| Summary: | 碩士 === 國立高雄大學 === 應用數學系碩士班 === 98 === Preconditioner plays an important role in solving linear system Ax=b. A suitable preconditioner makes the system converge quickly and more stable. In thisthesis, we’ll discuss the effects of different preconditioners in the linear system which comes from an eigenvalue problem derived from the governing Maxwell equation. The purpose is to find a suitable preconditioner to accelerate convergent rate of the eigenvalue problem. We conduct our experiment by combining Arnoldi’s method, Jacobi-Davidson’s method and Krylov-Schur method with preconditioners like Jacobi, SSOR, FFT, etc. Finally, we find that the Krylov-Schur method with FFT preconditioner is more effective than other common preconditioners.
|
|---|
