From: Stephen Kirkup Subject: Re: Boundary Elements Date: Mon, 23 Jul 2001 15:10:55 +0100 Newsgroups: sci.math.num-analysis Summary: Comparison of Finite-element- and Boundary-element-methods Boundary elements are still alive and well actually. They are not as widely applicable as finite elements, but have advantages when they can be used - such as that only the boundary or surface needs to be discretised (very good for "exterior" problems) and the matrices are smaller. Some resources are given on my web site: www.boundary-element-method.co.uk Sbdpumbaa wrote: > Whatever happened to these? Browsing the library I found a book on the Boundary > Element Method, talking of its advantages over finite elements. I'm curious - > if BEM has the advantages claimed (which appear to be simplicity, parsimony, > speed, reduced data needs) why didn't it take over? > > Thanks. > > Jeremy ============================================================================== From: lpa@cutter.rexx.com (Pierre Asselin) Subject: Re: Boundary Elements Date: Mon, 23 Jul 2001 14:55:36 GMT Newsgroups: sci.math.num-analysis Victor Eijkhout writes: >In fact, if I take an FEM matrix and eliminate the interior, I get >a smaller but dense matrix, and it couples the unknowns on the boundary only. >Is there in fact such an equivalence between FEM and BEM? It's been a while since I looked at that... If I remember correctly, your procedure yields a matrix A = M^t M , where M is the matrix the BEM people would have been working with. Their matrices are typically not symmetric, even in self-adjoint problems, but their condition numbers are better. -- Pierre Asselin Westminster, Colorado