A method of unconstrained global optimization

摘要: Abstract The method that is defined in the following finds maximum or minimum of a real-valued function many variables even if has local maxima minima. methods iterative and guaranteed to converge for polynomials several up fourth degree. It can also be used successfully other types functions. approximates automatically it not polynomial degree four less. shown global optimization solve systems equations any speed convergence analysed theoretically empirically. been applied by author collaborators solution nonlinear (up 100 variables), determination rate constants differential (systems identification), chemical equilibrium equations, curve fitting sums exponentials, pattern recognition, analysis spectra with superposition (Bremermann Lam [11]). applications will reported elsewhere.