我有两个错误：'RuntimeWarning：除以double_scalars中的零'; 'RuntimeWarning：在减法中遇到无效的值'

Question

我不明白python如何被零除。这是我的整个代码：

from scipy import optimize
import numpy as np
from sympy import *

from tkinter import *
#import ipdb;ipdb.set_trace()
y=[100,200]
alp=[0.5,0.5]
end=[1,2]
e, t, a, w = symbols('e t a w')
u= ((a ** a) * (24 * (1 - t) * w + e) * ((1 - a) ** (1 - a))) * (((1 - t) * w) ** (a-1))

print(latex(u))
def main(x, n):

    r = 0
    p = 0
    #j=-1 * ((alp[i] ** alp[i]) * (24 * (1 - x[i]) * y[i] + end[i]) * ((1 - alp[i]) ** (1 - alp[i]))) / ((1 - x[1] * y[i]) ** (1 - alp[i]))
    for i in range(len(y)):
        if i >= n:

            r += -1 * ((alp[i] ** alp[i]) * (24 * (1 - x[1]) * y[i] + end[i]) * ((1 - alp[i]) ** (1 - alp[i]))) * (((1 - x[1]) * y[i]) ** (alp[i]-1))


        elif i < n:

            p += -1 * ((alp[i] ** alp[i]) * (24 * y[i] + x[0] + end[i]) * ((1 - alp[i]) ** (1 - alp[i]))) * ((y[i]) ** (alp[i]-1))
    return r + p

def tax_rev(n):
    r = 0
    for i in range(n, len(y)):
        r += y[i]

    return r

bounds = [(0, np.inf), (0, 1)]

mat = np.zeros((len(y), 4))


for i in range(len(y)):
    # print(i)
    def constraint1(x):
        return 1 * (x[1] * tax_rev(i) - x[0] * (len(y) -i))

    cons1 = {'type': 'ineq', 'fun': constraint1}

    def co(x):
        return main(x, i)

    max = optimize.shgo(co, bounds=bounds, constraints=cons1)

    mat[i, 0] = round(max.x[0] ,5)
    mat[i, 1] = round(max.x[1] ,5)
    mat[i, 2] = -1 * round(max.fun ,5)
    mat[i, 3] = max.success

tax_from= np.unravel_index(np.argmax(mat, axis=None), mat.shape)

fin_tex = f'{100 * round(float(mat[tax_from[0] ,1]) ,15)}  % from those endowed with {y[tax_from[0]]}  or more. Transfer {round(float(mat[tax_from[0] ,0]) ,15)} units of money to those endowed with less than  {y[tax_from[0]]} .This would maximize utility in the society.'
print(fin_tex)
print(mat)

我正在使用此代码优化可以在乳胶中看到的方程式（变量u中的方程式）。我不明白为什么当我不将任何东西都除以0时为什么会有零除。我认为这也在优化时引起问题。

除以零，根据我收到的错误消息在这里：

def main(x, n):

    r = 0
    p = 0
    #j=-1 * ((alp[i] ** alp[i]) * (24 * (1 - x[i]) * y[i] + end[i]) * ((1 - alp[i]) ** (1 - alp[i]))) / ((1 - x[1] * y[i]) ** (1 - alp[i]))
    for i in range(len(y)):
        if i >= n:

            r += -1 * ((alp[i] ** alp[i]) * (24 * (1 - x[1]) * y[i] + end[i]) * ((1 - alp[i]) ** (1 - alp[i]))) * (((1 - x[1]) * y[i]) ** (alp[i]-1))


        elif i < n:

            p += -1 * ((alp[i] ** alp[i]) * (24 * y[i] + x[0] + end[i]) * ((1 - alp[i]) ** (1 - alp[i]))) * ((y[i]) ** (alp[i]-1))
    return r + p

在我命名为r的变量中发生。

我是Python新手，欢迎提供有关代码的任何帮助/建议：）

Answer 1

完整警告是：

RuntimeWarning: divide by zero encountered in double_scalars
  if __name__ == '__main__':
/usr/local/lib/python3.6/dist-packages/scipy/optimize/slsqp.py:63: RuntimeWarning: invalid value encountered in subtract
  jac[i] = (func(*((x0+dx,)+args)) - f0)/epsilon
/usr/local/bin/ipython3:9: RuntimeWarning: invalid value encountered in double_scalars

因此，优化程序（slsqp）正在评估jacobian，然后除以epsilon。并进行减法。

这个问题有很多不必要的包.。例如说sympy/latex业务。您不要在优化中使用它。

基于警告上下文，您需要编写代码来测试优化，而没有该i循环。