Numpy 数组切片的views and copy

检查发现自己代码中的bug是由如下原因导致，

1. numpy数组的切片是views而不是copy

2. += 符号没有copy新的对象，直接覆盖原地址，类似使用pandas的时候设置inplace=True

我下面给出一个示例代码以及代码的输出结果，以后写的时候需要注意。

import numpy as np


a = np.array([1, 2, 3, 4, 5])
b = a[2:4]
print("views of a[2:4]", b)
b += 10
print("b is changed to", b)
print("a is changed to", a)

b = b + 10
print("b is new object. ", b)
print(f"a is not changed. {a}")

输出结果：
views of a[2:4] [3 4]
b is changed to [13 14]
a is changed to [ 1  2 13 14  5]
b is new object.  [23 24]
a is not changed. [ 1  2 13 14  5]