python實現模擬退火演算法
摘要：
問題
求F(x)在定義域[5,8]上的最大值
原始碼
# 模擬退火法
import numpy as np
import math
# 定義域x從5到8閉區間
BOUND = [5,8]
tmp = 1e2
tmp_min = 1e-3
alpha = 0.98
beta ...
問題
求F(x)在定義域[5,8]上的最大值
原始碼
# 模擬退火法
import numpy as np
import math
# 定義域x從5到8閉區間
BOUND = [5,8]
tmp = 1e2
tmp_min = 1e-3
alpha = 0.98
beta = 1
def F(x):
return math.sin(x*x)+2.0*math.cos(2.0*x)
def judge(de,tmp):
if de > 0:
return 1
else:
if math.exp(de/tmp) > np.random.rand():
return 1
else:
return 0
x = np.random.rand()*(BOUND[1]-BOUND[0])+BOUND[0]
f = F(x)
counter = 0
while tmp > tmp_min:
delta = (np.random.rand()-0.5)*beta
x_new = x + delta
if x_new < BOUND[0]:
x_new = x_new + BOUND[1] - BOUND[0]
if x_new > BOUND[1]:
x_new = x_new - BOUND[1] + BOUND[0]
f_new = F(x_new)
de = f_new - f
flag = judge(de,tmp)
if(flag):
f = f_new
x = x_new
if de > 0:
tmp = tmp * alpha
counter += 1
print('current x {}, y {},tmp {},counter {}'.format(x,f,tmp,counter))
執行結果
調調引數看怎樣收斂比較快
# 更優的引數配搭 tmp = 1e2 tmp_min = 1e-3 alpha = 0.89 beta = 1.2
今天也要開心呀