Search-space smoothing for combinatorial optimization problems

Commonly there are two types of local search approaches known to treat combinatorial optimization problems with very complex search-space structure: One is to introduce very complicated types of local move classes, allowing a bypass of high energetic barriers separating different minima. The second is introducing a control-parameter (i.e. temperature in physics terminology) dependent state space walker, which is — depending on this control parameter — more or less easily able to climb over barriers. A third, less well-known, but very obvious approach is to smooth the search space, i.e. to eliminate barriers between low-energy configurations and therefore to allow a fast and easy approach to the global optimum. This procedure will be discussed in depth in the following work.