Did you know that Python actually has built-in decorators that can add a cache to any function? (what are decorators?)

Let's say I have a function that requires a bit a computing power. For this example, I'll use a fibonacci sequence calculator.

def fib(x):
    if x <= 1:
        return 1
    return fib(x - 1) + fib(x)

Using Python's builtin functools library, I can add a "cache" in one line. A cache will remember the outputs of a compute-heavy function so the values don't need to be calculated multiple times.

If a set of inputs have already been calculated, the cache decorator will return the calculated result without running the fib() function.

from functools import cache

@cache
def fib(x):
    if x <= 1:
        return 1
    return fib(x - 1) + fib(x)

However, @cache will store an unlimited number of inputs and outputs. If we want to limit it to only, say, 1,000 cached values, we can use a "least-recently-used" or "LRU" cache.

With this type of cache, Python will only cache the 1,000 most recently used values. If a new value is calculated, the 1,001st value will be "evicted" from the cache (removed from memory).

Example 1 (note that we do not need parentheses):

from functools import lru_cache

@lru_cache
def fib(x):
    if x <= 1:
        return 1
    return fib(x - 1) + fib(x)

Example 2 (maximum size of 1000 elements):

from functools import lru_cache

@lru_cache(max_size=1000)
def fib(x):
    if x <= 1:
        return 1
    return fib(x - 1) + fib(x)