python 迭代 3D 数组的 Pythonic 方式
声明：本页面是StackOverFlow热门问题的中英对照翻译，遵循CC BY-SA 4.0协议，如果您需要使用它，必须同样遵循CC BY-SA许可，注明原文地址和作者信息，同时你必须将它归于原作者（不是我）：StackOverFlow
原文地址: http://stackoverflow.com/questions/1316068/
Warning: these are provided under cc-by-sa 4.0 license. You are free to use/share it, But you must attribute it to the original authors (not me):
StackOverFlow
Pythonic way of iterating over 3D array
提问by Nathan Fellman
I have a 3D array in Python and I need to iterate over all the cubes in the array. That is, for all (x,y,z)in the array's dimensions I need to access the cube:
我在 Python 中有一个 3D 数组，我需要遍历数组中的所有多维数据集。也就是说，对于(x,y,z)数组的所有维度，我都需要访问多维数据集：
array[(x + 0, y + 0, z + 0)]
array[(x + 1, y + 0, z + 0)]
array[(x + 0, y + 1, z + 0)]
array[(x + 1, y + 1, z + 0)]
array[(x + 0, y + 0, z + 1)]
array[(x + 1, y + 0, z + 1)]
array[(x + 0, y + 1, z + 1)]
array[(x + 1, y + 1, z + 1)]
The array is a Numpy array, though that's not really necessary. I just found it very easy to read the data in with a one-liner using numpy.fromfile().
该数组是一个 Numpy 数组，尽管这并不是必需的。我刚刚发现使用numpy.fromfile().
Is there any more Pythonic way to iterate over these than the following? That simply looks like C using Python syntax.
有没有比以下更多的 Pythonic 方法来迭代这些？这看起来就像使用 Python 语法的 C。
for x in range(x_dimension):
for y in range(y_dimension):
for z in range(z_dimension):
work_with_cube(array[(x + 0, y + 0, z + 0)],
array[(x + 1, y + 0, z + 0)],
array[(x + 0, y + 1, z + 0)],
array[(x + 1, y + 1, z + 0)],
array[(x + 0, y + 0, z + 1)],
array[(x + 1, y + 0, z + 1)],
array[(x + 0, y + 1, z + 1)],
array[(x + 1, y + 1, z + 1)])
回答by Otto Allmendinger
Have a look at itertools, especially itertools.product. You can compress the three loops into one with
看看itertools，尤其是itertools.product。您可以将三个循环压缩为一个
import itertools
for x, y, z in itertools.product(*map(xrange, (x_dim, y_dim, z_dim)):
...
You can also create the cube this way:
您还可以通过以下方式创建多维数据集：
cube = numpy.array(list(itertools.product((0,1), (0,1), (0,1))))
print cube
array([[0, 0, 0],
[0, 0, 1],
[0, 1, 0],
[0, 1, 1],
[1, 0, 0],
[1, 0, 1],
[1, 1, 0],
[1, 1, 1]])
and add the offsets by a simple addition
并通过简单的加法添加偏移量
print cube + (10,100,1000)
array([[ 10, 100, 1000],
[ 10, 100, 1001],
[ 10, 101, 1000],
[ 10, 101, 1001],
[ 11, 100, 1000],
[ 11, 100, 1001],
[ 11, 101, 1000],
[ 11, 101, 1001]])
which would to translate to cube + (x,y,z)in your case. The very compact version of your code would be
cube + (x,y,z)在你的情况下会转化为。您的代码的非常紧凑的版本将是
import itertools, numpy
cube = numpy.array(list(itertools.product((0,1), (0,1), (0,1))))
x_dim = y_dim = z_dim = 10
for offset in itertools.product(*map(xrange, (x_dim, y_dim, z_dim))):
work_with_cube(cube+offset)
Edit: itertools.productmakes the product over the different arguments, i.e. itertools.product(a,b,c), so I have to pass map(xrange, ...)with as *map(...)
编辑：itertools.product使产品通过不同的参数，即itertools.product(a,b,c)，所以我必须通过map(xrange, ...)as*map(...)
回答by nosklo
import itertools
for x, y, z in itertools.product(xrange(x_size),
xrange(y_size),
xrange(z_size)):
work_with_cube(array[x, y, z])
