for t in time_LW:
nn.step(
np.concatenate((np.zeros(3), np.zeros(4), body.state())))
body.step(stepsize_LW,
nn.output() + np.random.normal(0.0, noisestd))
#body.step(stepsize_LW, nn.output())
#body.step(stepsize_LW, np.concatenate(((nn.output() + np.random.normal(0.0,noisestd)),np.zeros(2))))
fit += body.cx / duration_LW
fitness3 = (fit / total_trials_LW) / MaxFit
np.save('Fitness_legged', fitness3)
#np.load('Fitness_legged.npy')
return fitness1 * fitness2 * fitness3
# Evolve and visualize fitness over generations
ga = mga.Microbial(fitnessFunction, popsize, genesize, recombProb, mutatProb,
demeSize, generations, boundaries)
ga.run(tournaments)
ga.showFitness()
# Get best evolved network and show its activity
af, bf, bi = ga.fitStats()
#print(ga.avgHistory)
#print(ga.bestHistory)
ah = ga.avgHistory
bh = ga.bestHistory
np.save('average_history', ah)
np.load('average_history.npy')
np.save('best_history', bh)
np.load('best_history.npy')
# -*- coding: utf-8 -*-
"""
Created on Tue Oct 1 16:13:40 2019
@author: benso
"""
import mga
import numpy as np
population_size = 20
genotype = 10
#total_pop = np.zeros((population_size, genotype))
transfectprob = 0.5
mutationprob = 0.5
generations = 5
num_tournaments = generations * population_size
def fitness_func(genotype):
fitness_scores = np.sum(genotype)
return fitness_scores
ga = mga.Microbial(fitness_func, population_size, genotype, transfectprob,
mutationprob)
ga.run(num_tournaments)
ip_count += 1
return ip_count
# EA Params
popsize = 10
genesize = nnsize * nnsize + 2 * nnsize
recombProb = 0.5
mutatProb = 0.1
generations = 50
tournaments = generations * popsize
# Evolve and visualize fitness over generations
ga = mga.Microbial(fitnessFunction, popsize, genesize, recombProb, mutatProb)
ga.run(tournaments)
ga.showFitness()
# Get best evolved network and show its activity
af, bf, bi = ga.fitStats()
time = np.arange(0.0, duration, stepsize)
nn = ctrnn.CTRNN(nnsize)
nn.setParameters(
ga.pop[bi], WeightRange, BiasRange, TimeConstMin, TimeConstMax
) #edited to give genotype of bi rather than just bi, which is an address in ga.pop
nn.initializeState(np.zeros(nnsize))
outputs = np.zeros((len(time), nnsize))
step = 0
for t in time:
nn.step(stepsize)
t1 = (t1 + 1) % n
t2 = (t2 + 1) % n
avg_dist = cum_dist / (n * numpy.sqrt(dim))
return 1 - avg_dist
# EA Params
popsize = 500
demesize = 5
genesize = nnsize * nnsize + 2 * nnsize
recombProb = 0.5
mutatProb = 1 / genesize
generations = 1500
# Evolve and visualize fitness over generations
ga = mga.Microbial(large_n_fitness, popsize, genesize, recombProb, mutatProb,
demesize, generations, 1)
ga.run()
#ga.showFitness()
af, bf, bi = ga.fitStats()
#id = int(sys.argv[1])
numpy.save("avg_fitness_{0}_{1}.npy".format(targ, num),
ga.avgHistory) #"avg_fitness_"+str(id)+".npy"
numpy.save("best_fitness_{0}_{1}.npy".format(targ, num),
ga.bestHistory) #"best_fitness_"+str(id)+".npy"
numpy.save("best_individual_{0}_{1}.npy".format(targ, num),
bi) #"best_individual_"+str(id)+".npy"
#avg_fit = open("avg_fitness_{0}.npy".format(num), 'rb')
#numpy.save(avg_fit, ga.avgHistory)
#avg_fit.flush()
