[SCIP] FW: Most Infeasible Branching for MIQCP

[Stefan Vigerske]

Hi,

No.
Some branching rules have only implementations for branching candidates 
that are integer variables with fractional solutions in the LP relaxation.
In other rules, the implementations are just very different, e.g., 
regarding how to apply pseudo-costs.
Only for rules like mostinfeasible, leastinfeasible, random, and 
inference, the algorithms for branching on integer variables with 
fractional solutions in the LP relaxation or variables occurring in 
nonconvex terms of violated nonlinear constraints are quite similar.

Looking into the code might give some idea what is exactly happening.

Stefan

On 04/22/2016 07:22 PM, Ahmed Ibrahim wrote:
>
> ________________________________________
> From: Ahmed Ibrahim
> Sent: Friday, April 22, 2016 11:32 AM
> To: Timo Berthold
> Subject: RE: [SCIP] Most Infeasible Branching for MIQCP
>
> So, for each branching rule in SCIP (like most infeasible, reliability , pscost, cloud inference, etc.) is it exactly the same procedures for each rule that is used when branching on continuous variables as on integer variables for an MIQCP?
> ________________________________________
> From: Timo Berthold [berthold at zib.de]
> Sent: Friday, April 22, 2016 2:51 AM
> To: Ahmed Ibrahim
> Cc: scip at zib.de
> Subject: Re: [SCIP] Most Infeasible Branching for MIQCP
>
> Dear Ahmed,
>
>> So does the most
>> infeasible branching in SCIP consider only integrality constraints
> Yes.
>> MIQCP or does it consider quadratic constraints too?
> No.
>
> Per default, branching on violated nonconvexities will only happen when
> all integer variables take integral values in the current LP solution.
> Thus, as long as most infeasible (or other LP-based  branching rules like
> relpscost) finds a violated integer to branch on, it will do so. If,
> however, the LP happens to be integral, then most infeasible will branch
> on the most violated nonconvexity.
>
> Hope that helps.
>
> Best regards,
> Timo
>
>
>> Hi All,
>>
>> According to my my understanding, most infeasible branching is an LP
>> branching technique and selects the integer variable whose fractional
>> component is closest to 0.5. Is it the same for an MIQCP? The reason I'm
>> asking this is that in MIQCP linear outer approximation is used for
>> quadratic constraints. Hence some quadratic constraints could also be
>> violated besides integrality constraints can be violated. So does the most
>> infeasible branching in SCIP consider only integrality constraints in
>> MIQCP or does it consider quadratic constraints too?
>>
>> Regards,
>> Ahmed
>> _______________________________________________
>> Scip mailing list
>> Scip at zib.de
>> http://listserv.zib.de/mailman/listinfo/scip
>>
>
>
>
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> http://listserv.zib.de/mailman/listinfo/scip
>


-- 
http://www.gams.com/~stefan