Getting infeasible solution back using JuMP with IPOPT and/or KNITRO

[Huckleberry Febbo]

I am solving an optimization problem that I set up using JuMP and solve with both IPOPT and KNITRO. In some cases, the optimization converges to an infeasible point. How do I get the data from this infeasible solution back? In MATLAB, when I use IPOPT for example, I get back result if it is feasible or infeasible and in the infeasible case I can look at the constraints that are violated so that I can understand why the optimization is failing.  Now that I am using julia and JuMP with IPOPT (and now KNITRO as well), the infeasible solution is not passed back, I only get a result if the solution is feasible.

[Miles Lubin]

Yes it is (until we move to discourse, that is). I would encourage you to dig into the JuMP source code to see what's going on here.

On Nov 16, 2016 4:35 PM, "Huckleberry Febbo" <fe...@umich.edu> wrote:

Also, is this the right venue for questions such as this?

On Wednesday, November 16, 2016 at 4:31:20 PM UTC-5, Huckleberry Febbo wrote:

I am solving an optimization problem that I set up using JuMP and solve with both IPOPT and KNITRO. In some cases, the optimization converges to an infeasible point. How do I get the data from this infeasible solution back? In MATLAB, when I use IPOPT for example, I get back result if it is feasible or infeasible and in the infeasible case I can look at the constraints that are violated so that I can understand why the optimization is failing.  Now that I am using julia and JuMP with IPOPT (and now KNITRO as well), the infeasible solution is not passed back, I only get a result if the solution is feasible.

--
You received this message because you are subscribed to the Google Groups "julia-opt" group.
To unsubscribe from this group and stop receiving emails from it, send an email to julia-opt+unsubscribe@googlegroups.com.
Visit this group at https://groups.google.com/group/julia-opt.
For more options, visit https://groups.google.com/d/optout.

[Huckleberry Febbo]

Miles, I figured it out. Thanks!

On Thursday, November 17, 2016 at 12:34:14 AM UTC-5, Miles Lubin wrote:

Yes it is (until we move to discourse, that is). I would encourage you to dig into the JuMP source code to see what's going on here.

On Nov 16, 2016 4:35 PM, "Huckleberry Febbo" <fe...@umich.edu> wrote:

Also, is this the right venue for questions such as this?

On Wednesday, November 16, 2016 at 4:31:20 PM UTC-5, Huckleberry Febbo wrote:

I am solving an optimization problem that I set up using JuMP and solve with both IPOPT and KNITRO. In some cases, the optimization converges to an infeasible point. How do I get the data from this infeasible solution back? In MATLAB, when I use IPOPT for example, I get back result if it is feasible or infeasible and in the infeasible case I can look at the constraints that are violated so that I can understand why the optimization is failing.  Now that I am using julia and JuMP with IPOPT (and now KNITRO as well), the infeasible solution is not passed back, I only get a result if the solution is feasible.

--
You received this message because you are subscribed to the Google Groups "julia-opt" group.

To unsubscribe from this group and stop receiving emails from it, send an email to julia-opt+...@googlegroups.com.