How works python key=operator.itemgetter(1))?

Question

I have a matrix and I need to find max element and its number. How to rewrite it without operator (with for)?

    for j in range(size - 1):
        i, val = max(enumerate(copy[j::, j]), key=operator.itemgetter(1))
        copy = change_rows(copy, j, i)
        P = change_rows(P, j, i) 

And actually maybe you can explain what this string means?

i, val = max(enumerate(copy[j::, j]), key=operator.itemgetter(1))

Answer 1

Let's decompose this line.

i, val = max(enumerate(copy[j::, j]), key=operator.itemgetter(1))

First, enumerate() creates an iterator over copy[j::,j] that yields index-value pairs. For example,

>>> for i, val in enumerate("abcd"):
...     print(i, val)
...
0 a
1 b
2 c
3 d

Next, the max() function is for finding the largest item in a sequence. But we want it to target the values of copy[j::,j] , not the indices that we are also getting from enumerate() . Specifying key=operator.itemgetter(1) tells max() to look at the (i,val) pairs and find the one with the largest val .

This is probably better done with np.argmax() , especially because val goes unused.

>>> import numpy as np

>>> for j in range(size - 1):
...     i = np.argmax(copy[j::, j])    # Changed this line.
        copy = change_rows(copy, j, i)
        P = change_rows(P, j, i)