Elliptic partial differential equations (PDEs) arise in many areas of computational sciences such as computational fluid dynamics, biophysics, engineering, geophysics and more. They are difficult to solve due to their global nature and sometimes ill-conditioned operators. We review common discretization methods for elliptic PDEs such as the finite difference, finite volume, finite element, and spectral methods and the linear systems they form. We also provide an overview of classic to modern solution methods for the linear systems formed by these discretization methods. These methods include splitting and Krylov methods, direct methods, and hierarchical methods. Finally, we show applications that would benefit from fast and efficient solvers for elliptic PDEs, including projection methods for the incompressible Navier-Stokes equations and the shallow water wave equations with dispersive corrections.
翻译:在计算科学的许多领域,如计算流体动力学、生物物理学、工程学、地球物理等许多领域,都出现了椭圆部分差异方程式(PDEs),由于这些方程式具有全球性,有时是条件差的操作者,因此难以解决。我们审查了椭圆部分方程式的共同离散方法,如有限差异、有限体积、有限元素和光谱方法及其形成的线性系统。我们还概述了由这些离散方法形成的线性系统的经典到现代的解决方案方法。这些方法包括分解和Krylov方法、直接方法和等级方法。最后,我们展示了可受益于快速高效的椭圆方程式解决方案的应用,包括不可压缩的导航-斯托克斯方程式的投影方法,以及具有分解性校正的浅水波方程式。