多维数组的切片(slicing)

# coding: utf-8
import numpy as np

# arr = np.arange(24).reshape((2, 3, 4))
# print(arr)
#  arr=[[[ 0  1  2]
#   [ 3  4  5]]
#
#  [[ 6  7  8]
#   [ 9 10 11]]]
# print(arr[1, ...])               # 等价于 arr[1, :, :]
#  arr[1, ...]=  [[ 6  7  8]
#  [ 9 10 11]]
# print(arr[..., 1])               # 等价于 arr[:, :, 1]
#  arr[..., 1]=[[ 1  4]
#  [ 7 10]]
arr = np.arange(12).reshape((2, 2, 3))
print(arr)
#  arr=[[[ 0  1  2]
#   [ 3  4  5]]
#
#  [[ 6  7  8]
#   [ 9 10 11]]]
print(arr[1, ...])               # 等价于 arr[1, :, :]
# arr[1, ...]=[[ 6  7  8]
#  [ 9 10 11]]
print(arr[..., 1])               # 等价于 arr[:, :, 1]
# arr[..., 1]=[[ 1  4]
#  [ 7 10]]

切片(slicing)操作

Numpy 中多维数组的切片操作与 Python 中 list 的切片操作一样，同样由 start, stop, step 三个部分组成

import numpy as np

arr = np.arange(12)
print 'array is:', arr

slice_one = arr[:4]
print 'slice begins at 0 and ends at 4 is:', slice_one

slice_two = arr[7:10]
print 'slice begins at 7 and ends at 10 is:', slice_two

slice_three = arr[0:12:4]
print 'slice begins at 0 and ends at 12 with step 4 is:', slice_three

array is: [ 0  1  2  3  4  5  6  7  8  9 10 11]
slice begins at 0 and ends at 4 is: [0 1 2 3]
slice begins at 7 and ends at 10 is: [7 8 9]
slice begins at 0 and ends at 12 with step 4 is: [0 4 8]

上述例子是一维数组的例子，如果是多维数组，将不同维度上的切片操作用 逗号 分开就好了

# coding: utf-8
import numpy as np

arr = np.arange(12).reshape((3, 4))
print 'array is:'
print arr

# 取第一维的索引 1 到索引 2 之间的元素，也就是第二行
# 取第二维的索引 1 到索引 3 之间的元素，也就是第二列和第三列
slice_one = arr[1:2, 1:3]
print 'first slice is:'
print slice_one

# 取第一维的全部
# 按步长为 2 取第二维的索引 0 到末尾 之间的元素，也就是第一列和第三列
slice_two = arr[:, ::2]
print 'second slice is:'
print slice_two

array is:
[[ 0  1  2  3]
 [ 4  5  6  7]
 [ 8  9 10 11]]
first slice is:
[[5 6]]
second slice is:
[[ 0  2]
 [ 4  6]
 [ 8 10]]

对于 slice_two，如果 arr 是用 Python 的 list 表示的，那么要得到相同的结果得像下面这样，相对来说就麻烦多了:

import numpy as np

arr = np.arange(12).reshape((3, 4)).tolist()

slice_two = [
    row[::2] for row in arr
]
print slice_two

[[0, 2], [4, 6], [8, 10]]

对于维数超过 3 的多维数组，还可以通过 '…' 来简化操作

# coding: utf-8
import numpy as np

arr = np.arange(24).reshape((2, 3, 4))

print arr[1, ...]               # 等价于 arr[1, :, :]
print arr[..., 1]               # 等价于 arr[:, :, 1]

[[12 13 14 15]
 [16 17 18 19]
 [20 21 22 23]]
[[ 1  5  9]
 [13 17 21]]