def search_es(max_gens, search_space, pop_size, num_children, prefix):
f, tmpdir = init_search(prefix,names)
population = es.init_population(search_space, pop_size, id=True)
cost = run_particles(population,tmpdir+'/'+prefix)
for p in population:
p["fitness"] = cost[p["id"]]
best = sorted(population, key=lambda k: k["fitness"])[0]
try:
for gen in range(max_gens):
children = [es.mutate(population[i], search_space, id=i) for i in range(num_children)]
cost = run_particles(children,tmpdir+'/'+prefix)
for c in children:
c["fitness"] = cost[c["id"]]
union = children+population
union = sorted(union, key=lambda k: k["fitness"])
if union[0]["fitness"] < best["fitness"]: best = union[0]
population = union[:pop_size]
f.write("%i %.6g " % (gen, best["fitness"]))
[f.write("%.6g "%p) for p in best["pos"]]
f.write("\n")
print " > gen %i, fitness=%.4g" % (gen, best["fitness"])
print [names[i] + ": %.6g " % p for i,p in enumerate(best["pos"])]
except KeyboardInterrupt:
print "\033[91mInterruption by user. Exiting...\033[0m"
pass
f.close()
return best
def search_pso(max_gens, search_space, pop_size, c1, c2, prefix):
f, tmpdir = init_search(prefix,names)
vel_space = pso.generate_vel_space(search_space,frac=1)
pop = [pso.create_particle(search_space, vel_space, id=i) for i in range(pop_size)]
cost = run_particles(pop,tmpdir+'/'+prefix)
for p in pop:
p["cost"] = cost[p["id"]]
p["b_cost"] = p["cost"]
gbest = pso.get_global_best(pop)
part_best = {}
part_best["pos"] = gbest["pos"]
part_best["cost"] = gbest["cost"]
try:
for gen in range(max_gens):
for particle in pop:
pso.update_velocity(particle, gbest, vel_space)
pso.update_position(particle, search_space)
cost = run_particles(pop,tmpdir+'/'+prefix)
for particle in pop:
particle["cost"] = cost[particle["id"]]
pso.update_best_position(particle)
gbest = pso.get_global_best(pop, gbest)
if gbest["cost"] < part_best["cost"]:
part_best["pos"] = list(gbest["pos"])
part_best["cost"] = gbest["cost"]
f.write("%i %.6g " % (gen, part_best["cost"]))
[f.write("%.6g "%p) for p in part_best["pos"]]
f.write("\n")
print " > gen %i, fitness=%.6g" % (gen, part_best["cost"])
print [names[i] + ": %.6g " % p for i,p in enumerate(part_best["pos"])]
except KeyboardInterrupt:
print "\033[91mInterruption by user. Exiting...\033[0m"
pass
f.close()
return part_best
def search_hs(search_space, max_iter, mem_size, consid_rate, adjust_rate, nrange, prefix):
f, tmpdir = init_search(prefix,names)
memory = [hs.create_random_harmony(search_space,id=i) for i in range(mem_size*3)]
cost = run_particles(memory,tmpdir+'/'+prefix)
for m in memory:
m["fitness"] = cost[m["id"]]
memory = sorted(memory, key=lambda k: k["fitness"])[:mem_size]
best = memory[0]
try:
for iter in range(max_iter):
harm = hs.create_harmony(search_space, memory, consid_rate, adjust_rate, nrange)
harm["id"] = 0
cost = run_particles([harm],tmpdir+'/'+prefix)
harm["fitness"] = cost[harm["id"]]
if harm["fitness"] < best["fitness"]: best = harm
memory.append(harm)
memory = sorted(memory, key=lambda k: k["fitness"])
del memory[-1]
f.write("%i %.6g " % (iter, best["fitness"]))
[f.write("%.6g "%p) for p in best["pos"]]
f.write("\n")
print " > iteration %i, fitness=%.4g" % (iter, best["fitness"])
print [names[i] + ": %.6g " % p for i,p in enumerate(best["pos"])]
except KeyboardInterrupt:
print "\033[91mInterruption by user. Exiting...\033[0m"
pass
f.close()
return best
def search(search_space, pop_size, prefix):
f, tmpdir = init_search(prefix,names)
#pop = init_population_random(search_space, pop_size)
pop = init_population_lhs(search_space, pop_size)
try:
cost = run_particles(pop,tmpdir+'/'+prefix)
for i,candidate in enumerate(pop):
candidate["cost"] = cost[candidate["id"]]
f.write("%.6g " % candidate["cost"])
[f.write("%.6g "%p) for p in candidate["pos"]]
f.write("\n")
#print candidate["pos"]
except KeyboardInterrupt:
print "\033[91mInterruption by user. Exiting...\033[0m"
pass
f.close()
def search_rs(search_space, max_iter, prefix):
f, tmpdir = init_search(prefix,names)
pop = []
for i in range(max_iter):
candidate = {}
candidate["id"] = i
candidate["pos"] = rs.random_vector(search_space)
pop.append(candidate)
best = None
cost = run_particles(pop,tmpdir+'/'+prefix)
for i,candidate in enumerate(pop):
candidate["cost"] = cost[candidate["id"]]
if best is None or candidate["cost"] < best["cost"]: best = candidate
f.write("%i %.6g " % (i, best["cost"]))
[f.write("%.6g "%p) for p in best["pos"]]
f.write("\n")
print " > iteration %i, best=%.4g" % (i, best["cost"])
print [names[i] + ": %.6g " % p for i,p in enumerate(best["pos"])]
f.close()
return best
def search_ca(max_gens, search_space, pop_size, num_accepted, prefix):
f, tmpdir = init_search(prefix,names)
# initialize
pop = [{"pos" : ca.random_vector(search_space) , "id" : i } for i in range(pop_size)]
belief_space = ca.initialize_beliefspace(search_space)
# evaluate
cost = run_particles(pop,tmpdir+'/'+prefix)
for p in pop:
p["fitness"] = cost[p["id"]]
best = sorted(pop, key=lambda k: k["fitness"])[0]
# update situational knowledge
ca.update_beliefspace_situational(belief_space, best)
try:
for gen in range(max_gens):
# create next generation
children = [ ca.mutate_with_inf(pop[i], belief_space, search_space, id=i) for i in range(pop_size)]
# evaluate
cost = run_particles(children,tmpdir+'/'+prefix)
for c in children:
c["fitness"] = cost[c["id"]]
best = sorted(children, key=lambda k: k["fitness"])[0]
# update situational knowledge
ca.update_beliefspace_situational(belief_space, best)
# select next generation
pop = [ca.binary_tournament(children + pop) for i in range(pop_size)]
# update normative knowledge
pop = sorted(pop, key=lambda k: k["fitness"])
acccepted = pop[0:num_accepted]
ca.update_beliefspace_normative(belief_space, acccepted)
# user feedback
f.write("%i %.6g " % (gen, belief_space["situational"]["fitness"]))
[f.write("%.6g "%p) for p in belief_space["situational"]["pos"]]
f.write("\n")
print " > gen %i, f=%.4g" % (gen, belief_space["situational"]["fitness"])
print [names[i] + ": %.6g " % p for i,p in enumerate(belief_space["situational"]["pos"])]
except KeyboardInterrupt:
print "\033[91mInterruption by user. Exiting...\033[0m"
pass
f.close()
return belief_space["situational"]