报告题目：Absolutely and strictly stable filters for time-stepping methods

报告时间： 11月17日（星期五）上午10:00 - 11:00

报告地点：尚贤楼706会议室

报告人：Adrian Hill 教授

主持人：蒋勇教授

数学与统计学院

2017年11月16日

报告摘要：Time-stepping methods generate unwanted parasitic components whose frequencies are known in advance from an analysis of the method. This talk explores how to design a filter whose purpose is to remove such parasitic components, and to transmit the principal component of the method, with high acccuracy. Such a filter should not amplify random signals (stability property). The design of two types of stable filters is investigated: (i) Strictly stable filters, which are suitable to be applied once to linear problems, at the end of the main computation; (ii) Absolutely stable filters, which are suitable for multiple intermittent applications to nonlinear problems throughout the main computation. Both types transmit the principal component with high accuracy. However, absolutely stable filters only do so with a time-delay.

报告人简介：Prof. Adrian T. Hill  1988年取得剑桥大学数学学士学位，1992年取得牛津大学数值数学方向博士学位，1995 获得Leslie Fox Prize in Numerical Analysis；1992年至今在英国巴斯大学任职，期间多次访问奥克兰皇家院士John C. Butcher。主要研究兴趣包括：Filters for weakly stable numerical methods以及结构保持的general linear methods等；在SIAM Journal on Mathematical Analysis，SIAM Journal on Numerical Analysis，IMA Journal of Numerical Analysis，BIT，Applied Numerical Mathematics等Top期刊发表29篇学术论文。