Iterative methods for dense linear systems
Jussi RaholaFinland, Center for Scientific Computing
Thursday September 9, 15.00 p.m. Parallel Algorithms Seminar CERFACS Conference Room
Abstract :
The numerical solution of integral equations gives rise to large dense linear systems. These systems can be solved with either direct methods or iterative methods. Iterative methods can be very efficient for integral equations, because for some problems the number of iterations can be made independent of the number of unknowns. Furthermore, in the iterative solver, the matrix-vector product can be replaced by a subroutine that computes the approximate action of the matrix to a given vector. Examples of fast matrix-vector products are the fast multipole method and methods based on the use of the fast Fourier transforms. These methods do not require the assembly of the full matrix. Thus the computational complexity and memory requirements are greatly reduced. In my talk I will give an overview of the various aspects of the application of iterative methods to dense linear systems.
Cerfacs' Conferences 1999-2001 Home Page
