项目名称: 符号数值混合计算中基于问题结构的算法和数值分析研究
项目编号: No.11301066
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 数理科学和化学
项目作者: 李冰玉
作者单位: 东北师范大学
项目金额: 22万元
中文摘要: 本项目紧扣符号数值混合计算在工程中的应用,从问题的结构出发,基于前沿的优化方法和数值计算方法研究近似多项式相关问题。拟研究的内容为: 1)稀疏优化在符号数值混合计算中的应用,着重研究矩阵核范数极小化和L1-范数稀疏优化方法在近似最大公因子问题中的应用; 2)多元多项式系统符号数值混合消元方法的高效实现问题; 3)计算具有给定根重数结构的距离最近的多元多项式系统问题; 4)几类广义的总体最小二乘问题的可计算型条件数理论。
中文关键词: 符号数值混合计算;结构矩阵;稀疏优化;稀疏插值;条件数
英文摘要: This project follows the applications of symbolic-numerical hybrid computations in engineering, studies numerical polynomial-related problems based on structures of problems and by means of advanced optimization methods and numerical computation methods. It consists of the following four parts: 1) the application of sparse optimization theories to solve symbolic-numerical hybrid computation problems, with a special interest in the application of the matrix nuclear norm minimization and the L1-norm sparse optimization to solve the approximate greatest common divisor problems; 2) the valid implementation of the symbolic-numerical hybrid elimination method for multivariate polynomial systems; 3) to ?nd the nearest multivariate polynomial system to a given one which has roots with prescribed multiplicity structure; 4) to establish theories for condition numbers which are suitable for computing for several kinds of generalized total least squares problems.
英文关键词: Symbolic-Numerical Hybrid Computation;Structured Matrix;Sparse Optimization;Sparse Interpolation;Condition Number