[SCIP] Absolute value or root of linexpr

[Stefan Vigerske]

Hi,

I'm going to assume that this is about using SCIP via PySCIPOpt, due to 
the mention of "linexpr" and since SCIP does not throw exceptions.

SCIP itself should be able to handle square roots and absolute value 
functions in nonlinear constraints. If PySCIPOpt does not reformulate 
nonlinear objective functions as constraint automatically, then try 
doing this on your side.

Otherwise, a short code snippet to reproduce this problem could also be 
useful. For PySCIPOpt, the issue tracker is at 
https://github.com/scipopt/PySCIPOpt/issues

Stefan

On 09/11/2022 21:26, Berry Vermin wrote:
> Hi all,
> 
> I'm trying to use a combined objective function for a job-shop scheduling problem, where one part is linked to the distribution of the workload. The most logical options are to minimise the standard deviation or mean absolute deviation of the individual workload. However, these would require either a square root or absolute value to be calculated from the product of a binary decision variable (SCIP linexpr) and some parameter which results in the exception "SCIP: a required plugin was not found".
> 
> What would be good options/alternatives here?
> 
> Kind regards,
> Berry
> 
> 
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> https://listserv.zib.de/mailman/listinfo/scip