30+ Years of Fast Multipole Methods – From Math To Code To Applications
The Fast Multipole Method combines optimal complexity O(N) and adjustable accuracy in one fast summation technique for a wide range of applications in modern numerical science. As one of the Top10 algorithms of the last century and a demonstrator for future Exascale systems it bridges theory and simulation for more than 30 years. Applications range from energy computations in N-body systems, governed by Coulomb or gravitational interactions, electromagnetic scattering, boundary element methods for Stokes, acoustic wave and elastostatic problems to preconditioners for dense linear algebra. As a mesh-free method the FMM shows significant advantages over mesh-based methods, especially for applications with spatial non-homogeneously distributed input. Flexibility of FMMs has been increased by the advent of kernel-independent implementations and a priori error control schemes.
Ivo Kabadshow ( Research Centre Jülich ) - Organiser
Godehard Sutmann ( Forschungszentrum Juelich ) - Organiser