Comments on: A Haskell monad for infinite search in finite time

By: Searchable Data Types « Luke Palmer

Searchable Data Types « Luke Palmer — Thu, 18 Nov 2010 03:48:38 +0000

[...] space" (infinite streams of bits). He then showed that spaces that can be thus searched form a monad (which I threw onto hackage), and wrote a paper about the mathematical foundations which is [...]

By: BMeph

BMeph — Thu, 08 Jan 2009 20:21:05 +0000

Sorry for commenting so late in the game, but I also noticed: aren’t forevery and forsome just folds, with the arguments reversed? In particular, using the “S is a Monad” trait, then forevery and forsome are like disjunctive and conjunctive monoid versions of MonadPlus, except that MonadPlus isn’t exactly a monoid (and thus, isn’t a Monoid, either).

By: Martin Escardo

Martin Escardo — Mon, 29 Dec 2008 23:53:59 +0000

And, of course, you define >>= recursively, in a way that is unlikely to produce a total function. Hence it is hard to see this as a generalization of the search monad.

By: Martin Escardo

Martin Escardo — Mon, 29 Dec 2008 23:39:32 +0000

This looks like an interesting idea. I have some questions. Have you checked that this is indeed a monad? Does “sequence” converge for an infinite lazy list? (When? Always?) In what sense is your invariant an invariant? What does the monad multiplication (=join) do? There is a candidate of a morphism from the search monad to your “inversion” monad, but not the other way round. And there is a morphism from the search monad to the continuation monad. But there aren’t maps connecting the monads in the other directions (according to djinn).

By: migmit

migmit — Mon, 22 Dec 2008 20:02:45 +0000

Sorry, some symbols got lost; I’ve uploaded all code (with a minor changes in notation) here: http://hpaste.org/13260