Large Matrix Solution Techniques Applied to an Electromagnetic Scattering Problem
Abstract
There are many numerical applications where one must solve a large system of linear equations. In order to do this, two issues must be addressed: Storage of the large matrices and finding an efficient technique to solve this large system of equations. This report looks at these issues as they apply to a given electromagnetic scattering problem. Surface integral equations are solved to find the surface currents and the far-field scattering from a three- dimensional conducting cube. This three-dimensional scattering problem is solved by magnetic-field integral equations (MFIE), augmented magnetic-field integral equations (AMFIE), or, the recently developed, dual-surface magnetic-field integral equations. Various methods are reported for dealing with the shortage of memory - unified extended memory (UEM), virtual memory, mass storage, and direct access fields. These methods can be used to increase the size of available matrix memory, without vastly increasing the CPU time beyond the time that would be required with central memory alone. Three techniques of matrix solution are discussed: Gaussian elimination, the conjugate gradient, and the biconjugate gradient methods. The computations reported here were done on three representative scientific computers: A Cyber 860 with NOS 2.5 operating system, a VAX 8650 and a Vax 11/780. Both Vaxes have Vax/VMS version 4.7 operating system. Keywords: Magnetic-field integral equation, Augmented magnetic-field integral equation, Dual-surface magnetic-field integral equation.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 1988
- Accession Number
- ADA206917
Entities
People
- Margaret B. Woodworth