def _select(self, pop):
node = Node(self.problem, pop, self.settings.pop_size).divide(sqrt(pop))
non_dom_leafs = node.nonpruned_leaves()
all_leafs = node.leaves()
# Counting number of evals
evals = 0
for leaf in all_leafs:
for row in leaf._pop:
if row.evaluated:
evals+=1
return non_dom_leafs, evals
def _select(self, pop):
node = Node(self.problem, pop,
self.settings.pop_size).divide(sqrt(pop))
non_dom_leafs = node.nonpruned_leaves()
all_leafs = node.leaves()
# Counting number of evals
evals = 0
for leaf in all_leafs:
for row in leaf._pop:
if row.evaluated:
evals += 1
return non_dom_leafs, evals
def run(algo, id=0):
gen = 0
best_solutions = []
size = algo.settings.pop_size
max_gens = per_core(algo.settings.gens)
population = Node.format(algo.problem.populate(size))
total_evals = 0
while gen < max_gens:
say(".")
selectees, evals = algo.select(population)
solutions, evals = algo.get_best(selectees)
best_solutions += solutions
total_evals += evals
# EVOLUTION
selectees, evals = algo.evolve(selectees)
total_evals += evals
population, evals = algo.recombine(selectees,
algo.settings.pop_size)
total_evals += evals
gen += 1
if RANK == 0:
for i in range(1, SIZE):
best_solutions += COMM.recv(source=i, tag=id)
return best_solutions
else:
COMM.send(best_solutions, dest=0, tag=id)
def run(self, init_pop=None):
gen = 0
best_solutions = []
max_gens = self.settings.max_gens
population = init_pop
total_evals = 0
try:
if population is None:
population = Node.format(self.problem.populate(self.settings.pop_size))
while gen < max_gens:
say(".")
selectees, evals = self.select(population)
total_evals += evals
solutions, evals = self.get_best(selectees)
best_solutions.append(solutions)
total_evals += evals
# EVOLUTION
selectees, evals = self.evolve(selectees)
total_evals += evals
population, evals = self.recombine(selectees, self.settings.pop_size)
total_evals += evals
gen += 1
except RuntimeException as e:
e.print_trace()
return best_solutions, total_evals
return best_solutions, total_evals
def run(algo, id = 0):
gen = 0
best_solutions = []
size = algo.settings.pop_size
max_gens = per_core(algo.settings.gens)
population = Node.format(algo.problem.populate(size))
total_evals = 0
while gen < max_gens:
say(".")
selectees, evals = algo.select(population)
solutions, evals = algo.get_best(selectees)
best_solutions += solutions
total_evals += evals
# EVOLUTION
selectees, evals = algo.evolve(selectees)
total_evals += evals
population, evals = algo.recombine(selectees, algo.settings.pop_size)
total_evals += evals
gen += 1
if RANK == 0:
for i in range(1, SIZE):
best_solutions += COMM.recv(source=i, tag = id)
return best_solutions
else:
COMM.send(best_solutions, dest=0, tag = id)
def _recombine(self, mutants, total_size):
remaining = total_size - len(mutants)
pop = []
for _ in range(remaining):
pop.append(self.problem.generate())
return mutants + Node.format(pop), 0
def _recombine(self, mutants, total_size):
remaining = total_size - len(mutants)
pop = []
for _ in range(remaining):
pop.append(self.problem.generate())
return mutants + Node.format(pop), 0
