Many problems in technical and commercial application areas cannot be formulated adequately as purely linear mixed-integer programs. Several processes and relations inherently lead to non-linear constraints in the mathematical models. If (the relaxations of) the resulting models are convex, then at least methods for convex optimisation can be applied. For the solution of general non-convex, non-linear models, no practically efficient methods are available so far.
For certain technical and financial processes, such as blending or segregation of different materials, however, the arising non-linearities have special structural properties, which can be exploited in the solution techniques. The aim of this Project is to identify and to study such special sub-structures and to develop efficient solution methods for the corresponding non-linear programs.
