[Python] Efficiently Finding the Indices of the N Largest Values in a NumPy Array Without...

I'm working with very large NumPy arrays (millions to billions of elements) and need to find the indices of the N largest values in the array. Using np.argsort() followed by slicing to get the last N elements is too slow and memory-intensive for my use case, as it sorts the entire array.

Is there a more efficient way to find the indices of the N largest values without sorting the entire array? Ideally, the solution would have a time complexity better than O(n log n) and avoid creating large intermediate arrays.

For example, given the array:

import numpy as np
arr = np.array([1, 5, 2, 8, 3, 9, 4, 7, 6])
N = 3


The desired output would be the indices of the 3 largest values (9, 8, 7):

[5, 3, 7] # Indices of 9, 8, and 7

Continue reading...