in range(knapsack.items):
base_solution[i] = not base_solution[i]
score = knapsack.evaluate(base_solution)
nb_eval += 1
if score > best_score:
best_score = score
index_solution = i
base_solution[i] = not base_solution[i]
if nb_eval >= max_evals:
break
if best_score > base_score:
base_solution[index_solution] = not base_solution[index_solution]
base_score = best_score
else:
break # Optimum local
return base_score, nb_eval
def __init__(self, mu: int, knapsack: Knapsack):
self.knapsack = knapsack
parents = [random_solution(knapsack) for i in range(mu)]
self._population = [
Individu(parent, knapsack.evaluate(parent)) for parent in parents
]
self.mu = mu