Stacking 2-d numpy arrays to get a 3-d array in python

Question

I am trying to stack a few 2-D arrays to obtain a 3-D array in python. My motivation behind doing this is to plot the spatial dependence of a variable: var(x,y) where x and y are two different dimensions, for different parameter values param. A minimum reproducible example for the same has been attached below:

while (param<=2):
 var(x,y)= fun  # fun is a function which returns var(x,y) as mentioned above
 param+=1

I want to keep on stacking the 2-D arrays var(x,y) for different values of param, along a 3rd dimension. My output should be of the form :- For param=0 : var(x,y), For param=1 : var(x,y), For param=2 : var(x,y). Any help regarding the construction of the 3-D array would be highly appreciated. Please note that var(x,y) is a 2-D array.

Usually we recommend collecting the arrays in a list, and doing one array construction step at the end. Whether you use np.array, np.stack or one of the other variations on np.concatenate depends on the desired shape(s). — hpaulj
– hpaulj, Commented Jun 2, 2020 at 6:21
But for this problem, I already have the 2d arrays ready. Could you suggest me how to reconstruct a 3d array from this. — Anweshan
– Anweshan, Commented Jun 2, 2020 at 6:28
Because I don't want to complicate things more. The reason for already having 2-d arrays built is that it gives a better physical essence to the problem in terms of the spatial dependence of the variable. — Anweshan
– Anweshan, Commented Jun 2, 2020 at 6:33

Carlo · Accepted Answer · 2020-06-02 07:15:12Z

2

As suggested by hpaulj, the key idea is to have a list of the 2D arrays and then assemble them into a 3D array. It's usually more efficient to create the array in one go (no appending), so the following should be ideal:

import numpy as np

var_lst = []  # This is where you store the var arrays
while (param<=2):
    var = fun  # fun is a function which returns var(x,y) as mentioned above
    var_lst.append(var)
    param+=1

# Creates a 3D array with the following axes (x, y, param)
var_arr = np.stack(var_lst, 2)

# If later you want different axes, like (param, x, y), you can always swap them around with transpose
var_arr = var_arr.transpose(2, 0, 1)

edited Jun 2, 2020 at 7:15

answered Jun 2, 2020 at 7:08

Carlo

18511 bronze badges

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

4 Comments

hpaulj Over a year ago

With np.stack you can chose the axis and skip the swap.

Anweshan Over a year ago

Thanks for your answer. One question :- keeping up with generality, if I am sweeping param *** n*** number of times, then should I use var_arr=np.stack(var_list,(n-1)), one more question:- how do I access the elements of the 3-D array along the 3rd dimension. Will normal indexing technique work?

Carlo Over a year ago

1st question: no, the 2 in np.stack just indicated the axis on which the arrays within the list are to be stacked. For example, if var(x,y) is a 2x2 array, and if param=n, by doing var_arr = np.stack(var_lst, 2) the shape of var_arr will be 2x2xn. If you were to do var_arr = np.stack(var_lst, 0), the shape of var_arr would now be nx2x2, because you are stacking on the first axis. 2nd question: the normal indexing works, you if you want var(x, y) for param=3 you just do var_arr[:, :, 3]

Anweshan Over a year ago

My question was not for ``` param=n``` . I meant to ask : if i am sweeping the parameter a few times (let's say n number of times). The what should I do?

Collectives™ on Stack Overflow

Stacking 2-d numpy arrays to get a 3-d array in python

1 Answer 1

4 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

4 Comments

Your Answer

Sign up or log in

Post as a guest

Related