项目名称: 张量最优化中的若干理论和算法研究
项目编号: No.11271206
项目类型: 面上项目
立项/批准年度: 2013
项目学科: 数理科学和化学
项目作者: 杨庆之
作者单位: 南开大学
项目金额: 50万元
中文摘要: 张量计算或称数值多重线性代数是应用数学的一个新兴分支,在信息、通讯、高阶统计分析等领域有广泛而重要的应用,十多年前开始得到应用数学界的重视和快速发展。张量特征值是2005年定义的,其已在医疗影像、高阶马尔可夫链、多次型正定性、最佳秩一逼近、弹性力学等领域得到应用。本项目研究张量特征值及相关多项式优化问题的一些理论和算法问题,简称为张量最优化问题,研究内容包括张量特征值的性质,非负不可约张量Perron-Frobenius定理的进一步推广,模最大或最小张量特征值的计算,张量多项式在各种不同约束条件下的最大值或最小值的计算,张量的最佳低秩逼近等。我们还将有针对性地研究其中一些有重要应用背景的具体问题,比如非负张量情形,稀疏张量情形等,以便得到更好的结果。我们将充分利用非线性最优化和数值代数中新近发展出的技巧和结果处理我们研究的张量最优化问题。
中文关键词: 张量优化;张量特征值;非负不可约张量;算法;
英文摘要: The tensor computing or numerical multilinear algebra is a new branch in applied mathematics field, it has wide and important applications in many fields, such as information, communication and higher-order statistic analysis, and it has being payed the great attention and has being developed fast for over ten years. The eigenvalues of the tensor were defined explicitely in 2005 and from then on they have been applied in medical image, higher-order Markov chain, positivity of multi-homogeneous form, best rank-one approximation, elastic mechanics, and so on. This project plans to devote the study of the eigenvalues of tensors and related polynomial optimization, called tensor optimization, including the theory and algorithm aspects.In details,our research focuses on the properties of the tensor eigenvalues, the further generalization of Perron-Frobenius theorem of nonnegative irreducible tensors, the calculating of the eigenvalue of the tensor with largest or least modulus, the calculating of the optimal tensor polynomial under various different constraints, the best lower-rank approximation of a given tensor, and so on. Moreover we will study some particular problems with important applied background, such as those situations with nonnegative entries or separable structure or sparse structure, in order to get be
英文关键词: tensor optimization;tensors' eigenvalue;nonnegative irreducible tensor;algorithm;