Automatic Switching Algorithms in Hybrid Single-Objective Optimization

Hybrid optimization algorithms consist of a number of proven constituent optimization algorithms and a control algorithm that performs automatic switching among the constituent algorithms at each stage during the optimization when the rate of convergence becomes unsatisfactory, the process tends towards a local minimum, or some other undesirable aspect of the iterative process appears. Thus, hybrid optimization algorithms that utilize a number of gradient based and non-gradient based constituent optimizers are more robust and converge better than individual constituent optimization algorithms. The logic of designing the automatic switching algorithms in hybrid optimizers is surveyed in this paper focusing on the research performed by the authors in the area of hybrid single-objective optimization initiated in 1997.