Python timeit.timeit - 排序的片段版本比使用 lambda 运行得更快，为什么？

Question

I'm trying to measure the time of sorting using timeit.timeit for code in two versions - snippet or normal using lambda:

bubble='''
elements = [1,22,4,632,12,3432,54531,2,2,2,2,2,4,54,3,2,1,54,43,11,22,32]
def bubble_sort(elements):
    n=0
    for i in range(len(elements)-n):
        for j in range(len(elements)-1):
            if elements[j]>elements[j+1]:
                elements[j+1],elements[j] = elements[j], elements[j+1]
                n+=1
    return elements
'''

and normal:

elements = [1,22,4,632,12,3432,54531,2,2,2,2,2,4,54,3,2,1,54,43,11,22,32]
def bubble_sort(elements):
    n=0
    for i in range(len(elements)-n):
        for j in range(len(elements)-1):
            if elements[j]>elements[j+1]:
                elements[j+1],elements[j] = elements[j], elements[j+1]
                n+=1
    return elements

When i run timeit.timeit(stmt=bubble, number=1000) it give me result of 0.00026829999978872365 but when I use

 t = Timer(lambda: bubble_sort(elements))
print(t.timeit(number = 1000))

it gives 0.04926030000024184 which is much slower. Can someone explain me why?

Answer 1

To confirm @Craig's astute observation, actually calling the function within the snippet version confirms they're pretty much exactly as fast.

import timeit

bubble='''
def bubble_sort(elements):
    n=0
    for i in range(len(elements)-n):
        for j in range(len(elements)-1):
            if elements[j]>elements[j+1]:
                elements[j+1],elements[j] = elements[j], elements[j+1]
                n+=1
    return elements
bubble_sort([1,22,4,632,12,3432,54531,2,2,2,2,2,4,54,3,2,1,54,43,11,22,32])
'''

def bubble_sort(elements):
    n=0
    for i in range(len(elements)-n):
        for j in range(len(elements)-1):
            if elements[j]>elements[j+1]:
                elements[j+1],elements[j] = elements[j], elements[j+1]
                n+=1
    return elements

print(timeit.timeit(stmt=bubble, number=10000))
print(timeit.timeit(stmt=lambda: bubble_sort([1,22,4,632,12,3432,54531,2,2,2,2,2,4,54,3,2,1,54,43,11,22,32]), number=10000))
print(timeit.timeit(stmt=lambda: list(sorted([1,22,4,632,12,3432,54531,2,2,2,2,2,4,54,3,2,1,54,43,11,22,32])), number=10000))

And, of course, that you shouldn't be rolling your own bubble sort except for practice.

0.569853096
0.5863851120000001
0.00920665399999998