I am skeptical that this is useful for any argument but 2 (or maybe powers of 2), but it does seem likely that we'll be able to optimize this effectively for constant powers, so the generalization probably doesn't hurt.
I am skeptical that this is useful for any argument but 2 (or maybe powers of 2), but it does seem likely that we'll be able to optimize this effectively for constant powers, so the generalization probably doesn't hurt.