Given two arrays A and B, how to efficiently combine them so that the output is an array in which each element is a tuple (Aij,Bij)?)

Question

Consider two arrays A and B, both of dimension NxN. I wish to generate a new array NxN such that each new element is a tuple (or list, doesn't really matter) of the type (A_ij,B_ij).

I can do it by running element by element like:

def recombine(A,B):
    NewArray=[]
    for i in range(len(A)):
        NewArray.append([])
        for j in range(len(B)):
            NewArray[i].append(A[i][j],B[i][j])
    return NewArray

which is a very memory consuming algorithm.

I was wondering if there is a more efficient way (taking advantage of numpy array for doing this).

To clarify, let's consider the following example (where, for simplicity, I am considering simple lists instead of np arrays):

A=[[1,2],[3,4]]
B=[[10,20],[30,40]]

#I want an output in the form:

[[(1,10),(2,20)],[(3,30),(4,40)]]

One possibility is np.stack([A, B], axis=-1). But that doesn't produce an array of tuples - it produces a 3D array. That's more memory efficient than an array of tuples. Do you need tuples specifically? — Nick ODell
– Nick ODell, Commented Oct 20, 2023 at 2:33
Thank you, @NickODell! Your suggestion works perfectly. I don't need tuples specifically, I just used them to make it more visually clear. — delta_omega
– delta_omega, Commented Oct 20, 2023 at 3:16

tdelaney · Accepted Answer · 2023-10-20 02:53:44Z

1

You can use nested zip. zip yields tuples from a set of iterables where each item in the tuple is the next value of each iterable in turn. First you zip A and B which yields sublists in A and B, then a second zip to flip them.

>>> A=[[1,2],[3,4]]
>>> B=[[10,20],[30,40]]
>>> [list(zip(*zipped)) for zipped in zip(A,B)]
[[(1, 10), (2, 20)], [(3, 30), (4, 40)]]

answered Oct 20, 2023 at 2:53

tdelaney

77.9k6 gold badges91 silver badges129 bronze badges

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

Comments

vht981230 · Accepted Answer · 2023-10-20 02:33:52Z

1

I think np.dstack can be used in this case but. You can use the built-in zip function if you don't want to use external libraries. Theoretically, the performance should be relatively the same O(n*n) since you have to iterate through all elements in the 2D array to generate a new array but numpy should be slightly more efficient due to memory, locality optimization

A=[[1,2],[3,4]]
B=[[10,20],[30,40]]

def recombine(A,B):
    return [list(zip(A[i], B[i])) for i in range(len(A))]

print(recombine(A,B))

import numpy as np

print(np.dstack((A,B)))

answered Oct 20, 2023 at 2:33

vht981230

5,3252 gold badges11 silver badges18 bronze badges

Comments

Chams Agouni · Accepted Answer · 2023-10-20 02:43:21Z

1

import numpy as np

A = np.array([[1, 2], [3, 4]])
B = np.array([[10, 20], [30, 40]])

def combine(A, B):
    result_array = np.dstack((A, B))
    return [[tuple(pair) for pair in row] for row in result_array]

print(combine(A, B))

answered Oct 20, 2023 at 2:43

Chams Agouni

3742 silver badges13 bronze badges

Collectives™ on Stack Overflow

Given two arrays A and B, how to efficiently combine them so that the output is an array in which each element is a tuple (Aij,Bij)?)

3 Answers 3

Comments

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

Comments

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related