为什么while循环比for循环更有效？

Question

I'm working on the Sieve of Eratosthenes and I'm analyzing how I could filter non-prime numbers from the range of candidates.

For testing, I'm using this command:

import timeit
print timeit.timeit(stmt="sieve(1000)", setup='from sieve import sieve', number=1000)

This is what I currently have:

def sieve(limit):
    primes = [2]
    pRange = range(3, limit, 2)
    while pRange:
        p = pRange.pop(0)
        filter_range = range(p**2, limit, p << 1)
        pRange = filter(lambda x: x not in filter_range, pRange)
        primes.append(p)
    return primes

Seconds => 5.42698788643

I then thought that if I was going to pop the first element from range, I might want to write it like this:

def sieve(limit):
    primes = [2]
    pRange = range(3, limit, 2)
    for p in pRange:
        filter_range = range(p**2, limit, p << 1)
        pRange = filter(lambda x: x not in filter_range, pRange)
        primes.append(p)
    return primes

but it turns out to be much slower at 15.7085180283 Seconds.

Why is this happening? Is the for loop iterating over the initial value of pRange and isn't updating?

Answer 1

for p in pRange:

This evaluates pRange , turning it into a list (Python 2) or a generator (Python 3), and then iterates through the items in the range. It does not re-evaluate pRange .

Either way will result in more loop iterations than the while loop would, because the latter is re-evaluating the range each time through the loop.

Remember, in Python, variables are all references. Even something like a = 5 means "set a to refer to an integer constant with the value 5 ." Likewise, pRange = filter(...) means "take the object that filter(...) returned, and set pRange to refer to it. Release the reference to whatever value pRange used to refer to (and possibly garbage collect it later if nothing refers to it)."