class GENEALG:
def __init__(self, play_back=False):
if play_back:
f = open('robot.p', 'rb')
self.parent = pickle.load(f)
f.close()
else:
self.parent = POPULATION(c.popSize)
self.parent.Initialize()
self.parent.Evaluate()
def Evolve(self):
for g in range(c.numGens):
print(g, end=' ')
self.parent.Print()
child = POPULATION(c.popSize)
child.Fill_From(self.parent)
child.Evaluate()
self.parent.ReplaceWith(child)
def Show_Best(self):
bestIndividual = self.parent.Best_Individual()
self.parent.p[bestIndividual].Start_Evaluation(pb=False, pp=True)
self.parent.p[bestIndividual].Compute_Fitness()
print(self.parent.p[bestIndividual].fitness)
def Save(self):
f = open('robot.p', 'wb')
pickle.dump(self.parent, f)
f.close()
from population import POPULATION
import copy
parents = POPULATION(10)
parents.Initialize() # 09
parents.Evaluate(pp=False, pb=True)
parents.Print()
for g in range(1, 200):
children = copy.deepcopy(parents)
children.Mutate()
children.Evaluate(pp=False, pb=True)
parents.ReplaceWith(children)
#parents.Sort()
print g,
parents.Print()
parents.Evaluate(pp=True, pb=False)
import copy
from robot import ROBOT
import pickle
from population import POPULATION
from individual import INDIVIDUAL
import constants as c
from environments import ENVIRONMENTS
from environment import ENVIRONMENT
envs = ENVIRONMENTS()
parents = POPULATION(c.popSize)
parents.Initialize()
parents.Evaluate(envs, pp=False, pb=True)
#
parents.Print()
for g in range(1, c.numGens):
children = POPULATION(c.popSize)
children.Fill_From(parents)
children.Evaluate(envs, pp=False, pb=True)
print(g),
children.Print(),
print(),
parents.ReplaceWith(children)
if (g == c.numGens - 1):
print(parents.p[0].fitness)
for e in range(0, c.numEnvs):
parents.p[0].Start_Evaluation(envs.envs[e], pp=False, pb=False)
import pyrosim
import matplotlib.pyplot as plt
import random
import numpy
import copy
import pickle
from individual import INDIVIDUAL
from robot import ROBOT
from population import POPULATION
parents = POPULATION(
10) # creates a empty population but stores the population size
parents.Initialize() #start a population of individuals of size 'i'
parents.Evaluate(True) # Evaluates the fitness of each individual
#exit() # exit the program immediately
parents.Print() #print ID and fitness of each individual
for i in range(1, 200):
children = POPULATION(5)
#children.Print()
children.Fill_From(parents)
#children.Print()
children.Evaluate(True)
children.Print()
parents = children
parents.Evaluate(False)
parents.Print()
from robot import ROBOT
from individual import INDIVIDUAL
from population import POPULATION
import pyrosim
import random
import copy
import pickle
pop = POPULATION(5)
pop.Evaluate()
pop.Print()
# parent = INDIVIDUAL()
# parent.Evaluate(True)
# print(parent.fitness)
# for i in range (0, 500):
# child = copy.deepcopy( parent )
# child.Mutate()
# child.Evaluate(True)
# print('[g:' ,i , '] [p:' ,parent.fitness , ']', '[c:' ,child.fitness , ']')
# if ( child.fitness > parent.fitness ):
# parent = child
# parent.Evaluate(True)
# f = open('robot.p','wb')
# pickle.dump(parent , f )
# f.close()
fitnesses[0] = highest highest = -100 for g in range(1,c.numGens): children = POPULATION(c.popSize) #FOR RANDOM SEARCH ONLY, COMMENT OUT OTHERWISE #for i in parents.p: #parents.p[i].fitness = 0 children.Fill_From(parents) children.Evaluate(envs, pp=False, pb=True) for i in children.p: if (children.p[i].fitness > highest): highest = children.p[i].fitness fitnesses[g] = highest highest = -100 print(g) children.Print() parents.ReplaceWith(children) if(g == c.numGens-1): for e in range(0, c.numEnvs): children.p[0].Start_Evaluation(envs.envs[e],pp=True, pb=False) low = fitnesses[0] hi = fitnesses[c.numGens-2] print(fitnesses) f = plt.figure() panel = f.add_subplot(111) panel.set_xlim(0, 100) panel.set_ylim(0, 1.2) #panel.set_xlim(0, c.numGens) #panel.set_ylim(low, hi) plt.plot(fitnesses) plt.show()
file_2.close()
file_3.close()
# skip parent evaluation, child evolution, and saving in movie mode
if (not c.movieMode):
#---EVALUATE PARENTS---#
# evaulate parents in respective environments types
parents_1.Evaluate(envs_1, True, pp=False)
parents_2.Evaluate(envs_2, True, pp=False)
parents_3.Evaluate(envs_3, True, pp=False)
# print results for each parent population
print("\nfitness of parents 1 in envs 1:"),
parents_1.Print()
print("\nfitness of parents 2 in envs 2:"),
parents_2.Print()
print("\nfitness of parents 3 in envs 3:"),
parents_3.Print()
print("\n")
###################
# EVOLVE CHILDREN #
###################
startTime = time.time()
g = 0
while (time.time() - startTime < c.runTime):
# initialize children
children_1 = POPULATION(c.popSize)
import copy
import pickle
from individual import INDIVIDUAL
from population import POPULATION
# import matplotlib.pyplot as plt
parent = POPULATION(5)
parent.Evaluate()
for g in range(100):
print(g, end=' ')
parent.Print()
child = copy.deepcopy(parent)
child.Mutate()
child.Evaluate()
parent.ReplaceWith(child)
# parent = INDIVIDUAL()
# parent.Evaluate(False)
# for g in range(0, 100):
# child = copy.deepcopy(parent)
# child.Mutate()
# child.Evaluate(True)
# print('[g: ', g, ' ] ', '[pw: ', parent.genome, ' ] ',
# '[p: ', parent.fitness, ' ] ', '[c: ', child.fitness, ' ]')
# if child.fitness > parent.fitness:
# parent = child
# parent.fitness = child.fitness
# parent.Evaluate(False)
# f = open('robot.p', 'wb')
# pickle.dump(parent, f)
# f.close()
from population import POPULATION
import copy
import constants as c
from environments import ENVIRONMENTS
import pickle
envs = ENVIRONMENTS()
parents = POPULATION(c.popSize)
parents.Initialize()
parents.Evaluate(envs, True, True, c.evalTime)
parents.Print()
parents2 = POPULATION(c.popSize)
parents2.Initialize2()
parents2.Evaluate(envs, True, True, c.evalTime)
parents2.Print()
for g in range(1, c.numGens):
children = POPULATION(c.popSize)
children.Fill_From(parents)
# children.Fill_From2(parents)
children.Evaluate(envs, False, True, c.evalTime)
parents.ReplaceWith(children)
# parents.ReplaceWith2(children)
print('Whegged: '),
print(g),
children.Print()
children2 = POPULATION(c.popSize)
children2.Fill_From(parents2)
