Numpy on the superposition of two arrays

numpyThere's a lot about botharray的操作,How they stack up and what dimensions they stack up to are often confusing,The total price of this article.

一、 Overlay operation

1. add
   逐元素操作,若是array + 常量broadcast to every element,若是array + arraytwo are requiredarray的shape一致.

end = '\n'+'*'*50+'\n'
a = np.array([[1,2],[3,4]])
b = np.array([[5,6], [7,8]])
print(a+1, end=end)
print(a+b)

输出：

[[2 3]
 [4 5]]
**************************************************
[[ 6  8]
 [10 12]]

2. concatenate
   Put two in the specified dimensionarray进行拼接,不会改变array的维度,要求两个array的维度相同,And in the specified splice dimensionarray size也必须相等,值得注意的是,两个array的shape不一定一致,比较拗口,看下例子：

end = '\n'+'*'*50+'\n'
a = np.array([[1,2],[3,4]])
b = np.array([[5,6], [7,8]])
c = np.array([[5,6]])
d = np.array([5,6])
f = np.array([[5]])

a & c 的shape并不相同：

维度不一致：
长度不一致：

3. stack & hstack
   stack Splicing two verticallyarray（可以指定axis）,要求两个array的shape严格一致,The result adds dimension;
   hstack 在水平方向（第0维,Not availableaxis）拼接两个array,The dimensions of both are required to be the same,且在第0维的size也相同,The stitching result does not change the original dimension .
   stack:The result will be added to the original dimension1
   
   hstack,a和c在0轴的size不同：
   hstack,Normal stitching after transposition：

二、 size & ndim & shape

These three concepts are easily confused,其中：
size：指的是“元素个数”;
ndim：维度,指的是“How many levels of nesting there are”,That is, the number of square brackets,同时和shape的len相等,axis=0Refers to the outermost square brackets,Incrementally inward;
shape：size和ndim综合起来就是shape,两个array的shapeSame means bothndimthe same and in every oneaxis上的size也相同.
看下例子：

值得注意的是,ndim和size以及shape没有必然联系,一个ndim不为0的arrayPossibly none of the elements：

size相同,shape和ndim也不一定相同：