在二维匿名函数数组变量上使用 scipy.optimize.fsolve 时，I/O 形状不匹配 [英] I/O shape mismatch when using scipy.optimize.fsolve on 2-dimensional anonymous function array variable

问题描述

这是源代码:

def lambdatest():F=lambda y: y-np.array([[1,2],[3,4]])y0=np.array([[3,4],[8,7]])Y=scipy.optimize.fsolve(F,y0)返回 Y

我得到的错误是:

 引发 TypeError(msg)类型错误:fsolve:func"参数<lambda>"的输入和输出形状不匹配.

我环顾四周，但似乎无法理解.

解决方案

F(fsolve) 必须返回标量或一维数组.fsolve 不处理高维数组.

您可以做的是使用 ravel() 方法，然后将 fsolve 返回的解重新整形为二维数组:

def lambdatest():F = lambda y: y - np.array([[1,2],[3,4]]).ravel()y0 = np.array([[3,4],[8,7]])Y = scipy.optimize.fsolve(F, y0.ravel()).reshape(y0.shape)返回 Y

结果如下:

<预><代码>>>>拉姆达测试()数组([[ 1., 2.],[ 3., 4.]])

Here is the source code:

def lambdatest():
    F=lambda y: y-np.array([[1,2],[3,4]])
    y0=np.array([[3,4],[8,7]])    
    Y=scipy.optimize.fsolve(F,y0)
    return Y

And the error I get is:

    raise TypeError(msg)
TypeError: fsolve: there is a mismatch between the input and output shape of the 'func' argument '<lambda>'.

I have looked around but can't seem to make sense of it.

解决方案

F (the func argument of fsolve) must return either a scalar or a one-dimensional array. fsolve doesn't handle higher dimensional arrays.

What you can do is flatten the 2-d array to a 1-d array using the ravel() method, and then reshape the solution returned by fsolve into a 2-d array:

def lambdatest():
    F = lambda y: y - np.array([[1,2],[3,4]]).ravel()
    y0 = np.array([[3,4],[8,7]])            
    Y = scipy.optimize.fsolve(F, y0.ravel()).reshape(y0.shape)
    return Y

Here's the result:

>>> lambdatest()
array([[ 1.,  2.],
       [ 3.,  4.]])

这篇关于在二维匿名函数数组变量上使用 scipy.optimize.fsolve 时，I/O 形状不匹配的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！