Abstract
In this paper we describe an effective and inherently parallel approximate inverse preconditioner based on Frobenius-norm minimization that can be easily combined with the fast multipole method. We show the numerical and parallel scalability of the preconditioner for solving large-scale dense linear systems of equations arising from the discretization of boundary integral equations in electromagnetism. We introduce simple deflating strategies based on low-rank matrix updates that can enhance the robustness of the approximate inverse on tough problems. Finally, we illustrate how to improve the locality of the preconditioner by using nested iterative schemes with different levels of accuracy for the matrix-vector products. Experiments on a set of model problems representative of realistic scattering simulations in industry illustrate the potential of the proposed techniques for solving large-scale applications in electromagnetism.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
