时间：     2018-08-22    09:00-10:00

报告地点： 宝山校区东区机电大楼604(机自学院会议室)

报告人：   Prof.  Jan Heiland

报告人Jan Heiland 教授简介：

Prof. Jan Heiland obtained his Master's degree in applied mathematics from the Technical University Berlin in Germany in 2009. He then joined Bombardier Transportation where he worked on the design and optimization of high-speed trains. In 2010 he returned to TU Berlin where he obtained his PhD in 2014. Already in 2013 he joined the Max Planck Institute Magdeburg (Germany) where he became the team leader of the group "Computer Aided Design of Control Systems". In 2018, Jan Heiland was appointed associate professor at the Otto-von-Guericke University Magdeburg.

Prof. Jan Heiland frequently publishes in high-ranking journals in the fields of control and systems theory, numerical analysis, mathematical modelling, and scientific computing. He also serves the community in reviewing for many prestigious journals and acting as chair and organizer of sessions, workshops and conferences -- most notably the 2016 Sino-German Symposium "Modeling, Model Reduction and Optimization of Flows" held in Shanghai.

报告摘要：

The approach of Proper Orthogonal Decomposition (POD) is a well known and commonly used model reduction method. Basically, POD collects snapshots of the solution trajectory and identifies the most prominent solution patterns. These so-called modes are used as the basis for a reduced order model. If the model is a discretized partial differential equation (PDE), the POD reduced model can be seen as particular optimized Galerkin projection of the PDE.

In our talk we propose a two-fold generalization of the classical POD method. First, we generalize the collection of snapshots towards testing against certain measurment functions. Second, we show that in such a framework POD can also be applied to provide optimized Galerkin time discretizations. We illustrate the basic principles of the new POD approach and show its efficiency in the optimal control of a Burgers equation.