From: Andrej Bauer <andrej.bauer@andrej.com>
To: Gerd Stolpmann <gerd@gerd-stolpmann.de>
Cc: saptarshi.guha@gmail.com, caml-list@yquem.inria.fr
Subject: Re: [Caml-list] The need to specify 'rec' in a recursive function defintion
Date: Wed, 10 Feb 2010 08:19:10 +0100 [thread overview]
Message-ID: <7d8707de1002092319j4d6bfca5s169914a992d57be2@mail.gmail.com> (raw)
In-Reply-To: <1265752863.5482.42.camel@flake.lan.gerd-stolpmann.de>
On Tue, Feb 9, 2010 at 11:01 PM, Gerd Stolpmann <gerd@gerd-stolpmann.de> wrote:
> For defining the operational meaning of a recursive function a special
> helper is needed, the Y-combinator. It gets quite complicated here from
> a theoretical point of view because the Y-combinator is not typable. But
> generally, the idea is to have a combinator y that can be applied to a
> function like
> y (fun f arg -> expr) arg
> and that "runs" this function recursively, where "f" is the recursion.
A small correction: y is typeable. It's a fixed-point operator, so it's type is
('a -> 'a) -> 'a
or if we only care about recursive functions it is:
# let rec y f x = f (y f) x ;;
val y : (('a -> 'b) -> 'a -> 'b) -> 'a -> 'b = <fun>
(Yes, I know I used "let rec", it doesn't matter for checking that y
has a type). Let's check that it works:
# let fac = y (fun f n -> if n = 0 then 1 else n * f (n - 1)) ;;
val fac : int -> int = <fun>
# fac 7 ;;
- : int = 5040
What Gerd probably meant was that the usual untyped lambda-calculus
definition of y, which gets us recursion out of nothing isn't typeable
because self-application requires unrestricted recursive types. But
again, it's an intermediate definition that is not typeable, not y
itself (here adapted so that it works for functions in an eager
language):
$ ocaml -rectypes
Objective Caml version 3.11.1
# let z = fun u v w -> v (u u v) w ;;
val z : ('a -> ('b -> 'c -> 'd) -> 'b as 'a) -> ('b -> 'c -> 'd) -> 'c
-> 'd = <fun>
# let y = z z ;;
val y : (('_a -> '_b) -> '_a -> '_b) -> '_a -> '_b = <fun>
Observe that y has a non-recursive type, it's the auxiliary z that
requires a recursive one. And this y also works:
# let fac = y (fun f n -> if n = 0 then 1 else n * f (n - 1)) ;;
val fac : int -> int = <fun>
# fac 7 ;;
- : int = 5040
Can anyone get rid of the pesky underscores in the type of y above, so
that it becomes truly polymorphic?
With kind regards,
Andrej
next prev parent reply other threads:[~2010-02-10 7:19 UTC|newest]
Thread overview: 25+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-02-09 20:50 Saptarshi Guha
2010-02-09 21:55 ` [Caml-list] " Guillaume Yziquel
2010-02-09 22:14 ` Saptarshi Guha
2010-02-09 22:01 ` Gerd Stolpmann
2010-02-09 21:58 ` Guillaume Yziquel
2010-02-09 22:34 ` Gerd Stolpmann
2010-02-10 0:07 ` Guillaume Yziquel
2010-02-10 3:10 ` Alain Frisch
2010-02-09 22:16 ` Saptarshi Guha
2010-02-09 23:29 ` Jon Harrop
2010-02-10 10:15 ` rossberg
2010-02-10 7:19 ` Andrej Bauer [this message]
2010-02-10 9:36 ` Francois Maurel
2010-02-10 10:12 ` rossberg
2010-02-09 23:33 ` Jon Harrop
2010-02-09 22:31 ` Saptarshi Guha
2010-02-10 0:12 ` Jon Harrop
2010-02-10 22:01 ` Stefan Monnier
2010-02-10 22:25 ` [Caml-list] " Till Varoquaux
2010-02-11 1:48 ` Jon Harrop
2010-02-15 15:46 ` Stefan Monnier
2010-02-15 17:33 ` [Caml-list] " Jon Harrop
2010-02-15 20:36 ` Stefan Monnier
2010-02-16 14:42 ` Stefan Monnier
2010-02-16 16:21 ` [Caml-list] " Ashish Agarwal
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=7d8707de1002092319j4d6bfca5s169914a992d57be2@mail.gmail.com \
--to=andrej.bauer@andrej.com \
--cc=caml-list@yquem.inria.fr \
--cc=gerd@gerd-stolpmann.de \
--cc=saptarshi.guha@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).