Thanks Wren. That was my guess too, but it seems not necessary: http://stackoverflow.com/questions/12103309/when-is-a-composition-of-catamorphisms-a-catamorphism
On 8/24/12 3:44 AM, Sebastien Zany wrote:>From <http://comonad.com/haskell/catamorphisms.html> we have the law:
More specifically (assuming I understood the statement correctly):
Suppose I have two base functors F1 and F2 and folds for each: fold1 :: (F1
a -> a) -> (μF1 -> a) and fold2 :: (F2 a -> a) -> (μF2 -> a).
Now suppose I have two algebras f :: F1 μF2 -> μF2 and g :: F2 A -> A.
When is the composition (fold2 g) . (fold1 f) :: μF1 -> A a catamorphism?
Given
F, a functor
G, a functor
e, a natural transformation from F to G
(i.e., e :: forall a. F a -> G a)
g, a G-algebra
(i.e., f :: G X -> X, for some fixed X)
it follows that
cata g . cata (In . e) = cata (g . e)
The proof of which is easy. So it's sufficient to be a catamorphism if your f = In . e for some e. I don't recall off-hand whether that's necessary, though it seems likely
--
Live well,
~wren
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe