List comprehension confusion...

Manuel M. Garcia mail at manuelmgarcia.com
Sun Feb 2 05:01:47 CET 2003


On Sun, 02 Feb 2003 03:10:27 GMT, Afanasiy <abelikov72 at hotmail.com>
wrote:

>I am confused about the purpose of list comprehensions.
>What can they do which an explicit for loop cannot?

I like that now we have a pretty compact notation for 2 dimensional
lists

list_2d = [ [0] * 3 for _ in range(5) ]
list_2d[3][1] = 7
assert list_2d == [ [0,0,0], [0,0,0], [0,0,0], [0,7,0], [0,0,0] ]

### THIS DOES NOT WORK! ###
list_2d = [ [0] * 3 ] * 5
list_2d[3][1] = 7
assert list_2d == [ [0,0,0], [0,0,0], [0,0,0], [0,7,0], [0,0,0] ]

also list comprehensions are slightly faster:

-----------------------

import time

t1 = time.clock()
out0 = []
for i in xrange(5000000):
    if not (i % 3):
        out0.append(i * 0.5)
print time.clock() - t1

t1 = time.clock()
out0 = [i * 0.5 for i in xrange(5000000) if not (i % 3)]
print time.clock() - t1

-----------------------

6.31521073209
5.78931763674

-----------------------

But this is hardly a reason to use them, even with such a large list,
the amount of time saved is so tiny.

I probably abuse them:

import operator

def normalize(list0):
    t = reduce(operator.add, list0) * 1.0
    return [ x / t for x in list0 ]

def normalize(list0):
    t = 0.0
    for x in list0:
        t += x
    out0 = []
    for x in list0:
        out0.append(x / t)
    return out0

One is shorter, the other is easier to understand if you are not used
to functional programming constructs.

Manuel




More information about the Python-list mailing list