[SCIP] Problems reading NL files from Pyomo with Scipampl

[Stefan Vigerske]

Hi,

I couldn't reproduce this on Linux or macOS with that .nl file and 
valgrind didn't signal anything.
You may have to debug a bit where this comes from 
(https://www.scipopt.org/doc-7.0.3/html/FAQ.php#howtodebug).

There is also a scipampl executable of SCIP 7.0.1 for macOS available at
https://www.scipopt.org/download.php?fname=scipampl-7.0.1.darwin.x86_64.gnu.opt.spx2.zip

Stefan

On 8/17/21 11:27 AM, Edwin Reynolds wrote:
> Hi all,
> 
> I have installed SCIP 7.0.2 and the AMPL interface from source on OSX, and
> I am using it with Pyomo in Python. The interface is via .NL files. This
> all works perfectly when my model is linear.
> 
> However, when I try to solve a non-linear problem, Scipampl cannot read it.
> 
> I have reproduced the issue with a very small quadratic problem - Python
> code to generate the NL, and the NL itself are both attached.
> 
> When I use Scipampl from the command line on this NL file, I get a
> segmentation fault. When I change the model to be linear, it works fine.
> 
> Do you know where I am going wrong? Any help much appreciated!
> 
> Thanks,
> Edwin
> 
> 
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> https://listserv.zib.de/mailman/listinfo/scip
>