我正在尝试使用 SciPy 中的 Differential_evolution 。我有三个矩阵:x、y 和 P - 大小均为 (14,6)。我必须使用以下公式:
z= np.log10(g)+ np.log10(c)*np.log10(P)
找到 c 的值(0 到 2 之间的实数)，从而最小化:
numpy.median(z**2)
这个表达式。我尝试的是这样的(为了方便起见，我提供了数据):
import numpy as np
from scipy.optimize import differential_evolution
def func(c, args):
z = args[0] + np.log10(c)*np.log10(args[1])
return np.median(z**2)
if __name__ == '__main__':
bounds = [(0, 2)]
x = np.array([[126581.94951205, 97601.85624482, 59659.00330833,
27646.48551627, 9202.50377458, 4840.25789068],
[213571.84886437, 148750.52154776, 85979.81139937,
38757.37831212, 11775.99906427, 4619.32027948],
[195684.50299021, 131818.78542437, 74376.55189913,
32793.21715377, 10288.70838873, 4042.58093119],
[177598.13865746, 120942.50439911, 68866.09898276,
30819.5354775 , 10588.08746517, 5011.71808947],
[126433.18311483, 85863.57788065, 48923.64502157,
21828.60950911, 7907.37639781, 4410.61819399],
[103431.88029629, 67452.94418262, 37608.36861047,
16456.97701443, 6027.98704858, 3550.06927169],
[100689.06813945, 64380.21348052, 34764.02910376,
14849.85472635, 5607.19256065, 3605.5709208 ],
[ 96509.22946744, 63832.74512518, 36041.69174706,
15802.87650901, 6473.33232805, 4664.07058733],
[113078.63455882, 73227.02362359, 40861.09037499,
17385.89127848, 7074.98444924, 5136.84232454],
[121241.93118924, 78537.13681709, 44257.97654994,
18584.94999742, 7733.39219718, 5869.49536788],
[115948.06368262, 73995.07204278, 41536.21315507,
16851.59724901, 6736.25125909, 4851.5738275 ],
[115024.20359423, 72108.15245783, 40341.98473413,
15900.55422399, 6243.63777265, 4411.24859372],
[108754.83802899, 66210.25952459, 36485.42905112,
14577.73925124, 5553.23702141, 3736.5217322 ],
[ 95340.59125024, 58458.97552915, 32364.19705748,
13236.30114676, 4929.04023171, 3202.21731277]])
y = y=np.array([[118166.08 , 95784.692 , 68134.878 , 37119.959 , 17924.157 ,
7445.3083],
[ 99265.027 , 70679.135 , 43297.559 , 19822.017 , 8527.8497,
3404.7113],
[ 80227.797 , 50972.879 , 26648.604 , 11190.488 , 4836.6514,
2249.9172],
[ 68510.582 , 39288.19 , 19938.938 , 9312.6881, 4907.6661,
2681.2709],
[ 65193.15 , 36610.107 , 18612.181 , 9211.144 , 5416.1685,
3372.1282],
[ 67188.918 , 37227.699 , 20132.92 , 11663.275 , 7315.3472,
4648.1669],
[ 64802.06 , 38885.622 , 22008.537 , 13100.638 , 8043.0185,
5049.2097],
[ 68104.867 , 41212.89 , 23247.898 , 14134.707 , 8805.2547,
5526.1014],
[ 74180.595 , 41268.904 , 22868.016 , 13841.437 , 8660.1413,
5401.245 ],
[ 78920.685 , 42743.389 , 23932.305 , 13910.089 , 8439.3342,
5141.7051],
[ 91329.012 , 45733.772 , 25430.818 , 14144.185 , 8273.7953,
5016.5839],
[ 92217.594 , 44984.3 , 23353.596 , 13467.631 , 8099.728 ,
4948.26 ],
[ 94508.441 , 48114.879 , 24735.311 , 13358.097 , 7821.8587,
4806.7923],
[108211.73 , 53987.095 , 25872.772 , 13189.61 , 7552.7164,
4497.2611]])
P=10000*np.array([[0.6011,0.6011,0.6011,0.6011,0.6011,0.6011],
[0.9007,0.9007,0.9007,0.9007,0.9007,0.9007],
[1.1968,1.1968,1.1968,1.1968,1.1968,1.1968],
[1.4178,1.4178,1.4178,1.4178,1.4178,1.4178],
[1.5015,1.5015,1.5015,1.5015,1.5015,1.5015],
[1.439,1.439,1.439,1.439,1.439,1.439],
[1.2721,1.2721,1.2721,1.2721,1.2721,1.2721],
[1.0616,1.0616,1.0616,1.0616,1.0616,1.0616],
[0.8543,0.8543,0.8543,0.8543,0.8543,0.8543],
[0.6723,0.6723,0.6723,0.6723,0.6723,0.6723],
[0.5204,0.5204,0.5204,0.5204,0.5204,0.5204],
[0.3963,0.3963,0.3963,0.3963,0.3963,0.3963],
[0.2990,0.2990,0.2990,0.2990,0.2990,0.2990],
[0.2211,0.2211,0.2211,0.2211,0.2211,0.2211]])
g=np.log10(y) - np.log10(x)
args = (g,P)
result = differential_evolution(func, bounds, args=args)
print(func(bounds, args))
我收到此错误: TypeError: func() 恰好需要 2 个参数(给定 3 个参数) 。有什么办法可以解决这个问题吗？
最佳答案
def func(c, g, P):
z = g + np.log10(c)*np.log10(P)
return np.median(z**2)
if __name__ == '__main__':
# Your arrays go here
g = np.log10(y) - np.log10(x)
args = (g, P)
result = differential_evolution(func, bounds, args=(g, P))
# will print the value of c and value of the optimized function
print (result.x, result.fun)
关于python - SciPy 中的差分进化，我们在Stack Overflow上找到一个类似的问题： https://stackoverflow.com/questions/49308372/