为什么会这样？ （关于矩阵行的交换）

Question

It is given that

A = np.array([[1, 2, 3],
       [4, 5, 6],
       [6, 5, 4],
       [3, 2, 1]])

and I want to exchange the second row and the third row. It doesn't matter how I should write the code, but I found something so strange.

Case 1:

B = A
A[1,:] = B[2,:]
A[2,:] = B[1,:]

Case 2:

B = []
B.append(A[1,:])
B.append(A[2,:])
A[2,:] = B[0]
A[1,:] = B[1]

Neither the case cannot realize what I want, since B is always changed even if just changing A . Why does it happen? In another different case but with a similar spirit, that is, given a=2 and b=1 , if we run

c = b
b = a
a = c

then the exchange will work well? Why?

Answer 1

Arrays are called by reference, atomic variables are called by value. Meaning, in the last example, c gets the value of b , but in the first case A and B are actually pointing to the same array, so your second line overwrites A . So, if you did B=np.copy(A) , all would be well.