Dear Panagis, Unfortunately, the callback mechanism was designed under the assumption that the non-linear part of the formulation was known and fixed. I think the best approach in the long run is going to be to re-design it so that the non-linear constraints and costs are handled in a generic fashion as well, providing code references to handle the function, Jacobian and Hessian evaluations.
I can't think of an easy way to keep things from getting very complicated and messy otherwise. -- Ray Zimmerman Senior Research Associate 419A Warren Hall, Cornell University, Ithaca, NY 14853 phone: (607) 255-9645 On Jun 17, 2012, at 4:30 AM, Panagis Vovos wrote: > Dear All, > > I am trying to add new variables in the OPF formulation and costs > associated only with them in the objective function. I have read in > the manual that this can be done using callbacks, so that the new OPF > characteristics can be built as an independent MATPOWER module (so > that someone can include it or not in his simulation). New z variables > can be defined, together with their zmin and zmax boundaries. The same > applies for the definition of costs However, it is not clear to me how > can the Hessians and the Jacobians be defined, so that the ability to > use external solvers (other than MIPS) can be preserved. I can alter > the Hessians and the Jacobians manually altering the appropriate code > in MATPOWER, but this generally makes things very complicated and > ruins the ability of building modules using callbacks (like in the > reserves example). Do you have any suggestions, ideas, or workarounds? > > Thank you for your time and help in advance, > > Panagis Vovos >
