[Scip] SCIP 3.1 is slower than 3.0?

[Alessia Violin]

Hello Gerald,

thanks a lot for your help. Indeed this solves the problem, my 
propagation method is fundamental for having the correct branching 
decisions in the tree :-)

I am sorry if you spent a lot of time for this, as I had some errors for 
other methods in which new parameters where added I thought it was ok 
once I solved them, and didn't check the warnings. For sure I will 
consider them more carefully from now on!

Best,

Alessia


On 03/29/2014 05:53 PM, Gerald Gamrath wrote:
> Hi Alessia,
>
> I finally found some time for having a deep look at your code and 
> found the reason for your code being slower with SCIP 3.1 after some 
> debugging.
>
> The CONSPROP-callback got a new parameter nmarkedconss, which you did 
> not add to the propagation method of your NPP constraint handler (both 
> in cpp and h). Without that, your propagation method was never called, 
> but instead the default propagation method of the objective constraint 
> handler, which does nothing.
>
> Seems like your propagation method is quite important for the 
> performance of your code. :-) After adjusting the parameter list of 
> scip_prop in your constraint handler, it showed the same behavior as 
> with SCIP 3.0.1.
>
> BTW: Yet another reason for always removing all compiler warnings. ;-)
>
> Best,
> Gerald
>
>
>
>
>

-- 
Alessia Violin
Service Graphes et Optimisation Mathématique (G.O.M.)
Université Libre de Bruxelles
C.P. 210/01
Boulevard du Triomphe
B-1050 BRUXELLES
Tel: 02 650 58 80 - Fax: 02 650 59 70
Email: aviolin at ulb.ac.be
Webpage: http://homepages.ulb.ac.be/~aviolin/