Discontinuous Petrov–Galerkin method based on the optimal test space norm for steady transport problems in one space dimension |
| |
Institution: | 1. Institute of Applied Physics and Computational Mathematics, Beijing 100088, PR China;2. Software Center for High Performance Numerical Simulation, Beijing 100088, PR China;3. Laboratory of Computational Physics, Beijing 100088, PR China |
| |
Abstract: | We revisit the finite element analysis of convection-dominated flow problems within the recently developed Discontinuous Petrov–Galerkin (DPG) variational framework. We demonstrate how test function spaces that guarantee numerical stability can be computed automatically with respect to the optimal test space norm. This makes the DPG method not only stable but also robust, that is, uniformly stable with respect to the Péclet number in the current application. We employ discontinuous piecewise Bernstein polynomials as trial functions and construct a subgrid discretization that accounts for the singular perturbation character of the problem to resolve the corresponding optimal test functions. We also show that a smooth B-spline basis has certain computational advantages in the subgrid discretization. The overall effectiveness of the algorithm is demonstrated on two problems for the linear advection–diffusion equation. |
| |
Keywords: | |
本文献已被 ScienceDirect 等数据库收录! |
|