Learn how easy it is to sync an existing GitHub or Google Code repo to a SourceForge project! See Demo
Quick question and sorry if this is better directed to the COBRA google user group.
Will COBRA2.0 continue to work with ILOG CPLEX from IBM rather than TomLab? I realize that I wont' be able to do QP optimiztion using ILOG CPLEX but at least the license is free.
I believe that the ILOG/CPLEX solver works as it did with previous versions for solving LP.
Basically. I prefer ILOG/CPLEX to tomlab as all tomlab does is provide an interface to the ILOG/CPLEX solvers. I believe tomlab become more prominent solely for historical reasons; I believe a past developer was using it for solving NLPs and liked that the tomlab interface accesses the ILOG/CPLEX solvers in the the fastest method. If you're creating problems with ILOG/CPLEX and don't understand how the solver's structured, it is very very easy to make the creation and population of your problem take much much longer than solving it; however, if you spend the time to know your tool you can speed a number of things up.
If you're interested in updating the ILOG/CPLEX support for the code, I'd gladly look in to merging it with the core. All of the components, except the c13 module which uses an NLP solver would benefit from updated interfaces to ILOG/CPLEX for QP, MILP, MIQP.
For my research, I've developed python code that either uses glpk or the ILOG/CPLEX solvers for solving LP, QP, MILP, MIQP problems; eventually, this will be hosted on the opencobra site.
I would be interested to use IBM's ILOG CPLEX from COBRA, rather than Tomlab CPLEX. I hope it will not requier much effort to adapt the code at places where Tomlab is used.
If you're interested in writing a patch to return support for ILOG/CPLEX to the COBRA Toolbox, it'd be welcome. I don't believe any of the current developers have plans to do so themselves; most of the contributors use gurobi or the tomlab interface.
If you don't mind using Python or devlopment code, ILOG/CPLEX is an option for solving LPs/QPs with the pyCOBRA package and basic MILP/MIQP support should be available within a month or so. Just be ware that we haven't tagged an official release for pyCOBRA so there might be a lag in responses to questions, especially if they reflect an inexperience with python. We're shooting for an official release early next year; however, the code is functional and usable if you've got the savviness.