Підходи до класифікації методів неперервної нелінійної глобальної оптимізації

Abstract

The great variety of continuous nonlinear global optimization (CNLGO) methods were developed for different optimization problems. It leads to difficulties of their classification. This article contains an attempt to derive overwhelming integral taxonomy of CNLGO methods from existing classifications and formal definition of CNLGO method and problem model. Proposed formal definition of CNLGO method differs from existing in strict distinction between features of optimization algorithm and features of problem model. Considered classifications are divided on two groups according to its basic criterion – morphological and functional. Proposed integral taxonomy doesn’t contradict with present morphological and the most of functional classifications and reflects state-of-the-art in global optimization theory. Potential developers of optimization software may be more confident in projecting of interfaces for classes after reading this paper