Memoization of function results to avoid redundant computation.

Indicates that the decorated function's results should be cached for performance optimization.

When a pure function is called repeatedly with the same arguments and the computation is expensive (e.g., recursive algorithms, expensive calculations, I/O‑bound functions).

Performance‑critical code, recursive functions (e.g., Fibonacci), expensive pure functions, API call caching, dynamic programming.

@functools.lru_cache(maxsize=None)

@functools.lru_cache

maxsize=<int or None>

• fibonacci
• fib
• expensive_func
• recursive_func
• api_call
• dp_func

• maxsize=128
• maxsize=256
• maxsize=64
• @functools.lru_cache() (default maxsize=128)
• @functools.cache (unbounded
• Python 3.9+)

Using unhashable arguments (e.g., lists, dicts) as function parameters; forgetting that the cache grows without bound when maxsize=None; assuming the cache is automatically cleared; confusing @lru_cache with @cache.

@functools.lru_cache(maxsize=128) (bounded cache), @functools.lru_cache() (default maxsize=128), @functools.cache (unbounded cache, Python 3.9+), @functools.lru_cache(maxsize=None, typed=True) (separate cache per argument type).

Confusing @lru_cache with @cache; assuming arguments are hashable when they are not; expecting automatic cache invalidation; memory growth with an unbounded cache.

• functools decorators
• memoization patterns

Setting maxsize=None provides an unlimited cache, which can lead to unbounded memory consumption if the function is called with many distinct argument combinations.

Signals that the function is pure (deterministic) and expensive, indicating an intentional optimization that trades memory for speed.

@lru_cache

Use functools.cache (Python 3.9+) for unbounded caching; beware of unbounded memory growth with maxsize=None.

@functools.cache (unbounded cache introduced in Python 3.9) or using functools.lru_cache with a finite maxsize for bounded caching