Python/Numpy: Vectorizing repeated row insertion in a 2D array

Question

Is it possible to vectorize the insertion of rows?

I have a large 2D numpy array arr (below) and a list of indices. For each index of arr in indices I would like to insert the row at that index back into arr row 5 times at that same index.

indices = [2, 4, 5, 9, 11, 12, 16, 18, 19]

Currently I'm just looping through all the indices and inserting new rows. This approach is slow for a large list of thousands of rows, so for performance reasons I'm wondering is it possible to vectorize this multi-point tile-type insertion?

arr = [       
        [' ', ' ', 'd'],
        [' ', 'd', ' '],
        [' ', 'd', 'd'],    # <-- reinsert arr[2] here 5 times
        ['d', ' ', ' '],
        ['d', ' ', 'd'],    # <-- reinsert arr[4] here 5 times
        ['d', 'd', ' '],    # <-- reinsert arr[5] here 5 times
        ['d', 'd', 'd'],
        [' ', ' ', 'e'],
        [' ', 'e', ' '],
        [' ', 'e', 'e'],    # <-- reinsert arr[9] here 5 times
        ['e', ' ', ' '],
        ['e', ' ', 'e'],    # <-- reinsert arr[11] here 5 times
        ['e', 'e', ' '],    # <-- reinsert arr[12] here 5 times
        ['e', 'e', 'e'],
        [' ', ' ', 'f'],
        [' ', 'f', ' '],
        [' ', 'f', 'f'],    # <-- reinsert arr[16] here 5 times
        ['f', ' ', ' '],
        ['f', ' ', 'f'],    # <-- reinsert arr[18] here 5 times
        ['f', 'f', ' ']     # <-- reinsert arr[19] here 5 times
    ]

Example of first insertion of desired result:

arr = [       
        [' ', ' ', 'd'],
        [' ', 'd', ' '],
        [' ', 'd', 'd'],    # <-- arr[2]
        [' ', 'd', 'd'],    # <-- new insert
        [' ', 'd', 'd'],    # <-- new insert
        [' ', 'd', 'd'],    # <-- new insert
        [' ', 'd', 'd'],    # <-- new insert
        [' ', 'd', 'd'],    # <-- new insert
        ['d', ' ', ' ']
        #...
      ]

Kevin · Accepted Answer · 2021-05-21 15:59:45Z

3

You could use np.repeat for this:

indices = [2, 4, 5, 9, 11, 12, 16, 18, 19]
rpt = np.ones(len(arr), dtype=int)
rpt[indices] = 5

np.repeat(arr, rpt, axis=0)

edited May 21, 2021 at 15:59

answered May 21, 2021 at 15:38

Kevin

3,4182 gold badges16 silver badges55 bronze badges

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

1 Comment

Mad Physicist Over a year ago

I would suggest replacing 20 with len(arr) here to make it clear where it comes from

Collectives™ on Stack Overflow

Python/Numpy: Vectorizing repeated row insertion in a 2D array

1 Answer 1

1 Comment

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Linked

Related