Python Numpy memory layout

Question

I am trying to run a code

import numpy as np
a = np.zeros((3,2))
b = a.T.reshape(3*2)
b[0] = 99
a
array([[0,0],
       [0,0],
       [0,0]])

The question here is that the reshape function in numpy return a view of the original array and if any changes occurs in view object or in main object would be propogated throughout the views and main object.

But in the above stated case it is not happening. Please explain.

Answer 1

The last reshape you are doing can't be expressed in strides in the original memory layout:

orig                  1 2  3 4  5 6  ok, strides 2, 1

transpose             1 3 5  2 4 6   ok, strides 1, 2

reshaped transpose    1 3 5 2 4 6    impossible

I'm not 100% sure I understand this correctly (or, more precisely, I'm almost 100% sure I don't), but there may be some obscure cases when a copy writes back. In any case, this is not one of them.

Answer 2

The official document gives an answer to your question.

This will be a new view object if possible; otherwise, it will be a copy. Note there is no guarantee of the memory layout (C- or Fortran- contiguous) of the returned array.