主题：Structure exploiting algorithms for computational science and engineering

主讲人：德国马克斯普朗克复杂动力系统研究所 邱越博士

主持人：经济数学学院顾先明博士

时间：2019年4月11日（星期四）下午15:00-16:30

地点：西南财经大学柳林校区通博楼B412会议室

主办单位：经济数学学院科研处

报告人简介：

邱越，2015年12月于荷兰代尔夫特理工大学获得计算数学博士学位。2015年12月至今在德国马克斯普朗克复杂动力系统研究所从事博士后研究。研究方向为数值代数，不确定性量化，数据同化，模型降阶，能源网络以及数学与控制、信息学科交叉研究领域。研究结果主要发表在SIAM/ASA J. Uncertain. Quantif., Electron. Trans. Numer. Anal., Comput. Geosci.等专业期刊上。

主要内容：

For most problems in computational science and engineering (CSE), the biggest challenge for the numerical computations is the computational complexity due to the fact that the problems to be resolved are often large-scale. In the current talk, I will introduce how to exploit the hidden structure of the underlying problems to design fast numerical algorithms to reduce the computational complexity. The building block is the low-rank approximation, which is widely applied to beat the curse of dimensionality. I will show how to perform structured matrix and tensor computations to solve large-scale problems in PDE-constrained optimization, optimal flow control, uncertainty quantification, and energy networks.