[SCIP] Different Problems while using SCIP with YALMIP

[Stefan Vigerske]

Hi,

(1) This looks like a bug in the SCIP interface, which would be the OPTI 
toolbox. Please check with them.

(2) Sometimes, especially for nonlinear programs, it can happen that 
SCIP found a solution that is feasible for the internal (reformulated) 
problem, but then turns out to slightly violate constraints for the 
original problem. If you think this violation is too large, then 
decreasing the feasibility tolerance (numerics/feastol) should help.

(3) One would need to look at the SCIP log, especially the "SCIP status" 
line. It could be that SCIP finished with a status that OPTI or Yalmip 
does not understand, so it returns it as "unknown problem".

Stefan

On 10/31/2017 01:50 PM, Valerie Schrayvogel wrote:
> Hello,
> 
>   
> 
> I have a couple of questions regarding using SCIP (nonlinear) under YALMIP
> (interfaced for Matlab through the OPTI Toolbox).
> 
>   
> 
> (1) I get the message "Original variable <nlobj> not released when freeing
> SCIP. Consider releasing variable first.": I could not find out how this
> works under YALMIP.
> 
> As far as I understood, this is not necessary but quite useful to use the
> Memory-Management. Is there a way to handle this using YALMIP?
> 
>   
> 
> (2) Furthermore I am a litte irritated by the following message:
> 
>   
> 
> SCIP Status        : problem is solved [optimal solution found]
> 
> Solving Time (sec) : 0.22
> 
> Solving Nodes      : 1
> 
> Primal Bound       : +4.79653886000805e+03 (1 solutions)
> 
> Dual Bound         : +4.79653886000805e+03
> 
> Gap                : 0.00
> 
>    [linear] <lincon45>: <xvar33>[C] -<xvar37>[C] <= 1e-05;
> 
> violation: right hand side is violated by 1.50505108851939e-05
> 
> best solution is not feasible in original problem
> 
>   
> 
> Does it mean that the solution is not feasable (yalmip reports the problem
> 0, which means feasable solution found)?
> 
>   
> 
> (3) I calculate the same type of NLP (accurate: a heat exchanger network) a
> couple of times with different parameters (Temperatures at the in- an
> outlet). In most cases yalmip reports the problem "0" but in some cases it
> returns 9 (which is defined as "Unknown problem in solver" in YALMIP).
> 
> Do you have any idea what could be the problem here?
> 
>   
> 
> Thank you in advance!
> 
>   
> 
> Best,
> 
>   
> 
> Valerie
> 
> 
> 
> 
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> https://listserv.zib.de/mailman/listinfo/scip
>