摘要

Fast multipole methods (FMM) and their immediate predecessors, tree codes, were developed in response to the need for solving N-body problems that occur in applications as varied as biophysics, computational chemistry, astrophysics and electromagnetics. In all these areas, it is necessary to compute long range potentials of the form 1/R between a dense distribution of point charges, where R is the distance between any two charges. Often, repeated evaluation of these potentials is necessary. It is apparent that the cost of direct evaluation, which scales as O(N(2)) for N degrees of freedom, forms a fundamental bottleneck. FMM and tree methods ameliorate the cost associated with these computation; CPU times of these method scale as O(N). It stands to reason that FMM has had a seminal impact on a multitude of fields, so much so, that it was recognized as one of the top ten algorithms of the past century. A method to rapidly compute potentials of the form e(-jkR)/R soon followed. As the reader is aware, these potentials are the crux of integral equation based analysis tools in electromagnetics and the advent of these methods have transformed the face of computational electromagnetics. Consequently, the state of art of integral equation solvers has grown by leaps and bounds over the past decade. This paper attempts to present a detailed review of the state of art of FMM based methods that are used in computational electromagnetics, from the static to the high frequency regime.

  • 出版日期2009-4