南京理工大学学报（自然科学版）2018，Vol.42Issue(2)：189-194,6.DOI:10.14177/j.cnki.32-1397n.2018.42.02.009
基于近似Hessian矩阵的修正网格自适应直接搜索算法
Modified mesh adaptive direct search algorithm with approximate Hessian matrix
摘要
Abstract
In order to solve the problem that the mesh adaptive direct search algorithm is easily failing into slow convergence and low efficiency when optimizing,the modified mesh adaptive direct search algorithm with the second directional derivative-based Hessian matrix is presented. The orthogonal triangular decomposition is applied to the presented algorithm. In search step,using the Taylor series expansion,the rank-one update and the linear regression,the quadratic model of the function with linear constrains is built by means of a second directional derivative-based Hessian matrix. By solving the subprogram of the problem,the local solution is obtained. The poll step searchs the best point according to the new poll directions in the neighbor of the trial points,and the update of the poll directions and some parameters is altered in this method. The modified algorithm outperforms the original algorithm in terms of iteration times on some test problems.关键词
约束优化/修正网格自适应直接搜索算法/近似Hessian矩阵/二次模型函数/正交三角分解Key words
constrained optimization/modified mesh adaptive direct search algorithm/approximate Hessian matrix/quadratic model funotions/orthogonal triangular decomposition分类
数理科学引用本文复制引用
刘梅,刘红卫,杨善学,刘泽显,卢晓宁..基于近似Hessian矩阵的修正网格自适应直接搜索算法[J].南京理工大学学报（自然科学版）,2018,42(2):189-194,6.基金项目
国家自然科学基金(11461021) （11461021）
广西高校科研项目(2013YB236) （2013YB236）
