Coupled chaotic simulated annealing processes

In this paper we formulate methods of chaotic simulated annealing within the context of coupled local minimizers. Interpreted within the cellular nonlinear networks context, coupled local minimizers consider local optimization algorithms as cells with local connections between the cells. As a result, information exchange is taking place between the minimizers. Instead of taking local optimization methods as individual cells we explore here the use of chaotic signals as additional driving force as in continuous simulated annealing where deterministic chaos now plays the role of noise. On a number of examples, improved results are obtained by coupled chaotic annealing. In general, the coupling of the minimizers also leads to a variance reduction on the optimal cost function values simulated for many different runs.