Caching Referentially Transparent Functions (Python)
Because such a function will always produce the same output given a same input, it is quite easy to cache.
Note: this is an illustration, with shortcomings. For example: calling a function with named arguments in different orders will cause a cache miss, although this is still an equivalent call.
cache = {}
def cached(func):
def wrapper(*args, **kwargs):
global cache
try:
return cache[str((args, kwargs))]
except KeyError:
res = func(*args, **kwargs)
cache[str((args, kwargs))] = res
return res
return wrapper
@cached
def fibonacci(n):
if n == 0: return 0
if n == 1: return 1
else: return fibonacci(n-1) + fibonacci(n-2)
@cached
def make_full_name(first_name, last_name):
return ("%s %s" % (first_name, last_name))
Backward Induction
A blog by Oliver Stollmann
oliver@stollmann.net
