How to create diagonal array from existing array in numpy

Question

I am trying to make a diagonal numpy array from:

[1,2,3,4,5,6,7,8,9]

Expected result:

[[ 0,  0,  1,  0,  0],
 [ 0,  0,  0,  2,  0],
 [ 0,  0,  0,  0,  3],
 [ 4,  0,  0,  0,  0],
 [ 0,  5,  0,  0,  0],
 [ 0,  0,  6,  0,  0],
 [ 0,  0,  0,  7,  0],
 [ 0,  0,  0,  0,  8],
 [ 9,  0,  0,  0,  0]]

What would be an efficient way of doing this?

What's the relation between your input and expected output? have you tried anything so far? — Kasravnd
– Kasravnd, Commented May 11, 2017 at 20:10

MSeifert · Accepted Answer · 2017-05-11 20:19:52Z

2

You can use integer array indexing to set the specified elements of the output:

>>> import numpy as np
>>> a = [1,2,3,4,5,6,7,8,9]
>>> arr = np.zeros((9, 5), dtype=int)           # create empty array
>>> arr[np.arange(9), np.arange(2,11) % 5] = a  # insert a 
>>> arr
array([[0, 0, 1, 0, 0],
       [0, 0, 0, 2, 0],
       [0, 0, 0, 0, 3],
       [4, 0, 0, 0, 0],
       [0, 5, 0, 0, 0],
       [0, 0, 6, 0, 0],
       [0, 0, 0, 7, 0],
       [0, 0, 0, 0, 8],
       [9, 0, 0, 0, 0]])

answered May 11, 2017 at 20:19

MSeifert

154k41 gold badges356 silver badges377 bronze badges

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

1 Comment

hpaulj Over a year ago

np.fill_diagonal has a wrap mode like this, but it only works for the main diagonal. The other diag functions allow offsets, but no wrapping.

hpaulj · Accepted Answer · 2017-05-11 22:15:18Z

1

Inspired by np.fill_diagonal which can wrap, but not offset:

In [308]: arr=np.zeros((9,5),int)
In [309]: arr.flat[2:45:6]=np.arange(1,10)
In [310]: arr
Out[310]: 
array([[0, 0, 1, 0, 0],
       [0, 0, 0, 2, 0],
       [0, 0, 0, 0, 3],
       [0, 0, 0, 0, 0],
       [4, 0, 0, 0, 0],
       [0, 5, 0, 0, 0],
       [0, 0, 6, 0, 0],
       [0, 0, 0, 7, 0],
       [0, 0, 0, 0, 8]])

(though for some reason this has the 4th all zero row).

def fill_diagonal(a, val, wrap=False):
    ...       
    step = a.shape[1] + 1
    # Write the value out into the diagonal.
    a.flat[:end:step] = val

answered May 11, 2017 at 22:15

hpaulj

233k14 gold badges260 silver badges392 bronze badges

5 Comments

MSeifert Over a year ago

you lost the 9 and the fourth row seem wrong. Using a step works great until you have one value in the last column and expect the next value in the next row, first column :)

hpaulj Over a year ago

Yes, I"m puzzled about that. I'll leave the answer up in case someone sees the flaw in this 'strided' solution.

MSeifert Over a year ago

well, the flaw is that you need different "steps". 6 for row 1->2, 2->3, 4->5, 5->6, 6->7, 7->8 but only 1 for row 3->4 and 8->9.

hpaulj Over a year ago

This style of wrapping was the default for fill_diagonal until a version 1.6.

hpaulj Over a year ago

So when lined up just right the stride of 6 skips a whole row of length 5.

Collectives™ on Stack Overflow

How to create diagonal array from existing array in numpy

2 Answers 2

1 Comment

5 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

5 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related