numpy中a[][][] 和 a[, , ]两种索引方式的区别

之前一直对a[][][] 和 a[, , ]两种索引方式的区别不清楚，今天试验了一下。
根据三个维度分别是：还是数字，共2^3^ =8种情况进行了实验。

1、a[:][:][:] 与 a[:, :, :]（等价于a[:]）， a[1][0][2] 与 a[1, 0, 2]

# -*- coding: utf-8 -*-
import numpy as np

a = np.arange(24).reshape([2,3,4])
print('a = ')
print(a)

b1 = a[:][:][:]
c1 = a[:, :, :]
print('b1 = ')
print(b1)
print('c1 = ')
print(c1)

b2 = a[1][0][2]
c2 = a[1, 0, 2]
print('b2 = ')
print(b2)
print('c2 = ')
print(c2)

结果

a = 
[[[ 0  1  2  3]
  [ 4  5  6  7]
  [ 8  9 10 11]]

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


b1 = 
[[[ 0  1  2  3]
  [ 4  5  6  7]
  [ 8  9 10 11]]

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


c1 = 
[[[ 0  1  2  3]
  [ 4  5  6  7]
  [ 8  9 10 11]]

 [[12 13 14 15]
  [16 17 18 19]
  [20 21 22 23]]]
  
b2 = 
14

c2 = 
14

可以看到两种方式没有区别

2、形如 a[0][:][:] 与 a[0, :, :]

# -*- coding: utf-8 -*-
import numpy as np

a = np.arange(24).reshape([2,3,4])
print('a = ')
print(a)

b1 = a[0][:][:]
c1 = a[0, :, :]
print('b1 = ')
print(b1)
print('c1 = ')
print(c1)

b2 = a[:][0][:]
c2 = a[:, 0, :]
print('b2 = ')
print(b2)
print('c2 = ')
print(c2)

b3 = a[:][:][0]
c3 = a[:, :, 0]
print('b3 = ')
print(b3)
print('c3 = ')
print(c3)

结果

a = 
[[[ 0  1  2  3]
  [ 4  5  6  7]
  [ 8  9 10 11]]

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


b1 = 
[[ 0  1  2  3]
 [ 4  5  6  7]
 [ 8  9 10 11]]
 
c1 = 
[[ 0  1  2  3]
 [ 4  5  6  7]
 [ 8  9 10 11]]
 
b2 = 
[[ 0  1  2  3]
 [ 4  5  6  7]
 [ 8  9 10 11]]
 
c2 = 
[[ 0  1  2  3]
 [12 13 14 15]]
 
b3 = 
[[ 0  1  2  3]
 [ 4  5  6  7]
 [ 8  9 10 11]]
 
c3 = 
[[ 0  4  8]
 [12 16 20]]

可以看到a[0][:][:] ，a[:][0][:] ，a[:][:][0]效果一样（都是a[0][:][:]的结果），而另一种方式是不同的

3、形如 a[0][0][:] 与 a[0, 0, :]

# -*- coding: utf-8 -*-
import numpy as np

a = np.arange(24).reshape([2,3,4])
print('a = ')
print(a)

b1 = a[0][0][:]
c1 = a[0, 0, :]
print('b1 = ')
print(b1)
print('c1 = ')
print(c1)

b2 = a[0][:][0]
c2 = a[0, :, 0]
print('b2 = ')
print(b2)
print('c2 = ')
print(c2)

b3 = a[:][0][0]
c3 = a[:, 0, 0]
print('b3 = ')
print(b3)
print('c3 = ')
print(c3)

结果

a = 
[[[ 0  1  2  3]
  [ 4  5  6  7]
  [ 8  9 10 11]]

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


b1 = 
[0 1 2 3]

c1 = 
[0 1 2 3]

b2 = 
[0 1 2 3]

c2 = 
[0 4 8]

b3 = 
[0 1 2 3]

c3 = 
[ 0 12]

同第二条，a[0][0][:] ，a[:][0][0] ，a[0][:][0]效果一样（都是a[0][0][:]的结果），而另一种方式是不同的

个人理解：a[][][]这种方式索引时会自动把:调整到最后，而a[, ,]这种方式会正确按照顺序去索引