From: Antonio Vieiro Varela <antonio@antispam.faraday.usc.es>
Newsgroups: comp.lang.fortran,sci.math
Subject: Re: [req] Jacobi rotation code
Date: Wed, 15 Oct 1997 19:54:18 +0100

On 15 Oct 1997, Hauke Reddmann wrote:

> Hi,
> I've got a problem with an extremely ill-conditioned
> matrix, so that simple Gauss-Jordan elimination
> fails. I've heard that Jacobi rotation is better
> in preserving the eigenvectors. 
> (Or what do you offer? Householder? Gram-Schmidt?)

  SVD? That should work, definitely.

> 
> Of course I could write this myself, but I surely
> don't have to reinvent the wheel? (The matrix is
> complex hermitian, and I only program in Fortran -
> said for the sci.math folks :-)

  "all" scientific FORTRAN code is at www.netlib.no
  (www.netlib.org works also for overseas connections ;^)

  You can also browse the GAMS page at http://gams.nist.gov
  and look for Hermitian Systems

  Cheers,
  Antonio

==============================================================================
From: osman@stargate.cinternet.net (Osman Buyukisik)
Newsgroups: sci.math.num-analysis
Subject: Re: Parabolic PDE solving routines
Date: 21 Dec 97 19:02:56 GMT

In article <673uu0$ka1$1@morgoth.sfu.ca>, Victor Aina wrote:
>
>I need to solve a  parabolic PDE. It has two 
>state variables (x, y) plus time (t):
>
>A*Uxx + B*Uxy + C*Uyy + D*Ux + E*Uy + F*U + Ut = 0
>
>where A, B, C, D, E, F are coefficients, functions of
>x and y.

How about finite element program freefem or kfem? The PDE is
entered in a similar way to the written above. Has minimal graphics.
try :
http://www.asci.fr/~prudhomm/gfem-html/index.html

>Another that I would like to solve has a single state 
>variable plus time.

For this one I suggest the program XTC from:
ftp://ftp.math.pitt.edu/pub/bardware/xtc*

Again nice graphical UI, and both programs are freely available with source code.

One or both may be available for windoze, but both work in various  un*x O/S.

Hope this helped.
Osman