[Coin-standards] Technical Item! CLP

[Leonardo B. Lopes]

On Thu, 11 Apr 2002, Irv Lustig wrote:

> At 11:52 AM 4/11/02 -0500, Leonardo B. Lopes wrote:
> >So what would be a good way of representing *instances* of CLP problems?
> >Where do they become models?
> 
> It's a model when there is no data associated with the problem.
> 
> It becomes an instance when you unroll "forall" or other looping constructs 
> and plug in the data.
> 

So that is not fundamentally different than LP. How 'bout search
directives? Do you consider that part of the model or the instance? Isn't
it the case that without those the problems can often be pretty hopeless?

> >Is it sufficient to represent CLP instances as Non-linearities? Can we
> >define a few functions (cumulative, etc...) and then have a closed
> >representation for CLP?
> 
> I think you can do this, but it won't be just a "few" functions.
> 

So what's a few these days? 50, 100? more? less?

> >Could people sketch (brainstorm quality sketch) what the characteristics
> >of the representation should be?
> 
> I'll back out of that and let people who are real CP people respond.
> 

Could you invite some ILOG CP person into the list?

Cheers,
Leo.

========================================================================
Leonardo B. Lopes                                       leo at iems.nwu.edu 
Ph.D. Student                                              (847)491-8470
IEMS - Northwestern University              http://www.iems.nwu.edu/~leo