Different Aspects of Numerical Schemes — Lesson 3

This lesson covers the continuation of the discussion on the discretization of partial differential equations and different aspects of numerical schemes. It delves into the finite difference methods and their application to the discretization of the Laplace equation. The lesson also explores the parabolic and hyperbolic partial differential equations, using the Stokes first problem and the linear wave equation as model equations. It further discusses the concept of upwind differencing, the explicit and implicit formulation of numerical schemes, and the accuracy and stability of these schemes. The lesson concludes with an examination of dissipative and dispersive errors in numerical schemes.

Video Highlights

01:02 - Introduction to the parabolic problem, the Stokes first problem, and its physical representation.
10:45 - Explanation of the concept of upwind differencing in the context of wave propagation or advection problems.
15:28 - Comparison of explicit and implicit formulations for the Laplace equation, Stokes first problem, and linear wave equation.
28:14 - Demonstration of the effects of dissipative and dispersive errors through numerical solution results for the first order upwind scheme and Lax-Wendroff scheme.

Key Takeaways

- Finite difference methods are used in the discretization of partial differential equations, including the Laplace equation.
- The Stokes first problem and the linear wave equation serve as model equations for studying parabolic and hyperbolic partial differential equations respectively.
- Upwind differencing is crucial in problems of wave propagation or advection, as it takes into account the direction from which the signal is propagating.
- Numerical schemes can be explicitly or implicitly formulated, with each having its own accuracy and stability characteristics.
- Dissipative and dispersive errors are two types of errors that can occur in numerical schemes. Dissipative errors result in the damping of sharp gradients, while dispersive errors lead to oscillations or "wiggles" in the solution.