On Fri, Nov 13, 2009 at 1:10 PM, Stephen Tetley
<stephen.tetley@gmail.com> wrote:
>> That is OK. Since understand the basic concept of monoid (I mean the
>> thing in actual math), the idea here is totally not hard for me. But
>> the sample here does not show why (or how) we use it in programming,
>> right?
Hi Magicloud
Conal Elliott has an interesting paper about designing your programs
in relation to the standard type classes:.
http://conal.net/papers/type-class-morphisms/
Thinking about the data structures and functions in your program with
regards the standard classes is very useful useful for clarifying your
design. And certainly if you decide your data structure fits the
Monoid interface then you will be presenting it to others who use your
program in the 'standard vocabulary'. But even for Monoid which
seemingly presents a simple interface (mempty, mappend) deciding
whether the _container_ you have is naturally a monoid can be
difficult.
A personal example, I've been developing a drawing library for a
couple of months and still can't decide whether a bounding box should
be a monoid (mempty, append) or a groupoid (just append) where append
in both cases is union. Even though I haven't resolved this problem,
having the framework of monoid versus groupoid at least gives me the
_terminology_ to consider the problem.
Watch out, in more common parlance, having just an binary operation is a magma, while having a category with full inverses yields a groupoid. I haven't seen many people use the older groupoid term for magmas, if only because they started to have naming conflicts with the category theory people, and Bourbaki's 'magma' was available and unambiguous. =)
And of course magma is not to be confused with the notion of a semigroup, which is a binary associative operation, and is therefore much more similar to a monoid in that all it lacks is a unit.
-Edward Kmett
Best wishes
Stephen