During our work, we have developed many codes related to various areas of optimization. Although not all them are ready (and worth) for a distribution, some are available under some different licensing regimes. Typically, we either use a GNU (GPL or LGPL) license, or an "Academic license" permitting a controlled distribution and modification of the codes for academic purposes, but allowing commercial use only via specific agreements (and possibly a fee). Please carefully check the license conditions to verify if you are eligible before downloading or requesting the codes.
These codes are provided "as is", without any explicit or implicit arranty that they will properly behave or they will suit you needs. Although codes reaching the distribution phase have usually been extensively tested, we cannot guarantee that they are absolutely bug-free (who can?). Furthermore, use of the codes is at you own risk: in no case we could be considered liable for any damage or loss you would evenctually suffer, either directly or indirectly, for having used them. More details about the non-warranty attached to these codes are available in the documents describing each licensing regime.
For some of the codes we also distribute test instances; please check our Data page.
The MCFClass project: several C++ solvers for the linear single-commodity Min Cost Flow problem.
The MMCFClass project: some C++ solvers for problems with a prevalent Multicommodity Min Cost Flow structure.
MSArbor: a C++ implementation of the ARBOR algorithm for the Minimal Spanning Arborescence problem on weighted directed (complete) graphs.
BTT: a C++ implementation of codes for solving convex quadratic problems with a special structure of the constraints, arising in Bundle methods for NonDifferentiable optimization and in other contexts.
The NDOSolver / FiOracle project: a suite of C++ solvers for NonDifferentiable Optimization problems whose interface is based on the pair of abstract classes NDOSolver and FiOracle.
SubGrad: a C++ implementation of a SubGradient algorithm for convex NonDifferentiable box-constrained Optimization.
The CQKnPClass Project: C++ implementations of a few different approaches for Continuous (Convex) Quadratic Knapsack Problems.
The VerySimple01Problem class, a simple C++ code solving the simplest problem in 0-1 variables and "efficiently" enumerating its solutions in objective function value order.