Creating numpy array from calculations across arrays

Question

I currently have the task of creating a 4x4 array with operations performed on the cells

Below you will see a function that takes in array into function the_matrix which returns adj_array

It then has a for loop that is supposed to loop through array, looking at the the cell in ref_array and upon finding the matching first two numbers in array (like 6,3") it will put that function lambda N: 30 into it's respective cell in adj_array, as it will do for all cells in the 4x4 matrix

Essentially the function should return an array like this

array([[inf, <function <lambda> at 0x00000291139AF790>,
<function <lambda> at 0x00000291139AF820>, inf],
[inf, inf, inf, <function <lambda> at 0x00000291139AF8B0>],
[inf, inf, inf, <function <lambda> at 0x00000291139AF940>],
[inf, inf, inf, inf]], dtype=object)

My work so far below

def the_matrix(array):
  

    ref_array = np.zeros((4,4), dtype = object)
    ref_array[0,0] = (5,0)
    ref_array[0,1] = (5,1)
    ref_array[0,2] = (5,2)
    ref_array[0,3] = (5,3)
    ref_array[1,0] = (6,0)
    ref_array[1,1] = (6,1)
    ref_array[1,2] = (6,2)       
    ref_array[1,3] = (6,3)       
    ref_array[2,0] = (7,0)     
    ref_array[2,1] = (7,1)       
    ref_array[2,2] = (7,2)      
    ref_array[2,3] = (7,3)
    ref_array[3,0] = (8,0)
    ref_array[3,1] = (8,1)
    ref_array[3,2] = (8,2)
    ref_array[3,3] = (8,3)
 
   
    for i in ref_array:
        for a in i: #Expecting to get (5,1) here, but's showing me array
            if a == array[0, 0:2]: #This specific slice was a test
             put the function in that cell for adj_array

    return adj_array

array = np.array([[5, 1, lambda N: 120],
                  [5, 2, lambda N: 30],
                  [6, 3, lambda N: 30],
                  [7, 3, lambda N: N/30]])

Have tried variations of this for loop, and it's throwing errors. For one, the a in the for loop is displaying the input argument array, which is weird because it hasn't been called in the loop at that stage. My intention here is to refer to the exact cell in ref_array.

Not sure where I'm going wrong here and how I'm improperly looping through. Any help appreciated

Your code is not working. ref_array has shape (4,4) but you are trying to do ref_array[5,0]? — Quang Hoang
– Quang Hoang, Commented Oct 28, 2021 at 14:24
@Quang Hoang - Sorry, numbers were totally wrong, now updated to suit — Renaissance Man
– Renaissance Man, Commented Oct 28, 2021 at 14:38
You are trying to place 2-element tuples into scalar elements. Was ref_array supposed to be (4, 4, 2)? — Mad Physicist
– Mad Physicist, Commented Oct 28, 2021 at 14:41
It seems that you need a basic tutorial on numpy arrays, which you think are fancy lists — Mad Physicist
– Mad Physicist, Commented Oct 28, 2021 at 14:42
@Mad Physicist Have now edited the question to be clearer. I do get what you're saying. My current approach is to use ref_array as the reference for what values the lambda function should be applied to, and inf for all others. I presume it's not the right one, but not sure what else I should be doing — Renaissance Man
– Renaissance Man, Commented Oct 28, 2021 at 14:59

hpaulj · Accepted Answer · 2021-10-28 19:53:24Z

1

Your ref_array is object dtype, (4,4) containing tuples:

In [26]: ref_array
Out[26]: 
array([[(5, 0), (5, 1), (5, 2), (5, 3)],
       [(6, 0), (6, 1), (6, 2), (6, 3)],
       [(7, 0), (7, 1), (7, 2), (7, 3)],
       [(8, 0), (8, 1), (8, 2), (8, 3)]], dtype=object)

Your iteration, just showing the iteration variables. I'm using `repr

In [28]: for i in ref_array:
    ...:     print(repr(i))
    ...:     for a in i:
    ...:         print(repr(a))
    ...: 
array([(5, 0), (5, 1), (5, 2), (5, 3)], dtype=object)
(5, 0)
(5, 1)
(5, 2)
(5, 3)
...

So i is a "row" of the array, itself a 1d object dtype array.

a is one of those objects, a tuple.

Your description of the alternatives is vague. But assume on tries to start with a numeric dtype array

In [30]: arr = np.array(ref_array.tolist())
In [31]: arr
Out[31]: 
array([[[5, 0],
        [5, 1],
        [5, 2],
        [5, 3]],

         ...
        [8, 2],
        [8, 3]]])
In [32]: arr.shape
Out[32]: (4, 4, 2)

now the looping:

In [33]: for i in arr:
    ...:     print(repr(i))
    ...:     for a in i:
    ...:         print(repr(a))
    ...: 
array([[5, 0],          # i is a (4,2) array
       [5, 1],
       [5, 2],
       [5, 3]])
array([5, 0])           # a is (2,) array....
array([5, 1])
array([5, 2])
array([5, 3])

If "the a in the for loop is displaying the input argument array", it's most likely because a IS a an array.

Keep in mind that object dtype arrays are processed at list speeds. You might as well think of them as bastardized lists. While they have some array enhancements (multidimensonal indexing etc), the elements are still references, and are processed as in lists.

I haven't paid attention as to why you are putting lambdas in the array. It looks ugly, and I don't see what it gains you. They can't be "evaluated" at array speeds. You'd have to do some sort of iteration or list comprehension.

edit

A more direct way of generating the arr, derived from ref_array:

In [39]: I,J = np.meshgrid(np.arange(5,9), np.arange(0,4), indexing='ij')
In [40]: I
Out[40]: 
array([[5, 5, 5, 5],
       [6, 6, 6, 6],
       [7, 7, 7, 7],
       [8, 8, 8, 8]])
In [41]: J
Out[41]: 
array([[0, 1, 2, 3],
       [0, 1, 2, 3],
       [0, 1, 2, 3],
       [0, 1, 2, 3]])
In [42]: arr = np.stack((I,J), axis=2)  # shape (4,4,2)

If the function was something like

In [46]: def foo(I,J):
    ...:     return I*10 + J
    ...:

You could easily generate a value for each pair of the values in ref_array.

In [47]: foo(I,J)
Out[47]: 
array([[50, 51, 52, 53],
       [60, 61, 62, 63],
       [70, 71, 72, 73],
       [80, 81, 82, 83]])

edited Oct 28, 2021 at 19:53

answered Oct 28, 2021 at 16:01

hpaulj

233k14 gold badges260 silver badges392 bronze badges

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

2 Comments

Renaissance Man Over a year ago

Thanks for your help. Put simply the issue I'm trying to get my head around is how to take in array into the function and spit out` adj_array` that looks the first bit of code in the question. Thats all it is, but I think my approach with ref_array is going off the rails with trying to create something to reference as ref_array and then finding conditions with the for loop. I just want the_matrix function to look at array and then broadcast each value to it's respective cell in the 4x4 adj_array. I'm getting too deep in the weeds

hpaulj Over a year ago

You talk in generalities about function and adj_array, but really don't show the connection with ref_array.

Collectives™ on Stack Overflow

Creating numpy array from calculations across arrays

1 Answer 1

edit

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

edit

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related