[SCIP] QCP local Search and RENS

[Timo Berthold]

Hi Ahmed,

there is a paragraph in the paper:
https://opus4.kobv.de/opus4-zib/frontdoor/index/index/docId/4761
and some more explanation in Stefan's PhD thesis:
http://edoc.hu-berlin.de/dissertationen/vigerske-stefan-2012-08-21/PDF/vigerske.pdf

The principle idea of the heuristic can be explained in one sentence:
If you find an MINLP-feasible solution (e.g. by MIP heuristics or from a
local branching subproblem), fix the integers to the value in this
solution, and solve the remaining NLP with global bounds (to local
optimality, though) in hope to find an even better solution.

Best regards,
Timo

> Hi Stephan,
> Thanks for the clarification. I'm glad I got that clarification from you
> as it seems that subNLP   was developed by you as the website indicates.
> Actually after I got your email I started searching for a paper that
> explains the heuristic but wasn't able to find one. Is there a paper that
> I could read for that?
>
> Regards,
> Ahmed
> ________________________________________
> From: Scip [scip-bounces at zib.de] on behalf of Stefan Vigerske
> [stefan at math.hu-berlin.de]
> Sent: Thursday, January 14, 2016 2:20 AM
> To: scip at zib.de
> Subject: Re: [SCIP] QCP local Search and RENS
>
> Hi,
>
> On 01/13/2016 10:01 AM, Jakob Witzig wrote:
>> Hi Ahmed,
>>
>> which QCP local search heuristic do you mean exactly?
>
> I think what is now called the subNLP primal heuristic was denoted as
> "QCP local search heuristic" in some papers.
>
> Stefan
>
>> The heuristic RENS works also for MIQCPs. There are two parameters you
>> have to take care of:
>> 1) rens/startsol: You can decide whether you want to use the LP or NLP
>> solution to fix variables. Per default RENS uses the LP relaxation,
>> which is also fine for MIQCPs. If you want to use the NLP solution you
>> should ensure that an NLP solver is included in SCIP.
>> 2) rens/uselprows: If this parameter is TRUE, RENS uses the LP rows to
>> create the subproblem. Otherwise RENS copies the CIP. In your case this
>> should be FALSE (which is the default).
>>
>> Best
>> Jakob
>>
>> Am 12.01.2016 um 23:04 schrieb Ahmed Ibrahim:
>>> Hi SCIP community,
>>>
>>> I was wondering if the QCP (Quadratic Constrained Program) local search
>>> heuristic is available in SCIP 3.2.0? If so can u tell me where can I
>>> access its relevant files? I can't see it in
>>> 'http://scip.zib.de/doc/html/group__PRIMALHEURISTICS.php'.
>>>
>>> Also, I'm wondering whether the current RENS is extended for MIQCPs?
>>>
>>> Regards,
>>> Ahmed
>>>
>>>
>>> _______________________________________________
>>> Scip mailing list
>>> Scip at zib.de
>>> http://listserv.zib.de/mailman/listinfo/scip
>>>
>>
>>
>
>
> --
> http://www.gams.com/~stefan
> _______________________________________________
> 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
>