替换numpy数组中的行时出错 [英] Error when replacing row in numpy array

问题描述

我遇到了一个问题，尝试用新行替换numpy二维数组中的行.

I am hitting a problem trying to replace a row in a numpy 2-d array with a new row.

我正试图编写一个函数，以 m 矩阵(n个长度为m的长度向量的样本)取 n ，并返回一个正交基数集.对于第一步，我要计算第一行的长度，然后除以该值(标准化第一行).当我尝试将归一化的行分配回原始矩阵时，出现错误:

I am trying to write a function to take an n by m matrix (n samples of m-length vectors) and return an orthonormal basis set. For the first step I am calculating the length of the first row and then dividing by that value (normalizing the first row). When I try to assign the normalized row back to the original matrix, I am hitting an error:

import numpy.linalg as la

def gauss_jordan(z):
    print(z[0])
    print(la.norm(z[0]))
    print(z[0] / la.norm(z[0])))
    print((z[0] / la.norm(z[0])).shape)
    z[0, :] = z[0, :] / la.norm(z[0])

    print(z)

结果:

[ 1  2 -2]
3.0
[ 0.33333333  0.66666667 -0.66666667]
(3,)
[[ 0  0  0]
 [-1  3  1]
 [-2  1  3]
 [ 1 -2  5]]

零行来自哪里?计算出的值都是正确的，但是我无法弄清楚我的作业出了什么问题.我也尝试过 z [0 ,:] = z [0]/la.norm(z [0]).

Where is the zero row coming from? The calculated values are all correct, but I can't figure out what is wrong with my assignment. I have tried z[0,:] = z[0] / la.norm(z[0]) as well.

推荐答案

在我看来，它采用的是舍入除法而不是普通的除法.如果您尝试

It looks to me like it's taking the rounded division instead of the normal one. What happens if you try to

z = z.astype('float64')
z[0] /= la.norm(z[0])

此外，您使用的是numpy和python的哪个版本?难道是python 2.X吗?

Also, what version numpy and python are you using? Is it by any chance python 2.X ?

这篇关于替换numpy数组中的行时出错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！