[Scip] Upper triangular or symmetric matrix for quadratic constraints

[Ambros Gleixner]

Dear Ramon,

how do you formulate your problem?  Are you using the callable library 
or a format like Zimpl, MPS, LP, ...?

Internally, SCIP does not make a difference between the two cases, 
because it does not store the matrix, but directly the product xy.

Kind regards,

ambros


Am 20.02.2014 10:44, schrieb Ramón Casero Cañas:
> Dear all,
>
> I have a question about how to format quadratic constraints in SCIP.
>
> Say we have [x y] as variables, and we want the constraint 0 <= xy <= 5.
>
> Is it better to aim for a symmetric matrix
>
> 0 <= [x y] [0.0  0.5] [x] <=5
>                 [0.5  0.0] [y]
>
> or for an upper triangular
>
> 0 <= [x y] [0.0  1.0] [x] <=5
>                 [0.0  0.0] [y]
>
> or it doesn't matter? Theoretically they both produce the same
> constraint, but maybe there are some numerical issues or  implementation
> details in SCIP by which one should be preferred.
>
> Many thanks,
>
> Ramon.
>
>
>
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> http://listserv.zib.de/mailman/listinfo/scip
>

-- 
____________________________________________________________
Ambros M. Gleixner
Zuse Institute Berlin - Matheon - Berlin Mathematical School
http://www.zib.de/gleixner