[Bonmin] Bonmin-BB as exact method

[Stefan Vigerske]

Hi,

If you cannot be sure that the relaxations are solved to global 
optimality (because they are nonconvex NLPs), then Bonmin may fathom 
nodes based on wrong bounding information.
Thus, it's not an exact method for nonconvex MINLPs.

See also the Section on getting good solutions for nonconvex MINLPs from 
the Bonmin manual:
http://www.coin-or.org/Bonmin/options_set.html#sec:opt%CB%99nonconv
(also in the GAMS/Bonmin solver manual, coin.pdf).

Stefan


On 08/02/2013 09:24 AM, Maximilian Reischl wrote:
> Hi all,
>
> i was wondering if BOMIN-BB can be considered a exact method when setting
> optcr=0. (Mip-Gap = 0).
> I do not mean guaranteed globally optimal (maybe this is the case?)
>   but locally (since i have a nonconvex problem).
>
> Thank you.
>
>
>
> _______________________________________________
> Bonmin mailing list
> Bonmin at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/bonmin
>