迭代和Python中的多维数组选择特定阵列 [英] Iterating and selecting a specific array from a multidimensional array in Python

问题描述

想象我有这样的事情：

 import numpy as np
 arra = np.arange(16).reshape(2, 2, 4)  

这使得

 array([[[0, 1, 2, 3],  
      [4, 5, 6, 7]],  
      [[8, 9, 10, 11],  
      [12, 13, 14, 15]]])     

和我想要沿着特定的子阵列上运行（在这种情况下，我想沿着每个'矩阵'的每个列运行）的循环，总结的结果（即，选择0和4，总结他们（4），在选择1和5以及对它们求和（6），...，选择3和7和对它们求和（10），选择8和12以及对它们求和（20），...，选择11和15和他们总结（26））。

and I want to make a loop that runs along specific subarrays (in this case, I want to run along each 'column' of each 'matrix') and sum the result (that is, selecting 0 and 4 and summing them (4), selecting 1 and 5 and summing them (6), ..., selecting 3 and 7 and summing them (10), selecting 8 and 12 and summing them (20), ..., selecting 11 and 15 and summing them (26)).

我曾想过这样做的显然是天然的：

I had thought doing that with the apparently natural:

 for i in arra[i, j, k]:  
     for j in arra[i, j, k]:  
         for k in arra[i, j, k]:  
             sum...  

问题是Python的肯定不允许做什么，我想用这种方式。如果它是一个二维数组，它会更容易些，因为我知道迭代器首先通过行运行，这样你就可以转沿列运行，但对于多维（3D在这种情况下）阵列（N，M，P）与N，M，P >> 1，我不知道如何可以做到。

The problem is that Python certainly doesn't allow to do what I want in this way. If it were a 2D array it would be easier as I know that the iterator first runs through the rows, so you can transpose to run along the columns, but for a multidimensional (3D in this case) array (N, M, P) with N, M, P >> 1, I was wondering how it could be done.

编辑：这个问题有一个在这里延续：<一href=\"http://stackoverflow.com/questions/35775040/choosing-and-iterating-specific-sub-arrays-in-multidimensional-arrays-in-python\">Choosing并遍历在Python中多维数组的特定子阵列

This question has a continuation here: Choosing and iterating specific sub-arrays in multidimensional arrays in Python

推荐答案

您可以使用的 地图 来完成这件事：

You can use map to get this done:

import numpy as np
arra = np.arange(16).reshape(2, 2, 4)  

然后命令：

map(sum, arra)

为您提供所需的输出：

gives you the desired output:

[array([ 4,  6,  8, 10]), array([20, 22, 24, 26])]

另外，您还可以使用名单COM prehension ：

res = [sum(ai) for ai in arra]

然后 RES 是这样的：

[array([ 4,  6,  8, 10]), array([20, 22, 24, 26])]

编辑：

如果您要添加相同的行 - 因为你在这个答案下面的评论中提到的 - 你可以这样做（使用的 拉链 ）：

If you want to add identical rows - as you mentioned in the comments below this answer - you can do (using zip):

map(sum, zip(*arra))

它给你所需要的输出：

which gives you the desired output:

[array([ 8, 10, 12, 14]), array([16, 18, 20, 22])]

有关完整起见也列表玉米prehension：

For the sake of completeness also the list comprehension:

[sum(ai) for ai in zip(*arra)]

它给你同样的输出。

which gives you the same output.

这篇关于迭代和Python中的多维数组选择特定阵列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！