[SCIP] Quadratic Programming Solver Evaluation for SCIP

[Ambros Gleixner]

Hi Lei,

It is correct that SCIP does not support nonlinear objective function
directly, but it can always be modeled using one or more auxiliary
variables in order to shift the nonlinear part of the objective into the
constraints:

   min f(x) ... -> min z st. f(x) <= z ...

Internally, SCIP recognizes and exploits this special structure during
the solving process, so SCIP should in general be perfectly fit for your
use case.

Best regards,
Ambros


Am 06.03.2017 um 19:45 schrieb Wang, Lei:
> Hi Felipe and Gleixner,
> 
>  
> 
> This is Lei Wang from SAP. Our team is in the process of evaluating
> different Quadratic Programming Solvers for supporting our use case. In
> our use case, the objective function has the standard quadratic format.
> The constraints include linear inequalities and linear equalities. It
> seems that a quadratic objective function is not supported by SCIP.
> Could you help confirm it or provide more insights on the topic? Any
> help will be appreciated.
> 
>  
> 
> Thanks,
> 
> Lei
> 
>  
> 
> *Lei Wang, PhD***
> 
> Director | Product Management, Supply Chain | Connected Supply Chain
> Management  |  Products and Innovation
> 
> *SAP Labs, LLC *| 6^th Avenue | Pittsburgh, PA | 15222
> 
> T: +1 412-454-4157 | M: +1 412-699-7180 | E: lei.wang05 at sap.com
> <mailto:lei.wang05 at sap.com>
> 
> http://www.sap.com <http://www.sap.com/>
> 
>  
> 
>  
> 

-- 
Ambros Gleixner, Zuse Institute Berlin, http://www.zib.de/gleixner