How to modify an array for a specific purpose

Question

I am working on some array T and a condition based on the array D:

import numpy as np
T = np.array([[ 1,  2],
              [ 3,  4],
              [ 5,  6],
              [ 7,  8]])
D = np.array([-6, 1, 5, -8])

What I want to do for each row of T:

if D<0 replace first value by the second value and second value by the opposite of first value, for example [ 1, 2] -> [ 2, -1] as D[0]<0
if D>0 replace first value by the opposite of second value and second value by the first value, for example [ 3, 4] -> [-4, 3] as D[1]>0

I did this by using this (ugly) code:

import numpy as np
T = np.array([[ 1,  2],
              [ 3,  4],
              [ 5,  6],
              [ 7,  8]])
D = np.array([-6, 1, 5, -8])
n = len(T)
NrT = T.copy()
for idx in range(n):
    if D[idx] > 0:
        NrT[idx,0] = -T[idx,1] 
        NrT[idx,1] = +T[idx,0] 
    else:
        NrT[idx,0] = +T[idx,1] 
        NrT[idx,1] = -T[idx,0] 
NrT

Would it be possible to use some vectorized function lambda to do the same thing more elegantly, in the possible shape:

f = lambda *some expression I was not able to find*
NrT = np.apply_along_axis(f, axis=1, arr=T)

I'm not used to using lambda functions. Is there perhaps an easier way to do it?

maybe create normal function def f() instead of lamba. And later you may try to reduce it to lamba. But maybe normal function would be simpler and more readable. And normal function you still can use in apply_along_axis — furas
– furas, Commented Nov 4 at 18:57
apply_along_axis iterates on the other axis/axes, sending 1d arays to your (lambda) function. With a 2d array like T you could just as easily (and as fast) iterate on axis 0, passing T[i] to the function. So it's not a true 'vectorization'. And apply doesn't work with 2 arrays. — hpaulj
– hpaulj, Commented Nov 4 at 19:09
Think about using idx1=D>0 and idx2=D<=0. And then NrT[idx1] = ...T[idx1] etc — hpaulj
– hpaulj, Commented Nov 4 at 19:16

strawdog · Accepted Answer · 2025-11-04 19:17:09Z

3

imho, the easiest way is to prepare an array to multiply original one with, based on your reshaped D array. And then mutiply it with your flipped original T array.

m = np.where(D[:, None]>0, [-1, 1], [1, -1])
res = np.flip(T,1)*m

res:

[[ 2 -1]
 [-4  3]
 [-6  5]
 [ 8 -7]]

P.S. a side-note: I believe using capital letters in array names is not a good idea, especially T - this letter used for transpose method and may make your code look confusing.

edited Nov 4 at 19:17

answered Nov 4 at 19:03

strawdog

7321 gold badge7 silver badges14 bronze badges

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

Collectives™ on Stack Overflow

How to modify an array for a specific purpose

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related