def total_crossover(pop):
dims = len(pop[0].genes)
psize = len(pop)
parent1 = pop[np.random.permutation(psize)]
parent2 = pop[np.random.permutation(psize)]
offsprings = np.ndarray(psize, dtype=object)
tmp = np.random.rand(psize, dims)
p1_selection = tmp >= 0.5
p2_selection = tmp < 0.5
for i in range(psize):
offsprings[i] = Chromosome(dims)
offsprings[i].genes[p1_selection[i]] = parent1[i].genes[
p1_selection[i]]
offsprings[i].genes[p2_selection[i]] = parent2[i].genes[
p2_selection[i]]
return offsprings
def crossover(p1, p2, alpha=np.random.rand(), ctype='uniform'):
if ctype == 'convex':
return Chromosome(alpha * p1.genes + (1 - alpha) * p2.genes)
# elif stype=='uniform':
# return max(np.random.choice(self.pop,k,False),key=lambda c:c.fitness)
else:
raise ValueError('Type of crossover which you entered is wrong')
def initialize(self):
gen_id = 0
if self.settings.get('evolution',
{}).get('checkpoint',
{}).get('initialize_enabled'):
gen_id = int(self.initialize_from_save()) + 1
else:
pop = []
for i in range(
0,
self.settings.get('evolution', {}).get('population_size')):
c = Chromosome(self.settings)
pop.append(c)
self.pop = pop
return gen_id
def get_pop_init(n, gn, init_func=np.random.rand, p_type='knapsack'):
"""[Initial Population Generator]
Arguments:
n {[int]} -- [Number of members of population]
gn {[int]} -- [Number of individual's genes]
init_func {[function]} -- the function which initialize each chromosome
Returns:
[np.ndarray] -- [Array of chromosomes]
"""
if p_type == 'knapsack':
return np.array([Chromosome(gn, init_func) for _ in range(n)])
elif p_type == 'double_pole':
return np.array([ChromosomePole(gn, init_func) for _ in range(n)])