Memoization decorator that caches function results using a least-recently-used (LRU) cache with unlimited size when maxsize=None.

Indicates that the decorated function should be memoized for performance, signaling that its results are safe to cache and reuse.

When a pure function is called repeatedly with the same arguments When computing expensive results that can be cached When implementing memoization for recursive functions like Fibonacci

Performance optimization Recursive algorithms API response caching Expensive computations with repeated inputs

functools.lru_cache(maxsize=None)(func)

functools.lru_cache(maxsize=None)(func)

{"func":"any callable function"}

• def return @ lambda recursive

• maxsize=128 maxsize=64 typed=True

Using lru_cache on functions with mutable arguments Forgetting that the cache is per-function instance Using lru_cache on methods without considering self

functools.lru_cache(maxsize=128) – bounded cache functools.cache – Python 3.9+ unbounded cache (equivalent to maxsize=None) functools.lru_cache(typed=True) – treats arguments of different types as distinct

Do not use on functions with side effects; results may be stale Mutable arguments (like lists, dicts) cannot be used as keys unless made hashable

• functools.lru_cache
• functools.cache
• functools.partial
• functools.wraps

When maxsize=None, the cache can grow without bound; use only when the set of possible arguments is bounded or memory is not a concern.

Signals to readers that the function is pure and its results are safe to cache, improving perceived performance.

Think of a librarian who remembers every book request forever (unlimited cache).

Equivalent to functools.cache in Python 3.9+; the decorator returns a wrapper that replaces the original function.

Consider using functools.cache (Python 3.9+) for clearer syntax when unbounded caching is desired.