Question

Can numpy evaluate N largest elements in a multidimensional array?
Say:

import numpy as np
a=[1,2,3]
b=a.max(num_of_max_elements=2) #I know it won't work, and I want it to be 2 and 3

Answer 1

The easiest way to do this is to sort and then use the last n numbers. This can also be done without changing the original array. Here’s an example how to find the largest 10 numbers in a 10×10 array:

a = numpy.random.randint(1000, size=100).reshape(10, 10)
indices = a.ravel().argsort()
print a.ravel()[indices[-10:]]

This code prints something like

array([909, 917, 926, 936, 952, 960, 961, 964, 969, 993])