Comparing Iterative Methods in iFlow
By Charles Xie ✉
The Jacobi, Gauss–Seidel, and successive over-relaxation (SOR) methods are three iterative techniques for solving a linear system of equations. While the Jacobi method is more suitable for parallel computing, the Gauss-Seidel and SOR methods are generally faster. All three methods are available to use with the Steady State FDM Solver Block provided in iFlow. The default is the Gauss-Seidel method. The following animation shows a comparison of performance of the three methods. As you can see, the SOR method achieves the solution rapidly. The animation shows the process as if the SOR method were about to crash, but it miraculously recovers from the seemingly chaotic trajectory to converge at the correct final state.
Click here
to play with the iFlow simulations