-----Original Message----- From: Zhou, Xingling (Xingling) Sent: Tuesday, September 04, 2001 5:36 PM To: 'raymonda@xxxxxxxxxxxxxxxxxxxxxxx' Subject: RE: [SI-LIST] spectral domain vs BEM (was non-negative......) Looks like you are discussing the mathematical aspects of different algorithms. But I think the original question is about the physical aspects. Should the off-diagonal elements be non-positive or non-negotive ? Fortunately, we can answer it directly based on the definitions, independent of any numerical examples. It does not matter how you calculate them. So, we can judge the mathematical results. It is very true that all algorithms have drawbacks which lead to numerical difficulties and errors. Mick -----Original Message----- From: Ray Anderson [mailto:raymonda@xxxxxxxxxxxxxxxxxxxxxxx] Sent: Tuesday, September 04, 2001 4:39 PM To: si-list@xxxxxxxxxxxxx Subject: [SI-LIST] spectral domain vs BEM (was non-negative......) One of the Apsim guys may want to comment, but I believe that the Apsim RLGC tool that I was using implements the spectral domain method which I think is similar to BEM in some respects but different in others. (I'm going to stop right here, I only know enough about this to be dangerous :) RLGC can optionally use the FEM, but it is recommended that FEM be utilized for more complex structures that are defined with multiple dielectrics than simple stripline constructs. -Ray > > >which brings into another question, >ray, >was the tool using FEM (most likely using ICCG) or >BEM (most likely using LU)? >chris > > >-----Original Message----- >From: Steve Corey [mailto:steve@xxxxxxxxxxxxxx] >Sent: Tuesday, September 04, 2001 12:23 PM >To: si-list@xxxxxxxxxxxxx >Subject: [SI-LIST] Re: non-negative off diagonal capacitive matrix >elements ?? > > > >The most common reason for limited precision when inverting a matrix is >that the condition of the matrix to be inverted was poor. This problem >affects both interative methods, such as conjugate gradient, and direct >methods such as LU decomposition. Most coders, if an iterative method >such as ICCG is convergent, will get it to converge as tightly as >possible, and the limit is imposed by the condition of the matrix. > >A poorly conditioned matrix is one that is close to singular in the >presence of the computer's roundoff error. Poor conditioning can be a >property of the problem to be solved, and/or a property of the algorithm >used to solve the problem. > >Naive as it may sound, I would suspect that in this case that it's due to >the problem being solved. Standard disclaimers apply, however: it's >been quite a few years since I've written any field solver code. > > -- Steve ------------------------------------------------------------------ To unsubscribe from si-list: si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field For help: si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field List archives are viewable at: //www.freelists.org/archives/si-list or at our remote archives: http://groups.yahoo.com/group/si-list/messages Old (prior to June 6, 2001) list archives are viewable at: http://www.qsl.net/wb6tpu ------------------------------------------------------------------ To unsubscribe from si-list: si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field For help: si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field List archives are viewable at: //www.freelists.org/archives/si-list or at our remote archives: http://groups.yahoo.com/group/si-list/messages Old (prior to June 6, 2001) list archives are viewable at: http://www.qsl.net/wb6tpu