Shengfeng Zhu, Professor, School of Mathematical Sciences, East China Normal University

Hehu Xie, Professor

On efficient numerical methods for eigenvalue optimization

10:00-11:00 August 10(Wednesday), Tencent Meeting ID: 882-549-827

We consider numerical solving several eigenfrequency-constrained shape and topology optimization problems. For shape optimization, both distributed and boundary types of Eulerian derivatives are derived with shape calculus. A priori error estimates for finite element discretizations of both shape gradients are shown. The distributed shape gradient has better convergence and is advocated to be used in numerical shape gradient optimization algorithms. Moreover, boundary shape gradient of correction type is introduced. For topology optimization, efficient two-grid and multi-level methods are developed. Numerical results are presented.