摘要
研究一维全局最优化问题的确定性求解方法。运用逐次建立目标函数的线性下界函数，将不含全局最优解的子区域删除，并基于非精确搜索结合下降算法而得出非精确搜索一维全局最优化方法，使计算量减少且使迭代收敛加快。迭代结束时该算法得到一维全局最优化问题的ε－全局最优解。该方法具有有限收敛性且不需精确的局部优化过程。文中的数值实例表明该算法的有效性。
The deterministic approach for global one dimensional optimization is studied in this paper. By means of constructing lower linear bounding functions for the objective function the sub regions, which do not contain the global solutions in the search domain are deleted progressively. Through incorporating the inexact search into the search domain contraction operation a global one dimensional optimization algorithen using Linear Bounding Functions(LBF s) and inexact search is formed. At the end of the iteration process an ε- global solution is reached. The proposed algorithm is finite convergent and independnt of exact local search. Numerical experience demonstrates that the proposed algorithm is efficient and of potential.
关键词
全局最优化
线性界限函数
非精确搜索
global optimization
linear bounding functions
inexact search
