搞不懂 numpy.sum 请过来

numpy.sum  官方给的例子里，是一会按列求和，一会按行求和，总有种不确定感（轴axis好术语化呀，没线性代数基础完全不懂呀）那两维以上的怎么办？

先看方法说明

numpy. sum ( a, axis=None, dtype=None, out=None ) ¶

Sum of array elements over a given axis.

Parameters :

a : array_like Elements to sum. axis : integer, optional Axis over which the sum is taken. By default axis is None,and all elements are summed. dtype : dtype, optional The type of the returned array and of the accumulator in whichthe elements are summed. By default, the dtype of a is used.An exception is when a has an integer type with less precisionthan the default platform integer. In that case, the defaultplatform integer is used instead. out : ndarray, optional Array into which the output is placed. By default, a new array iscreated. If out is given, it must be of the appropriate shape(the shape of a with axis removed, i.e.,numpy.delete(a.shape, axis)). Its type is preserved. Seedoc.ufuncs (Section “Output arguments”) for more details.

Returns :

sum_along_axis : ndarray An array with the same shape as a, with the specifiedaxis removed. If a is a 0-d array, or if axis is None, a scalaris returned. If an output array is specified, a reference toout is returned.

See also

ndarray.sum

Equivalent method.

cumsum

Cumulative sum of array elements.

trapz

Integration of array values using the composite trapezoidal rule.

mean, average

Notes

Arithmetic is modular when using integer types, and no error israised on overflow.

Examples

>>> np.sum([0.5, 1.5])
2.0
>>> np.sum([0.5, 0.7, 0.2, 1.5], dtype=np.int32)
1
>>> np.sum([[0, 1], [0, 5]])
6
>>> np.sum([[0, 1], [0, 5]], axis=0)
array([0, 6])
>>> np.sum([[0, 1], [0, 5]], axis=1)
array([1, 5])

If the accumulator is too small, overflow occurs:

>>> np.ones(128, dtype=np.int8).sum(dtype=np.int8)
-128

注意以下例子里的消失（taken)的维  

axis=0

>>> import numpy as np
>>> b=np.arange(24).reshape(2,3,4)
>>> b
array([[[ 0,  1,  2,  3],
        [ 4,  5,  6,  7],
        [ 8,  9, 10, 11]],

       [[12, 13, 14, 15],
        [16, 17, 18, 19],
        [20, 21, 22, 23]]])
>>> b.sum(axis=0)
array([[12, 14, 16, 18],
       [20, 22, 24, 26],
       [28, 30, 32, 34]])

（2,3,4）变为（3，4） 下标2的维消失了。和是怎么计算的呢？看下面b的展开样式：

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

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

对照  b.sum(axis=0) 结果 ，很容易看出怎么求和的了吧？  下标2的0维消失，却对下标4的2维的元素求和。

<pre name="code" class="python">>>> b.sum(axis=1)
array([[12, 15, 18, 21],
       [48, 51, 54, 57]])

（2,3,4）变为（2，4） 下标3的维消失了。和是怎么计算的呢？看下面b的展开样式：

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

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

>>> b.sum(axis=2)
array([[ 6, 22, 38],
       [54, 70, 86]])

（2,3,4）变为（2，3） 下标4的维消失了。和是怎么计算的呢？看下面b的展开样式：

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

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

总结：

看出规律了吗？不是是一会按列求和，一会按行求和，这么简单的！sum(axis=那个维) 那个维就消失，其他维不变，对消失的维求和

python 表示：

       b.sum(axis=0) 等同与  b[0,:,:]+b[1,:,:]

       b.sum(axis=1) 等同与  b[:,0,:]+b[:,1,:]+b[:,2,:]

       b.sum(axis=2) 等同与  b[:,:,0]+b[:,:,1]+b[:,:,2]+b[:,:,3]