Slicing Numpy array using List

Question

Consider 2D Numpy array A and in-place function x like

A = np.arange(9).reshape(3,3)
def x(M):
    M[:,2] = 0

Now, I have a list (or 1D numpy array) L pointing the rows, I want to select and apply the function f on them like

L = [0, 1]
x(A[L, :])

where the output will be written to A. Since I used index access to A, the matrix A is not affected at all:

A = array([[0, 1, 2],
           [3, 4, 5],
           [6, 7, 8]])

What I actually need is to slice the matrix such as

x(A[:2, :])

giving me the desired output

A = array([[0, 1, 0],
           [3, 4, 0],
           [6, 7, 8]])

The question is now, how to provide Numpy array slicing by the list L (either any automatic conversion of list to slice or if there is any build in function for that), because I am not able to convert the list L easily to slice like :2 in this case.

Note that I have both large matrix A and list L in my problem - that is the reason, why I would need the in-place operations to control the available memory.

Do you need to create a function to do this? Why not assign values directly by slicing without the function? — busybear
– busybear, Commented Dec 17, 2020 at 16:13
The reason is that my function is much more complicated than M[:,2] = 0 — Pavel Prochazka
– Pavel Prochazka, Commented Dec 17, 2020 at 16:16
Would passing both the list and array as inputs to the function work? — busybear
– busybear, Commented Dec 17, 2020 at 16:18
I do not understand the question - the function x operates on 2D Numpy array like in the example above — Pavel Prochazka
– Pavel Prochazka, Commented Dec 17, 2020 at 16:21

Quang Hoang · Accepted Answer · 2020-12-17 16:35:15Z

1

Can you modify the function so as you can pass slice L inside it:

def func(M,L): 
    M[L,2] = 0

func(A,L)

print(A)

Out:

array([[0, 1, 0],
       [3, 4, 0],
       [6, 7, 8]])

answered Dec 17, 2020 at 16:35

Quang Hoang

151k11 gold badges64 silver badges86 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

Pavel Prochazka Over a year ago

Thanks for the answer. I found that it is indeed not possible to make a slice from a list in general (e.g. scipy-cookbook.readthedocs.io/items/ViewsVsCopies.html). An alternative solution (and maybe preferable for me) is to rearrange the rows in A such that L would refer to consecutive sequence in the rearranged matrix.

Quang Hoang Over a year ago

@PavelProchazka no, AFAIK, even then A[:3][:2] might still give you a copy, and you can modify the original array.

Collectives™ on Stack Overflow

Slicing Numpy array using List

1 Answer 1

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related