What is row major and column major in numpy?

Question

I am new to numpy and i was reading numpy but i am not able to understand row major and column major in numpy can anyone explain with example with easiest way? Any answer would be appreciated thanks

Answer 1

Consider an array such as this:

>>> A = np.random.randint(low=1, high=9, size=(3,3))
>>> A   
array([[8, 7, 2],
       [4, 2, 5],
       [8, 6, 7]])

Using row-major means the values would be stored in memory like this (assuming 64-bit integers):

Memory address  0x00  0x08  0x10  0x18  0x20  0x28  0x30  0x38  0x40
Value              8     7     2     4     2     5     8     6     7

Whereas column-major storage would look like this:

Memory address  0x00  0x08  0x10  0x18  0x20  0x28  0x30  0x38  0x40
Value              8     4     8     7     2     6     2     5     7

Numpy stores in row-major order by default.

>>> A[0].__array_interface__['data']
(14502656, False)
>>> A[1].__array_interface__['data']
(14502680, False)  # 14502680 - 14502656 == 24

You can see the second row of data is 24 bytes (three int64 worth) away from the first. The transposed array offers a view into the original array data, not a copy, which is strided in a way that appears column major (the actual data in memory remains the same order):

>>> A.T[0].__array_interface__['data']
(14502656, False)
>>> A.T[1].__array_interface__['data']
(14502664, False)  # 14502664 - 14502656 == 8