[gecode-users] Reified Distinct Constraint

[Christian Schulte]

Hmmm. Implementing (a naïve one) is easy enough but I fear for us it would
be not enough of general interest.

We are currently working on techniques where reification of everything is
possible. Maybe that could be a solution.

Christian

--
Christian Schulte, www.it.kth.se/~cschulte/

-----Original Message-----
From: Max [mailto:chaosangel at gmx.de] 
Sent: Tuesday, February 17, 2009 10:49 AM
To: Christian Schulte
Cc: 'gecode list'
Subject: Re: [gecode-users] Reified Distinct Constraint

I'm building a SMT like appraoch for integrating CSP into ASP.
Each constraint is represented by a boolean atom that can be true or 
false, so ...
i would be great to support a non n^2 allDistinct for the variables.
With a reified version i could use the distinct constraint in the body 
of a rule as condition.
So for SMT like systems reified versions of all constraints are welcome.

Max

Christian Schulte wrote:
> No, not really. I do not know what this distinct would be good for. A more
> useful one would be a soft distinct where the number of violations is
> counted (there, one actually also could get some propagation). Check for
> example:
> 	http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.3.7071
>
> What do you need a reified distinct for?
>
> Christian
>
> --
> Christian Schulte, www.it.kth.se/~cschulte/
>
>
> -----Original Message-----
> From: Max [mailto:chaosangel at gmx.de] 
> Sent: Tuesday, February 17, 2009 10:32 AM
> To: Christian Schulte
> Cc: 'gecode list'
> Subject: Re: [gecode-users] Reified Distinct Constraint
>
> Thank you for the quick answer.
> Is this feature planned for the release of GeCode3.0?
> I just found a mail in the archive of somebody requesting the same
feature.
> Currently i do use O(n^2) constraints for this, but i wanted to show 
> that i can transform it into a constant number of rules.
> Thats why i would need the reified version.
>
> Max
>
> Christian Schulte wrote:
>   
>> No, there is no reified version. The best you can do at emulating is use
>> excessive reification: Create O(n^2) reified disequality constraints for
>>     
> all
>   
>> variable-variable pairs, and then the reified control variable is the
>> conjunction of all the control variables from the disequalities. As said,
>> not very good.
>>
>> Christian
>>
>> --
>> Christian Schulte, www.it.kth.se/~cschulte/
>>
>> -----Original Message-----
>> From: users-bounces at gecode.org [mailto:users-bounces at gecode.org] On
Behalf
>> Of Max Ostrowski
>> Sent: Monday, February 16, 2009 2:21 PM
>> To: gecode list
>> Subject: [gecode-users] Reified Distinct Constraint
>>
>> I just found the distinct constraint, that hopefully makes all variables
>> distinct. This could greatly reduce the number of constraints, as i am
>> translating from a logic program.
>> But i would need a reified version of the distinct constraint.
>> I there any or can i at least emulate thie behaviour?
>>
>> Thank you in advance,
>> Max
>>   
>>     
>
>
>
> _______________________________________________
> Gecode users mailing list
> users at gecode.org
> https://www.gecode.org/mailman/listinfo/gecode-users
>
>