how to use numpy.lexsort without cycles

Question

Hi I want to order a numpy.array using lexsort.

import numpy as np
surnames = ['Hertz', 'Galilei', 'Hertz']
names = ['Heinrich', 'Galilao', 'Gustav']
ind = np.lexsort((names,surnames))

But how can I have an array ordered according to this indexes?

I would do

surnames2 = surnames.copy()
for i, a in enumerate(ind):
     surnames2[i] = surnames[a]

But I think there should be another way without writing the cycle by myself.

Thanks

Answer 1

If you turn surnames into a numpy array then you can simply index it with the array of integers returned by lexsort :

sorted_surnames = np.array(surnames)[ind]

If you want to stick with plain Python lists, you could do something like this:

sorted_surnames = [n for (i, n) in sorted(zip(ind, surnames), key=lambda x: x[0])]