Python: Use lru_cache on lambda function, or other ways to create cache for lambda function

Question

Is there anyway to make it work?

func=i_want_it_to_cache_everything(lambda a,b:a+b)

And it has to be done in one line...

Update2:

I figured out the solution (thanks to everyone who replied!). But... There is an interesting phenomenon: caching slows down program?

import functools,datetime,timeit
@functools.lru_cache(maxsize=50000)
def euclidean_distance3(p1,p2):
    return (p1[0]-p2[0])**2+(p1[1]-p2[1])**2+(p1[2]-p2[2])**2+(p1[3]-p2[3])**2
euclidean_distance=(functools.lru_cache(maxsize=50000)(lambda p1,p2: (p1[0]-p2[0])**2+(p1[1]-p2[1])**2+(p1[2]-p2[2])**2+(p1[3]-p2[3])**2))
euclidean_distance2=lambda p1,p2: (p1[0]-p2[0])**2+(p1[1]-p2[1])**2+(p1[2]-p2[2])**2+(p1[3]-p2[3])**2
print(datetime.datetime.now())
def test1():
    for z in range(50):
        for i in range(200):
            for j in range(200):
                euclidean_distance((i,i,i,i),(j,j,j,j));
def test2():
    for z in range(50):
        for i in range(200):
            for j in range(200):
                euclidean_distance2((i,i,i,i),(j,j,j,j));
def test3():
    for z in range(50):
        for i in range(200):
            for j in range(200):
                euclidean_distance3((i,i,i,i),(j,j,j,j));
t1=timeit.Timer(test1)
print(t1.timeit(1))
t2=timeit.Timer(test2)
print(t2.timeit(1))
t3=timeit.Timer(test3)
print(t3.timeit(1))

print(euclidean_distance.cache_info())
print(euclidean_distance3.cache_info())

output:

9.989034592910151
4.936129879313011
10.528836308312947
CacheInfo(hits=1960000, misses=40000, maxsize=50000, currsize=40000)
CacheInfo(hits=1960000, misses=40000, maxsize=50000, currsize=40000)

Doesn't seem like a good way of timing, you should use the timeit module. — jamylak
– jamylak, Commented Apr 22, 2012 at 18:04
Caching is not free: there is real code involved in managing the cache, checking for cache hits, running your code if there is no cached result, etc. In this case you are doing a simple calculation which can execute faster than the cache implementation. If your function was more expensive caching might be a win. — Wichert Akkerman
– Wichert Akkerman, Commented Apr 22, 2012 at 18:06
@WichertAkkerman Still cached function is outperformed by simple function, even there are 2M hits and only 4k misses. See my update. — aaronqli
– aaronqli, Commented Apr 22, 2012 at 18:14
Putting items into a dictionary involves a bit of mathematics (hash functions) as well. Another factor are extra function calls: a functional is not a cheap operation and adding caching adds at least one extra function call for every operation. In general you only want to use this kind of caching for operations take a really long time or that require accessing an external system such as a SQL server — Wichert Akkerman
– Wichert Akkerman, Commented Apr 23, 2012 at 8:02

jamylak · Accepted Answer · 2012-04-22 17:55:44Z

2

>>> from functools import lru_cache
>>> x = lru_cache()(lambda a,b:a+b)
>>> x(2,3)
5
>>> x(4,2)
6
>>> x(2,3)
5
>>> x.cache_info()
CacheInfo(hits=1, misses=2, maxsize=100, currsize=2)

answered Apr 22, 2012 at 17:55

jamylak

135k30 gold badges238 silver badges240 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

aaronqli Over a year ago

I actually just worked out the same solution myself. But it looks like caching actually slows down the program. See edit.

Collectives™ on Stack Overflow

Python: Use lru_cache on lambda function, or other ways to create cache for lambda function

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related