Originally posted by: SystemAdmin

I am currently using ILOG CPLEX 12.1 Concert Technology with C++ to solve a mixed-integer quadratically constrained problem (MIQCP). I have noticed that CPLEX is not giving me the optimal solution everytime I solve and I was wondering if this is always the case with MIQCPs? Would it better if I used CPLEX 12.2?

Thank you for your time,
Andrew
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

It could be that CPLEX 12.2, CPLEX 12.2.0.2, or CPLEX 12.3 show improved behavior on your model. But one cannot say this in advance. It could very well be that your model is numerically challenging.

If possible, could you send your model to my email address achterberg at de dot ibm dot com? Then I could take a look and try to find out what is going on.
Tobias
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: EdKlotz

> Tobias Achterberg wrote:
> It could be that CPLEX 12.2, CPLEX 12.2.0.2, or CPLEX 12.3 show improved behavior on your model. But one cannot say this in advance. It could very well be that your model is numerically challenging.
>
> If possible, could you send your model to my email address achterberg at de dot ibm dot com? Then I could take a look and try to find out what is going on.
>
>
> Tobias

This is an old post, but in case ARCPLEX is still subscribed, I had a request for
MIQCP models from Tamas Terlaky of Lehigh to test some new disjunctive conic cuts
he has derived. Thank you for sending the model to us. If you are willing to share the model with Tamas, please send it to him at terlaky at lehigh dot edu.
Thanks.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: EdKlotz

> EdKlotz wrote:
> > Tobias Achterberg wrote:
> > It could be that CPLEX 12.2, CPLEX 12.2.0.2, or CPLEX 12.3 show improved behavior on your model. But one cannot say this in advance. It could very well be that your model is numerically challenging.
> >
> > If possible, could you send your model to my email address achterberg at de dot ibm dot com? Then I could take a look and try to find out what is going on.
> >
> >
> > Tobias
>
> This is an old post, but in case ARCPLEX is still subscribed, I had a request for
> MIQCP models from Tamas Terlaky of Lehigh to test some new disjunctive conic cuts
> he has derived. Thank you for sending the model to us. If you are willing to share the model with Tamas, please send it to him at terlaky at lehigh dot edu.
> Thanks.

One other thing, if you have other models you could share with Tamas, I'm sure he would appreciate it.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

To whom it may concern:

I am trying to find out if there were any significant improvements in handling MIQCP (mixed-integer quadratically-constrained problems) in CPLEX 12.3 over CPLEX 12.2. I figured I would post to one of my previous threads which dealt with a MIQCP I was running in 12.2. I am now working on a new MIQCP, but wanted to get an idea of how well CPLEX was handling these QCPs before implementing with quadratic constraints. (Basically, as a background to why I am posting this inquiry, I would like to utilize an L2-norm, but if no significant improvements are made in 12.3 versus 12.2 in solving quadratically-constrained programs, then I will most likely implement with a linearized version of an L1-norm.)

Any information on whether or not signicant updates to solving MIQCPs have been added to CPLEX 12.3 since version 12.2, it would be greatly appreciated.

Thank you for your time,
Andrew
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

Usually we significantly improve performance for all model types between two releases. Whether your special case can benefit from that is hard to tell without trying.
Since you want to upgrade anyway: any reason to not upgrade to the most recent version 12.5?
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

Thank you so much for the quick reply. I am using a university-based server to run my models, so it would be up to them whether or not to update to a newer version, but I will definitely let them know that there is a newer version available.

I figured it would probably be case-specific (as far as whether any updates would benefit), but I thought it would be useful to check to see if indeed updates had been performed to solve my type of model.

Thank you again,
Andrew
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

> ARcplex wrote:
> Thank you so much for the quick reply. I am using a university-based server to run my models, so it would be up to them whether or not to update to a newer version, but I will definitely let them know that there is a newer version available.
>
Note that version 12.5 no longer requires license files. So it is easier to maintain than previous versions. This may help you to convince them to upgrade :-)
#CPLEXOptimizers
#DecisionOptimization