[Couenne] solution output / heuristics

[Pietro Belotti]

Hi Ina,

there is no paper for the simple heuristic, while one for the feasibility 
pump should be out in a couple of months.

Regards,
Pietro

--
Pietro Belotti
Dept. of Mathematical Sciences
Clemson University
email: pbelott at clemson.edu
phone: 864-656-6765
web:   http://myweb.clemson.edu/~pbelott

On Tue, 2 Apr 2013, Liffey1986 at gmx.de wrote:

> Hi Pietro,
>  
> thanks for your help. Do you know if there exists a good paper with explanations how these 2
> heuristics work?
>  
> Best, Ina
>  
>   Gesendet: Freitag, 29. März 2013 um 04:42 Uhr Von: "Pietro Belotti" <pbelott at clemson.edu>
> An: Liffey1986 at gmx.de Cc: "Couenne mailing list" <couenne at list.coin-or.org> Betreff: Re: [Couenne]
> solution output / heuristics
> Hi Ina, > 1) Does this mean that Couenne found a new AND better solution to the > incumbent, or just a
> new solution that might not be better? It means that Couenne found a new and better solution. > 2) I
> solved some instances and the only heuristics that found solutions > are 'Init Rounding NLP' and
> 'Couenne Rounding NLP'. Does Couenne only > use these 2 heuristics? Yes, and actually they are the
> same heuristic, the "Init" version being something that is applied at the very beginning. Of course a
> solution can also be found after solving the LP relaxation (whose solution sometimes happens to be
> integer feasible), strong branching, etc. A feasibility pump heuristic is available in trunk and
> stable/0.4, but it is quite time consuming. You can enable it with the option feas_pump_heuristic yes
> in the couenne.opt option file. Regards, Pietro -- Pietro Belotti Dept. of Mathematical Sciences
> Clemson University email: pbelott at clemson.edu phone: 864-656-6765 web:
> http://myweb.clemson.edu/~pbelott On Tue, 26 Mar 2013, Liffey1986 at gmx.de wrote: > Hi all, >   > in the
> output Couenne gives information whenever a solution is found > (Integer solution of xxx found by
> xxx). > I have two questions about this: >   > 1) Does this mean that Couenne found a new AND better
> solution to the > incumbent, or just a new solution that might not be better? >   > 2) I solved some
> instances and the only heuristics that found solutions > are 'Init Rounding NLP' and 'Couenne Rounding
> NLP'. Does Couenne only > use these 2 heuristics? >   > Best, Ina