np.array index slicing bethween conditions

Question

I need to slice an array's index from where a first condition is true to where a second condition is true, these conditions are never true at the same time, but one can be true more than one time before the other occurs.
I try to explain:

array_filter = np.array([3,4,5,6,4,3,2,3,4,5])
array1 = np.array([2,3,4,6,3,3,1,2,3,4])
array2 = np.array([3,5,6,7,5,4,3,3,5,6])

array1_cond = array1 >= array_filter
array2_cond = array2 <= array_filter


                0  1  2   3  4  5  6   7  8  9
array_filter    3  4  5   6  4  3  2   3  4  5 
array1          2  3  4   6  3  3  1   2  3  4
array1_cond               ^     ^               (^ = True)
array2          3  5  6   7  5  4  3   3  5  6    
array2_cond     ^                      ^ 

expected_output 2  3  4 | 7  5  4  3 | 2  3  4
                 array1 |   array2   | array1

EXPECTED OUTPUT:

expected_output[(array2_cond) : (array1_cond)] = array1[(array2_cond) : (array1_cond)]
expected_output[(array1_cond) : (array2_cond)] = array2[(array1_cond) : (array2_cond)]
expected_output = [ 2, 3, 4, 7, 5, 4, 3, 2, 3, 4 ]

I'm so sorry if syntax is a little confusing, but idk how to make it better... <3
How can I perform this?
Is it possible WITHOUT LOOPS?

I don't see any loop in your code. Also, please take the tour and learn How to Ask. In order to get help, you will need to provide a minimal reproducible example. If your question include a pandas dataframe, please provide a reproducible pandas example — alec_djinn
– alec_djinn, Commented Oct 14, 2022 at 9:28
I'm sorry, but this is the best minimal reproducible example I can provide, if I knew how to make it, I wouldn't need help... @alec_djinn — Francesco Battisti
– Francesco Battisti, Commented Oct 14, 2022 at 9:32
Are you 1. looking for an index of the first element that satisfies a condition - let's call it j 2. Concatenate parts of two arrays, array_1[:j] + array_2[j:] - is this what you are doing? — Yulia V
– Yulia V, Commented Oct 14, 2022 at 9:34
Can you explain the logic behind the output? Why start with 2, 3, 4 and not 3, 5, 6? — T C Molenaar
– T C Molenaar, Commented Oct 14, 2022 at 9:35
Bacause 3, 5, 6 is in array2 and array2_cond is True at these points @TCMolenaar — Francesco Battisti
– Francesco Battisti, Commented Oct 14, 2022 at 9:46

user7138814 · Accepted Answer · 2022-10-14 13:24:31Z

1

This works for your example, with a, b in place of array1, array2:

nz = np.flatnonzero(a_cond | b_cond)
lengths = np.diff(nz, append=len(a))
cond = np.repeat(b_cond[nz], lengths)
result = np.where(cond, a, b)

If at the start of the arrays neither condition holds true then elements from b are selected.

answered Oct 14, 2022 at 13:24

user7138814

2,05112 silver badges12 bronze badges

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

3 Comments

Francesco Battisti Over a year ago

Honestly I don't understand but ok, works perfectly and at high speed...thank you so much <3 @user7138814

user7138814 Over a year ago

@Francesco I wish I knew a more straightforward solution with numpy. Personally I think in this case a simple loop accelerated with numba jit would be the best.

Francesco Battisti Over a year ago

>1ms calculations for an array of 5000 floats (with this method) is more than satisfating my friend :)

Collectives™ on Stack Overflow

np.array index slicing bethween conditions

1 Answer 1

3 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

3 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related