The cubic algorithm (Q1073547): Difference between revisions

Let \(X\subset R^ n\) be with \(\mu_ m(X)>0\), \(\mu_ m\) being the m- dimensional measure, with \(n\geq m\geq 1\). Let \(S: X\to R\), be a continuous single-valued computable procedure. By considering the problem of finding the value \(s^*=\inf_{x\in X}S(x)\) and the set \(K^*=\{x^*| x^*= \inf_{x\in X}S(x)\}\), \textit{Q. Zheng, B. Jiang} and \textit{S. Zhuang} [A method for finding the global extremum, Acta Math. Appl. Sin. 2, 161-174 (1978)] developed a discrete method to arrive the solution. This method was completed in several works due to Q. Zheng and the author. In this work the author proposes a new algorithm under the hypothesis that S is such that: \(| S(x)-S(x')| \leq L\| x- x'\|,\) \(L=const.>0\), x,x'\(\in X\), that is to say in the hypothesis of a nonlinear nonconvex, but Lipschitzian optimization problem. This algorithm guarantees to find the global minimum \(s^*\) value and the set \(K^*\) of all global minimizers.