How to categorize different numerical methods for PDE?
by Yi Zhang
All the numerical methods for PDE can be boiled down two things: the projection from infinite dimension to finite dimension where the numerical solution is seeked, and the discretization of the equaiton itself. Quote from page 1 of Hesthaven and Warburton’s book “Nodal Discontinuous Galerkin Methods“:
The construction of any numerical method for solving a partial differential equation requires one to consider the two choices:
- How does one represent the solution by an approximate solution?
- In which sense will the approximate solution satisfy the partial differential equation?