numpy array slicing, without for loop

Question

I have a question, how can I generate an array of slices from a numpy array. I can do it with for loop, what I am trying to do is:

def calc_matrix(arr, k): 
    x = np.zeros((len(arr) - k + 1, k))
    for i in range(len(arr) - k + 1):
        x[i, :] = arr[i: (i+k)]
    return x

also, can you clarify what the technical reason is that you want to get rid of the for loop ? — Edwin van Mierlo
– Edwin van Mierlo, Commented Apr 26, 2018 at 11:11
Avoiding for loops and using operations which work on many array elements is much faster in numpy for many operations. — allo
– allo, Commented Apr 26, 2018 at 11:16
@jpp arr = np.arange(20) and k is 4 for example, I meant that arr is 1D array, sorry for not mentioning that. — Django_kz
– Django_kz, Commented May 4, 2018 at 14:32

Paul Panzer · Accepted Answer · 2018-04-26 11:47:28Z

You can cut your desired result out of the Hankel matrix of arr (and an arbitrary second parameter; below we leave it out completely in which case zeros are used per default):

>>> import numpy as np                                                                                              
>>> from scipy import linalg
>>>
>>> arr = np.arange(10)**2 % 7   ##  just a random example      
>>> arr
array([0, 1, 4, 2, 2, 4, 1, 0, 1, 4])                                            
>>> k = 4                                                                                                           
>>> linalg.hankel(arr)[:arr.size-k+1, :k]                                              
array([[0, 1, 4, 2],                                                                                                
       [1, 4, 2, 2],                                                                                                
       [4, 2, 2, 4],                                                                                                
       [2, 2, 4, 1],                                                                                                
       [2, 4, 1, 0],
       [4, 1, 0, 1],
       [1, 0, 1, 4]])

Or you could use http://scikit-image.org/docs/dev/api/skimage.util.html#skimage.util.view_as_windows.

Collectives™ on Stack Overflow

numpy array slicing, without for loop

1 Answer 1

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related