def generate_new_population(population, weights, size, mutation_rate=0.1):
dna_set = [1, 2, 3, 4, 5, 6, 7, 8]
new_population = []
for _ in range(size):
ind_x = random_selection(population, weights)
ind_y = random_selection(population, weights)
xy = reproduce(ind_x, ind_y)
mutated_xy = mutation(xy, dna_set, mutation_rate)
new_population.append(mutated_xy)
return new_population
temp=pd.DataFrame([[interations,MinFunctionValue,AverageFunctionValue,MaxFitness,AverageFitness,BestIndex]], columns=metaDfVariableHeadings)
metaDf = pd.concat([metaDf,temp])
# print(metaDf,"meta")
#'BestFitnessValue','AverageFitness','BestIndexNumber']
# plotting
# metaDf.plot(, ax = ax) # Pass the axes to plot.
# print("max fitness value",df['FitnessValue'].idxmax())
mating_pool = utils.suss(fitness_scores,population,listOfVarStringLengths)
# selection - mating pool generation
# crossover of mating pool
utils.generalCrossover(crossoverProbability,mating_pool,utils.singlePointCrossover,listOfVarStringLengths,listOfPrecisionDigits)
# mutation of new mating pool
utils.mutation(mating_pool,mutationProbability,listOfVarStringLengths,listOfPrecisionDigits)
population = mating_pool
interations = interations + 1
f1 = plt.figure(1)
ax = plt.gca()
metaDf.plot(kind='line',x='GenerationNumber',y='AverageFunctionValue',color='blue',ax=ax)
metaDf.plot(kind='line',x='GenerationNumber',y='MinFunctionValue',color='red',ax=ax)
f2 = plt.figure(2)
ax = plt.gca()
metaDf.plot(kind='line',x='GenerationNumber',y='AverageFitness',color='blue',ax=ax)
metaDf.plot(kind='line',x='GenerationNumber',y='BestFitnessValue',color='red',ax=ax)
plt.show()
def LL_static(initial_solution,
instance,
lda1,
lda2,
k,
c,
crossover_choice,
max_evaluation,
ioh_output=None):
x = initial_solution
n = len(x.chromossome)
T = 0
function_value = []
neval = []
counter = 0
f = None
if ioh_output is not None:
f = open(ioh_output, 'at')
f.write(
'"function evaluation" "original f(x)" "best original f(x)" "transformed f(x) " "best transformed f(x)" "mutation_rate" "crossover_bias"'
)
while (x.value != instance.optimum):
#---------- mutation phase ------------
#sample l from the binomial
p = k / float(n)
l = np.random.binomial(n, p)
while l == 0:
l = np.random.binomial(n, p)
x_prime = utils.mutation(x, l)
x_prime.value = getattr(evaluate, instance.name)(x_prime, instance)
T += 1
for i in range(lda1 - 1):
x2_prime = utils.mutation(x, l)
x2_prime.value = getattr(evaluate, instance.name)(x2_prime,
instance)
T += 1
if x2_prime.value > x_prime.value:
x_prime = x2_prime
#-------- crossover phase
y, Ty = LL_crossover_phase(x=x,
x_prime=x_prime,
instance=instance,
implementation_choice=crossover_choice,
chromossome_length=n,
n_offsprings=lda2,
crossover_bias=c)
T += Ty
#-------- selection
if y.value > x.value:
x = y
if f is not None:
ls = [T, x.value, x.value, x.value, x.value, p, c] # IOH output
s = ' '.join([str(val) for val in ls])
f.write('\n' + s)
function_value.append(x.value)
neval.append(T)
counter += 1
if max_evaluation > 0 and T >= max_evaluation:
break
#End while
if f is not None:
f.write('\n' + str(T))
f.close()
return (T, counter, function_value, neval)
def LL_dynamic_02(initial_solution,
instance,
alpha,
beta,
gamma,
a,
b,
crossover_choice,
max_evaluation,
ioh_output=None):
x = initial_solution
n = len(x.chromossome)
lda = float(1)
T = 0
ldas = []
function_value = []
neval = []
counter = 0
min_prob = 1.0 / n
max_prob = 0.99
f = None
if ioh_output is not None:
f = open(ioh_output, 'at')
f.write(
'"function evaluation" "original f(x)" "best original f(x)" "transformed f(x) " "best transformed f(x)" "mutation_rate" "crossover_bias" "n_mutations" "n_offsprings" "iteration"'
)
while (x.value != instance.optimum):
p = max([min([alpha * lda / float(n), max_prob]), min_prob])
lda1 = int(round(lda))
lda2 = int(round(beta * lda))
c = max([min([gamma / lda, max_prob]), min_prob])
#----------- mutation phase
l = np.random.binomial(n, p)
while l == 0:
l = np.random.binomial(n, p)
x_prime = utils.mutation(x, l)
x_prime.value = getattr(evaluate, instance.name)(x_prime, instance)
T += 1
for i in range(int(round(lda1) - 1)):
x2_prime = utils.mutation(x, l)
x2_prime.value = getattr(evaluate, instance.name)(x2_prime,
instance)
T += 1
if x2_prime.value > x_prime.value:
x_prime = x2_prime
#---------- crossover phase
y, Ty = LL_crossover_phase(x=x,
x_prime=x_prime,
instance=instance,
implementation_choice=crossover_choice,
chromossome_length=n,
n_offsprings=lda2,
crossover_bias=c)
T += Ty
#---------- selection and update phase
if (y.value <= x.value):
#x = y
lda = min([lda * a, n - 1])
if (y.value > x.value):
x = y
lda = max([lda * b, 1])
if f is not None:
ls = [
T, x.value, x.value, x.value, x.value, p, c, lda1, lda2,
counter
] # IOH output
s = ' '.join([str(val) for val in ls])
f.write('\n' + s)
ldas.append(lda)
function_value.append(x.value)
neval.append(T)
counter += 1
if max_evaluation > 0 and T >= max_evaluation:
break
#End while
if f is not None:
f.write('\n' + str(T))
f.close()
return (T, counter, function_value, neval)
def LL_static_02(initial_solution,
instance,
lda,
alpha,
beta,
crossover_choice,
max_evaluation,
ioh_output=None):
x = initial_solution
n = len(x.chromossome)
T = 0
function_value = []
neval = []
counter = 0
while (x.value != instance.optimum):
#---------- mutation phase ------------
#sample l from the binomial
p = alpha / float(n)
l = np.random.binomial(n, p)
while l == 0:
l = np.random.binomial(n, p)
x_prime = utils.mutation(x, l)
x_prime.value = getattr(evaluate, instance.name)(x_prime, instance)
T += 1
for i in range(lda - 1):
x2_prime = utils.mutation(x, l)
x2_prime.value = getattr(evaluate, instance.name)(x2_prime,
instance)
T += 1
if x2_prime.value > x_prime.value:
x_prime = x2_prime
#-------- crossover phase
c = 1.0 / beta
y, Ty = LL_crossover_phase(x=x,
x_prime=x_prime,
instance=instance,
implementation_choice=crossover_choice,
chromossome_length=n,
n_offsprings=lda,
crossover_bias=c)
T += Ty
#-------- selection
if y.value > x.value:
x = y
function_value.append(x.value)
neval.append(T)
#counter += 1
if max_evaluation > 0 and T >= max_evaluation:
break
#if T % 1000 == 0:
# print(T)
# print(str(x.value) + '/' + str(instance.optimum))
#End while
return (T, counter, function_value, neval)