3

I found the below code that creates overlapping subarrays of a given length. It does what I want except that it also reverses the order of the elements, which I do not want. I couldn't really find documentation on the 'shape' and 'strides' arguments and so do not know how to change that behavior. How do I change the 'out = ...' line so that the elements are not reversed?

import numpy as np

x = np.array([2,3,1,0])
L = 3 # Row length
strided = np.lib.stride_tricks.as_strided
n = x.strides[0]
out = strided(x[L-1:],shape=(x.size-L+1,L),strides=(n,-n))
print out
user2099143
  • 65
  • 4
  • Possible duplicate of [Using strides for an efficient moving average filter](http://stackoverflow.com/questions/4936620/using-strides-for-an-efficient-moving-average-filter) – Benjamin Feb 06 '17 at 19:04

1 Answers1

1

you need to keep all the data :

out = strided(x,shape=(x.size-L+1,L),strides=(n,n))

For

[[2 3 1]
 [3 1 0]]

strided don't check that access is in the scope, so in case of mistake everything can happen.

B. M.
  • 18,243
  • 2
  • 35
  • 54