def mutate(num_objectives, X_pool, X_hi, X_lo, mui, pop_size, seed):
# The method "mutate" requires the following parameters:
# X_pool : The population from the previous generation
# X_hi : Max. Values of Xi
# X_lo : Min. Values of Xi
# pop_size : Population Size
# seed_gen : For random number gen
random.seed(seed)
num_params = len(X_hi)
X_mut = []
for i in range(0, pop_size, 1):
xm = []
rn = random.random()
rnmut = random.uniform(0, 1)
if rnmut < 0.1:
for j in range(0, num_params, 1):
val = X_pool[i][j] + (X_hi[j] - X_lo[j]) * delta_calculator(
rn, mui)
if val > X_hi[j]:
val = X_hi[j]
elif val < X_lo[j]:
val = X_lo[j]
xm.append(val)
x_ev = xm
for k in range(0, num_objectives, 1):
xm.append(evaluate(k, x_ev))
X_mut.append(xm)
else:
x_nil = X_pool[i]
for k in range(0, num_objectives, 1):
x_nil.append(0)
x_nil[num_params + k] = evaluate(k, x_nil)
X_mut.append(x_nil)
return X_mut
def initialize(X_hi, X_lo, pop_size, num_objectives):
X_init = [[0 for x in range(len(X_hi) + num_objectives)]
for y in range(pop_size)]
for i in range(0, pop_size, 1):
for j in range(0, len(X_hi), 1):
rn = random.uniform(0, 1)
X_init[i][j] = round(X_lo[j] + rn * (X_hi[j] - X_lo[j]), 4)
for k in range(0, num_objectives, 1):
X_init[i][len(X_hi) + k] = evaluate(k, X_init[i][:len(X_hi)])
return X_init
def selection(num_obj, X_trial, X_target, pop_size, num_params):
X_sel = [[0 for x in range(num_params)] for y in range(pop_size)]
for i in range(0, pop_size, 1):
trial_wins = 0
trial_F_val = []
target_F_val = []
for j in range(0, num_obj, 1):
trial_F_val.append(evaluate(j, X_trial[i]))
target_F_val.append(X_target[i][num_params + j])
if trial_F_val[j] < target_F_val[j]:
trial_wins = 1
if trial_wins == 1:
X_sel[i] = X_trial[i][:num_params] + trial_F_val
else:
X_sel[i] = X_target[i][:num_params] + target_F_val
return X_sel
def genetic_operator(num_objectives, X_pool, X_hi, X_lo, cxi, mui, pop_size,
num_params, seed):
mutated = 0
crossed = 0
p = 1
random.seed(seed)
X_daughters = []
for i in range(0, pop_size, 1):
# Performing Crossover with 90% probability
cxp = random.uniform(0, 1)
rn = random.sample(range(0, pop_size - 1),
3) # List of 3 random numbers
if cxp < 0.9:
xd1 = []
xd2 = []
for j in range(0, num_params, 1):
rn_beta = random.random()
beta = beta_calculator(rn_beta, cxi)
xd1_val = (0.5 * ((1 - beta) * X_pool[rn[0]][j] +
(1 + beta) * X_pool[rn[1]][j]))
if xd1_val > X_hi[j]:
xd1_val = X_hi[j]
elif xd1_val < X_lo[j]:
xd1_val = X_lo[j]
xd2_val = (0.5 * ((1 + beta) * X_pool[rn[0]][j] +
(1 - beta) * X_pool[rn[1]][j]))
if xd2_val > X_hi[j]:
xd2_val = X_hi[j]
elif xd2_val < X_lo[j]:
xd2_val = X_lo[j]
xd1.append(xd1_val)
xd2.append(xd2_val)
mup = random.uniform(0, 1)
if mup < 0.1:
xd1[j] = X_lo[j] + random.uniform(0,
1) * (X_hi[j] - X_lo[j])
xd2[j] = X_lo[j] + random.uniform(0,
1) * (X_hi[j] - X_lo[j])
xev1 = xd1
xev2 = xd2
for k in range(0, num_objectives, 1):
xd1.append(evaluate(k, xev1))
xd2.append(evaluate(k, xev2))
crossed = 1
mutated = 0
else:
XM = []
XP = X_pool[rn[2]]
for j in range(0, num_params, 1):
rn = random.random()
val = XP[j] + delta_calculator(rn, mui)
if val > X_hi[j]:
val = X_hi[j]
elif val < X_lo[j]:
val = X_lo[j]
XM.append(val)
x_ev = XM
for k in range(0, num_objectives, 1):
XM.append(evaluate(k, x_ev))
mutated = 1
crossed = 0
if crossed:
X_daughters.append(xd1)
X_daughters.append(xd2)
crossed = 0
elif mutated:
X_daughters.append(XM)
mutated = 0
return X_daughters