Numerical Analysis

Archive for January, 2010

Post-Optimal/Solution Analysis

     I tried this over the holidays with little success.
Hopefully, there will be some one out there now with some
thoughts they are willing to share.

     I am trying to get a feel for what practitioners do with the
output from models.  I’m not so concerned with how they use the
output to tweak/improve the model, but more how/if the output is
used to gain a better understanding of the system modeled.

     If you have any thoughts/suggestions, I would greatly
appreciate hearing from you.

gordon j. shamblin
shamb…@cherokee.nsuok.edu

.
posted by admin in Uncategorized and have No Comments

How? Invert indef. symm. matrix

Hello!

I want to know the fastest algorithm for inversion of an indefinite
symmetric 3×3 matrix. Actually I want ot invert a weight matrix to get
the covariance matrix. I know that the matrix has to be positiv definite
(Cholesky!) but due to numerical inaccuracies this is not always true.
I use a standard Cholesky routine but I get sometimes square root of
something negativ! Or should I better make sure that the matrix is always
positiv definite? How can I do that? Please, e-mail. Many thanks in advance.

– Dieter

———————————————————————–

Dieter Best
GSI/KP1                           There was an old lady of Kent      
Postfach 11 05 52                 Whose nose was most awfully bent.  
46220 Darmstadt                        One day, I suppose            
Germany                                She followed her nose,        
                                  For no one knows which way she went.
VOICE: 0049-06151-359-769
FAX  : 0049-06151-359-785
EMAIL: b…@DDAGSI5.bitnet              
       b…@rzri6c.gsi.de (RS6000/AIX)
       b…@v6000a.gsi.de (VAX/VMS)

———————————————————————–

posted by admin in Uncategorized and have Comments (6)

Looking for 2D/3D Navier-Stokes codes

Hello,

We are looking for any and all freely available CFD codes
solving 2D or 3D Navier-Stokes problems. Finite difference,
or finite elements – Fortran, C, C++. If anything turns up,
I’ll summarize to the net.

One package we do have is fec3dV1.1 from Babak Bagheri at
karazm.math.uh.edu. I’ve not tried it out yet, any comments?

Thanks in advance,

Paul J Turner
ptur…@amb4.ccalmr.ogi.edu

posted by admin in Uncategorized and have No Comments

Spectral Annealing

I have been asked what spectral annealing (sp?) is and I have no clue.
Anyone with any references to books or papers please email or post them.
Thanks
email to stockw…@canlon.physics.uwo.ca
(May be a image processing thing in regards to a boats wake or
something like that.)

posted by admin in Uncategorized and have No Comments

Winograd Fourier transforms

Numeric Recipes (Press’ et al) mentions the highly optimized Winograd Fourier
transorm algorithms (e.g. for N=2,3,4,5,7,8,11,13,16) in their section on
other FFT algorithms. They can run up to a factor of 2 faster than the
nearest power of 2 FFT, so it is claimed.

Does somebody have a reference, or perhaps even some code snippets lying
around?

 peter

       _/_/_/  _/_/_/  _/_/_/  _/_/_/  _/_/_/
      _/  _/  _/        _/    _/      _/  _/
     _/_/_/  _/_/_/    _/    _/_/_/  _/_/_/
    _/     _/         _/    _/      _/    _/
   _/     _/_/_/     _/    _/_/_/  _/     _/   _/

Peter J. Teuben         |  INTERNET  : teu…@astro.umd.edu         ,__0
Astronomy Department    |  FAX:        (301) 314-9067             _-\_<,    /
University of Maryland  |  MA-BELL   : (301) 405-1540 (office)   (*)/’(*)  /
College Park, MD 20742  |              (301) 405-1502 (secr.)   __________/
—————————————————————

       _/_/_/  _/_/_/  _/_/_/  _/_/_/  _/_/_/
      _/  _/  _/        _/    _/      _/  _/
     _/_/_/  _/_/_/    _/    _/_/_/  _/_/_/
    _/     _/         _/    _/      _/    _/
   _/     _/_/_/     _/    _/_/_/  _/     _/   _/

Peter J. Teuben         |  INTERNET  : teu…@astro.umd.edu         ,__0
Astronomy Department    |  FAX:        (301) 314-9067             _-\_<,    /
University of Maryland  |  MA-BELL   : (301) 405-1540 (office)   (*)/’(*)  /
College Park, MD 20742  |              (301) 405-1502 (secr.)   __________/
—————————————————————

posted by admin in Uncategorized and have Comment (1)

lapack and block size; linpack

Has anyone benchmarked the LAPACK routines dgetrf, dgetrs (for LU
decomposition) as functions of block size? Or compared speed against
the equivalent linpack routines (dgesl and dgefa)?

I am mainly interested in single-processor machines (e.g.
workstations), for matrices of order 100 to 400. The LAPACK manual
contains results for CRAYS, but it is difficult to determine whether the
improvement is due to blocking or the number of processors used.      

posted by admin in Uncategorized and have No Comments

QUICK Scheme

Dear Readers,

Is there a ftp site where I could get any info on the QUICK
scheme for 2D?

thanks

Kind Regards

Jerome

__________________________________________________________________
"It is a poor mind indeed which can’t think of at least two ways
to spell any word." – Andrew Jackson, 1833

Jerome Ellepola
Dept. of Mechanical and Production Engineering
National University of Singapore
Singapore 0511
Republic of Singapore

posted by admin in Uncategorized and have No Comments

resolution & deterministic interpolation

I’m looking for information about the "support"
issue when using deterministic interpolation
methods. If output is being represented in image
format, how is any particular resolution?
Is this a non-issue for methods which produce a
smooth surface, or is this data dependent?
Any pointers to relevant literature or keywords
is appreciated.

–darrel…@ids.net

posted by admin in Uncategorized and have Comment (1)

finite difference method (HELP!)

Hi there.

I am looking for a program (or the algorithm, but preferably the former) for
the finite difference(s?) method, in C or Fortran. Can anybody help?
I’ll be most grateful

Aspassia

posted by admin in Uncategorized and have Comment (1)

HELP ! Contour plots.

….. oh! Sorry. Didn’t see the previous postings
about this subject. Mea maxima culpa !
Please ignore the previous posting.

————————————————————————
 Christian Engstler, University of Innsbruck, A-6020 Innsbruck, Austria
 email: engst…@mat1.uibk.ac.at
————————————————————————

posted by admin in Uncategorized and have No Comments