how to compare current and last element iterated from numpy without loop-CodePudding

In 2d array get the last column and compare current value with its last iterated value and if the difference is == 1 then get row indexes of both of them . I able to do it using for loop but it gets slow when array grows as there are multiple conditions next after getting indexes

x=np.array ([[79, 50, 18, 55, 35], 
            [46, 71, 46, 95, 80], #1
            [97, 37, 71,  2, 79], #2
            [80, 96, 60, 85, 72],
            [ 6, 52, 63, 86, 38],
            [35, 50, 13, 93, 54], #5
            [69, 21,  4, 40, 53], #6           
            [18, 34, 91, 67, 89],
            [82, 16, 16, 24, 80]])

last_column = x[:,[-1]]
for kkk in range(1,len(last_column)):
    if last_column[kkk] == last_column[kkk-1]   1 \ 
    or last_column[kkk] == last_column[kkk-1] - 1 :
        print('range',last_column[kkk],last_column[kkk-1])

ouput is -
[80]
[79]
range [79] [80]
[72]
[38]
[54]
[53]
range [53] [54]
[89]
[80]

CodePudding user response：

I don't think there is a way of doing this in python that doesn't use a loop (or a function that utilizes a loop). But I would suggest something like this, it worked quite well for me:

import numpy as np

x = np.array([[79, 50, 18, 55, 35], [46, 71, 46, 95, 80], [97, 37, 71, 2, 79],[80, 96, 60, 85, 72], [ 6, 52, 63, 86, 38], [35, 50, 13, 93, 54], [69, 21,  4, 40, 53], [18, 34, 91, 67, 89], [82, 16, 16, 24, 80]])

last_column = x[:,-1]

for kkk in range(1,len(last_column)):
    if abs(last_column[kkk] - last_column[kkk-1]) == 1:
        print('range',last_column[kkk],last_column[kkk-1])

This performed quite well on my device in about, 0.00023s possibly less.

CodePudding user response：

You can use np.diff to compute the difference between all the numbers, and select those that are 1 or -1, and use np.pad to shift the resulting mask to get the n-1 items, and finally zip them together:

diff = np.insert(np.diff(x[:, -1]), 0, 0)
ix = (diff == -1) | (diff == 1)
z = zip(x[:,-1][ix], x[:,-1][np.pad(ix, (0,1))[1:]])

for first, second in z:
    print('range', first, second)

Output:

range 79 80
range 53 54