摘要
对于Hessian矩阵正定的情形,在求解二次函数模型信赖域子问题的隐式分段折线算法的基础上,提出一种求解信赖域子问题的改进的隐式Euler切线法,并分析该路径的性质.数值实验表明新算法是有效可行的,且较原算法具有迭代次数少、计算时间短等优点.
Based on the implicit piecewise dogleg algorithm for solving trust region subproblems of quadratic models,and aimed at the condition that the Hessian matrix was positive definite,an improved implicit Eulerian tangent algorithm for solving trust region subproblems was presented.Then the nature of the path was analyzed.The numerical experiments show that the new algorithm is effective and practicable,and the advantages of the new algorithm compared with the previous algorithms lie in the smaller number of iterations,the shorter computing time and so on.
出处
《应用数学和力学》
CSCD
北大核心
2017年第3期347-354,共8页
Applied Mathematics and Mechanics
基金
山西省自然科学基金(2008011013)
山西省"131"领军人才工程项目
关键词
隐式Euler切线法
信赖域子问题
微分方程模型
无约束优化
信赖域方法
implicit Eulerian tangent algorithm
trust region subproblem
differential equation model
unconstrained optimization
trust region method