地点：行健楼学术活动室526
邀请人：汪艳秋教授
摘要：In this talk, we discuss two kinds of numerical methods for solving PDEs. One is the least-squares finite element method and the other is a partially least-squares method, the Galerkin Least-Squares FEM. More specifically, we discuss a specific Galerkin Least-Squares method: the augmented mixed FEM. Both LSFEM and the augmented mixed FEM share many good properties: the formulations are automatically stable without the need of an inf-sup stable pair and the LS-functional can be used as a posteriori error estimator. On the other hand, the augmented mixed FEM as a partially LS method, is more flexible and can have properties that the original LSFEM does not have. We will the the parameter-robustness as an example to show that the original LSFEM is non-robust in a priori and a posteriori estimates while the augmented mixed FEM is robust in both estimates.
