Quick Python: Memoization

Brandon Rozek

March 30, 2020

There is often a trade-off when it comes to efficiency of CPU vs memory usage. In this post, I will show how the lru_cache decorator can cache results of a function call for quicker future lookup.

from functools import lru_cache

@lru_cache(maxsize=2**7)
def fib(n):
    if n == 1:
        return 0
    if n == 2:
        return 1
    return f(n - 1) + f(n - 2)

In the code above, maxsize indicates the number of calls to store. Setting it to None will make it so that there is no upper bound. The documentation recommends setting it equal to a power of two.

Do note though that lru_cache does not make the execution of the lines in the function faster. It only stores the results of the function in a dictionary.