home *** CD-ROM | disk | FTP | other *** search
- Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
- Path: sparky!uunet!paladin.american.edu!auvm!IBM.GWDG.DE!APYW
- Message-ID: <GDDM-L%93012605283728@POLYVM.POLY.EDU>
- Newsgroups: bit.listserv.gddm-l
- Date: Tue, 26 Jan 1993 10:30:19 MET
- Sender: The GDDM Discussion list <GDDM-L@POLYVM.BITNET>
- From: APYW%IBM.RZ.TU-CLAUSTHAL.DE@IBM.GWDG.DE
- Subject: ill-posed general linear equation system, inverse problem
- Lines: 37
-
- Dear Madame and Sir, Hello netters!
-
- I have to solve a ill-posed GENERAL linear equation system
- out of an inverse analysis with Boundary Element Method (BEM).
- The order of the matrix varies from 100 to 200. The condition
- number ranges from 0.10000000D+20 to 0.10000000D+30. All
- computations must take place in quadruple precision on
- IBM3090 machine and the solutions are (numerical) exact
- enough. Computations in double precision have no results.
-
- I have tried different GAUSS-methods (partial or complete pivoting,
- modified method according to residuals, LINPACK, NAPACK, LAPACK) and
- used the driver DGESVX of LAPACK version 1.0b to equalibrate the
- matrix. Newly we have used HOUSEHOLDER-TRANSFORMATION too. All these
- help hardly, because we would like to solve the system with double
- precision, not with quadruple precision, in order to transfer our
- theory into praxis.
-
- It is difficult for an engineer to start with regularization and/or
- singular value decomposition (SVD) of an ill-conditioned system.
-
- Therefore I wish to find PROGRAMS for the ill-posed GENERAL linear
- equation system. I hope to get some tips, algorithms, programs or
- related informations from You Experts on the list.
-
- Any further helps are very appreciated. Thanks a lot in advance.
-
- Yours sincerely
- -- Yaomin Wang
-
- Institute for Equipment and Plant Design
- Technical University Clausthal (TUC)
- C/O Dipl.-Ing. Yaomin Wang
- Zellbach 5
- D-3392 Clausthal-Zellerfeld
- Tel: +49 05323 722576
- Email: apyw@ibm.rz.tu-clausthal.de
-