def test_es():
#Define the fitness function
def FIT(individual):
"""Sphere test objective function.
F(x) = sum_{i=1}^d xi^2
d=1,2,3,...
Range: [-100,100]
Minima: 0
"""
y = sum(x**2 for x in individual)
return y
#Setup the parameter space (d=5)
nx = 5
BOUNDS = {}
for i in range(1, nx + 1):
BOUNDS['x' + str(i)] = ['float', -100, 100]
es = ES(mode='min',
bounds=BOUNDS,
fit=FIT,
lambda_=80,
mu=40,
mutpb=0.25,
cxmode='blend',
cxpb=0.7,
ncores=1,
seed=1)
x_best, y_best, es_hist = es.evolute(ngen=100, verbose=0)
def tune_fit(cxpb, mu, alpha, cxmode):
#--setup the parameter space
nx = 5
BOUNDS = {}
for i in range(1, nx + 1):
BOUNDS['x' + str(i)] = ['float', -100, 100]
#--setup the ES algorithm
es = ES(mode='min',
bounds=BOUNDS,
fit=sphere,
lambda_=80,
mu=mu,
mutpb=0.1,
alpha=alpha,
cxmode=cxmode,
cxpb=cxpb,
ncores=1,
seed=1)
#--Evolute the ES object and obtains y_best
#--turn off verbose for less algorithm print-out when tuning
x_best, y_best, es_hist = es.evolute(ngen=100, verbose=0)
return y_best #returns the best score
def tune_fit(cxpb, mu, alpha, cxmode, mutpb):
#--setup the ES algorithm
es = ES(mode='min',
bounds=BOUNDS,
fit=BEAM,
lambda_=80,
mu=mu,
mutpb=mutpb,
alpha=alpha,
cxmode=cxmode,
cxpb=cxpb,
ncores=1,
seed=1)
#--Evolute the ES object and obtains y_best
#--turn off verbose for less algorithm print-out when tuning
x_best, y_best, es_hist = es.evolute(ngen=100, verbose=0)
return y_best #returns the best score
#Define the fitness function
def FIT(individual):
"""Sphere test objective function.
F(x) = sum_{i=1}^d xi^2
d=1,2,3,...
Range: [-100,100]
Minima: 0
"""
y = sum(x**2 for x in individual)
return y
#Setup the parameter space (d=5)
nx = 5
BOUNDS = {}
for i in range(1, nx + 1):
BOUNDS['x' + str(i)] = ['float', -100, 100]
es = ES(mode='min',
bounds=BOUNDS,
fit=FIT,
lambda_=80,
mu=40,
mutpb=0.25,
cxmode='blend',
cxpb=0.7,
ncores=1,
seed=1)
x_best, y_best, es_hist = es.evolute(ngen=100, verbose=1)
bounds['x2'] = ['grid', (0.0625, 0.125, 0.1875, 0.25, 0.3125, 0.375, 0.4375, 0.5, 0.5625, 0.625)]
bounds['x3'] = ['float', 10, 200]
bounds['x4'] = ['float', 10, 200]
########################
# Setup and evolute HHO
########################
hho = HHO(mode='min', bounds=bounds, fit=Vessel, nhawks=50,
int_transform='minmax', ncores=1, seed=1)
x_hho, y_hho, hho_hist=hho.evolute(ngen=200, verbose=False)
assert Vessel(x_hho) == y_hho
########################
# Setup and evolute ES
########################
es = ES(mode='min', fit=Vessel, cxmode='cx2point', bounds=bounds,
lambda_=60, mu=30, cxpb=0.7, mutpb=0.2, seed=1)
x_es, y_es, es_hist=es.evolute(ngen=200, verbose=False)
assert Vessel(x_es) == y_es
########################
# Setup and evolute PESA
########################
pesa=PESA(mode='min', bounds=bounds, fit=Vessel, npop=60, mu=30, alpha_init=0.01,
alpha_end=1.0, cxpb=0.7, mutpb=0.2, alpha_backdoor=0.05)
x_pesa, y_pesa, pesa_hist=pesa.evolute(ngen=200, verbose=False)
assert Vessel(x_pesa) == y_pesa
########################
# Setup and evolute BAT
########################
bat=BAT(mode='min', bounds=bounds, fit=Vessel, nbats=50, fmin = 0 , fmax = 1,
1] #drop the cases with scores < 1 (violates the constraints)
bayesres = bayesres.sort_values(
['score'], axis='index',
ascending=True) #rank the scores from best (lowest) to worst (high)
print(bayesres.iloc[0:10, :]
) #the results are saved in dataframe and ranked from best to worst
#*************************************************************
# Part V: Rerun ES with the best hyperparameter set
#*************************************************************
es = ES(mode='min',
bounds=BOUNDS,
fit=BEAM,
lambda_=80,
mu=bayesres['mu'].iloc[0],
mutpb=bayesres['mutpb'].iloc[0],
alpha=bayesres['alpha'].iloc[0],
cxmode=bayesres['cxmode'].iloc[0],
cxpb=bayesres['cxpb'].iloc[0],
ncores=1,
seed=1)
x_best, y_best, es_hist = es.evolute(ngen=100, verbose=0)
print('Best fitness (y) found:', y_best)
print('Best individual (x) found:', x_best)
#---------------------------------
# Plot
#---------------------------------
#Plot fitness convergence
# Parameter Space
#---------------------------------
d = 20
space = {}
for i in range(1, d + 1):
space['x' + str(i)] = ['float', -32, 32]
#---------------------------------
# GA
#---------------------------------
ga = ES(mode='min',
bounds=space,
fit=ACKLEY,
lambda_=50,
mu=25,
mutpb=0.15,
alpha=0.5,
cxmode='blend',
cxpb=0.85,
ncores=1,
seed=1)
x_ga, y_ga, ga_hist = ga.evolute(ngen=150, verbose=0)
#---------------------------------
# GWO
#---------------------------------
gwo = GWO(mode='min', fit=ACKLEY, bounds=space, nwolves=50, ncores=1, seed=1)
x_gwo, y_gwo, gwo_hist = gwo.evolute(ngen=150, verbose=0)
#---------------------------------
# WOA
npop=sorted_res['npop'].iloc[0],
mu=int(sorted_res['frac'].iloc[0]*sorted_res['npop'].iloc[0]),
alpha_init=sorted_res['alpha_init'].iloc[0],
alpha_end=1.0,
cxpb=sorted_res['cxpb'].iloc[0],
mutpb=sorted_res['mutpb'].iloc[0],
c1=2.05, c2=2.05,
alpha_backdoor=0.1,
seed=1)
x_pesa, y_pesa, pesa_hist=pesa.evolute(ngen=300, x0=x0, verbose=False)
#Run GA
#use same optimized hyperparameters from PESA (or repeat tuning again for GA)
ga=ES(mode='min', bounds=bounds, fit=Vessel, cxmode='cx2point',
lambda_=sorted_res['npop'].iloc[0],
mu=int(sorted_res['frac'].iloc[0]*sorted_res['npop'].iloc[0]),
mutpb=sorted_res['mutpb'].iloc[0],
cxpb=sorted_res['cxpb'].iloc[0],
ncores=1, seed=1)
#filter initial guess for GA
#lambda_=int(sorted_res['npop'].iloc[0])
#x0_ga=x0[:lambda_]
x_ga, y_ga, ga_hist=ga.evolute(ngen=300, x0=None, verbose=0) #or use x0_ga if you like
#(random guess seems to be better)
#*************************************************************
# Part VI: Post-processing
#*************************************************************
#plot results
plt.figure()
plt.plot(pesa_hist, label='PESA')
