On Mon, Dec 7, 2015 at 8:31 AM Serhiy Storchaka <storchaka@gmail.com> wrote:

On 07.12.15 02:41, Michael Selik wrote:
> On Sat, Dec 5, 2015 at 6:11 PM Serhiy Storchaka
> <storchaka@gmail.com
> <mailto:storchaka@gmail.com>> wrote:
>
> On 04.12.15 23:44, Bill Winslow wrote:
> > def deterministic_recursive_calculation(input, partial_state=None):
> > condition = do_some_calculations(input)
> > if condition:
> > return deterministic_recursive_calculation(reduced_input,
> > some_state)
> >
> > I want to memoize this function for obvious reasons, but I need the
> > lru_cache to ignore the partial_state argument, for its value
> does not
> > affect the output, only the computation expense.
>
> Memoize a closure.
>
> def deterministic_calculation(input):
> some_state = ...
> @lru_cache()
> def recursive_calculation(input):
> nonlocal some_state
> ...
> return recursive_calculation(reduced_input)
> return recursive_calculation(input)
>
>
> This would provide the dynamic programming aspect for the recursion, but
> not cache across successive calls to the outer function. It does look
> cleaner than the OO version.

Decorate the outer function too.

Wouldn't that put us back where we started -- the cache is inappropriately keying on the state/shortcut?

@lru_cache()

def recursive(*args, shortcut=False):

@lru_cache()

def inner(*args):

if shortcut: return 'took a shortcut'

print('infinite recursion is fun!')

return inner(*args)

return inner(n)

Creating a separate wrapper would do it, so that the recursive function isn't re-created each time.

@lru_cache()

def recursive(*args):

if recursive.shortcut: return 'took a shortcut'

print('infinite recursion is fun!')

return recursive(*args)

recursive.shortcut = False

def wrapper(*args, shortcut=False):

recursive.shortcut = shortcut

return recursive(*args)