Re: [eclipse-clp-users] issues with ic and simplification

Ulrich Scholz wrote:
> Dear all,
> 
> it seems that ic on the order and kind of the constraints that are used.
> 
> For example, why does ECLiPSe find the correct answer with "[eclipse 5]" but
> delays with "[eclipse 7]"?
> 
The reason for the difference is that in the first case, the initial 
domain you assigned the two variables has two values: 1 and 100 
([1,100]), and so local propagation is able to reduce this to the empty 
interval when your two constraints are posted. Did you intend [1,100], 
rather than [1..100] (or actually [2..99], which is what you have in 
your second example):

[eclipse 2]: [X,Y] :: [2..99], X #= Y, X #> Y.
lists.eco  loaded in 0.01 seconds

X = X{2 .. 98}
Y = X{2 .. 98}


Delayed goals:
         -(X{2 .. 98}) + X #> 0
Yes (0.01s cpu)
[eclipse 3]:

> 
> 
> [eclipse 3]: integers([X]), ic:(X #>X).
> 
> No (0.00s cpu)
> [eclipse 4]: integers([X,Y]), X #= Y, X #>Y.
> 
> X = X{-1.0Inf .. 1.0Inf}
> Y = X{-1.0Inf .. 1.0Inf}
> 
> 
> Delayed goals:
>         -(X{-1.0Inf .. 1.0Inf}) + X #> 0
> Yes (0.00s cpu)
> [eclipse 5]: [X,Y] :: [0,100], X #= Y, X #> Y.
> 
> No (0.00s cpu)
> [eclipse 7]: integers([X,Y]), X#>1, Y#>1, X#<100, Y#<100, X#=Y, X#>Y.
> 
> X = X{2 .. 98}
> Y = X{2 .. 98}
> 
> 
> Delayed goals:
>         -(X{2 .. 98}) + X #> 0
> Yes (0.00s cpu)
> 
> 
> 
> To get failure fast, I could use labeling as follows.  Is there a better way?
> 
> testVariables(Vars) :-
>     setval(test, 0),
> 
>     (
>         labeling(Vars),
>         setval(test, 1),
>         fail
>     ;
>         true
>     ),
> 
>     getval(test, 1).

I am not sure I understand what you are trying to do here -- if labeling 
succeeds, it means that there is a consistent assignment of values to 
your variables in Vars, i.e. for your inconsistent examples, a call to 
labeling will fail eventually. It may take a while, as it may need to 
remove inconsistent values from your domain one by one...

Cheers,

Kish