local minimzers
Dear all,
recently, I asked help concerning available code for local
unconstrained optimization. The essentials of the replies
I received are the following (including some remarks I added):
----------------------------------------------------------
From: mike - at - mycenae.cchem.berkeley.edu
If you can find a copy of "Numerical Recipes" by Press et al, they
have a few unconstrained minimizers which use first derivative
information. The book is available in Fortran, C, Pascal, and perhaps
other languages. I don't think the codes in the book are available by
ftp (due to copyrights), but they are short, and could be typed into
a computer in under an hour. The authors grant the reader rights to
use the codes, so copyrights aren't a problem if you type it in yourself.
Mike Greenfield
F.Eisenmenger: I can strongly support this as a very good start to
understand minimization
----------------------------------------------------------
From: vkitzing - at - sunny.mpimf-Heidelberg.mpg.de (Eberhard von Kitzing)
%A Press, William H.
%A Teukolsky, Saul A.
%A Vetterling, William T.
%A Flannery, Brian P.
%D 1992
%T Numerical Recipes in C
%I Cambridge University Press
%7 2nd edition
%K numerical algorithms
----------------------------------------------------------
From: kit1 - at - molecular-recognition.chemistry.cambridge.ac.uk
The main site I have used is netlib (or one of its mirrors). Netlib is at
netlib.att.com which I find slow (as its in the USA) so I use
unix.hensa.ac.uk in directory netlib. Several sub-dirs are useful, such
as ./toms ./numeralgo ./opt
If you want more info on other sites then get /pub/inet.services.txt
from ftp.csd.uwm.edu.
Hope this helps
Keith Trollope.
F.E.:
ftp netlib.att.com (IP address 192.20.225.252)
name: anonymous
password: <your email address>
cd netlib
binary
get minpack.tar
----------------------------------------------------------
From cyrus - at - lassp.cornell.edu
For non-linear least-squares optimization, Peter Nightingale and I have
written a program that is an order of magnitude faster than MINPACK
for difficult cases. Furtheremore, it is much less likely to stop
prematurely than MINPACK. We are still improving upon it, but we will
eventually be happy to let others use it.
Cyrus Umrigar
----------------------------------------------------------
Thank you all for your help.
Frank Eisenmenger
Although I did not try this myself, I would like to suggest also:
anonymous ftp to infomeister - at - ccl.net
cd pub/chemistry/software/uncmin
mget *
+------------------------------------------------------------------------+
| Institute of Biochemistry |
| Medical Faculty of the Humboldt University (Charite) |
| |
| Hessische Str. 3-4 |
| 10115 Berlin, Germany |
| |
| Phone: +49-30-28468-612 or -239 FAX: +49-30-28468-600 |
| E-mail: eisenmen - at - comcom.rz.charite.hu-berlin.de
|
+------------------------------------------------------------------------+