[S] Simulated Annealing and Amoeba Summary

Humbolt, Allen (HumboltA@kochind.com)
Thu, 14 May 1998 09:17:07 -0500

Last week I asked about the availability of any new Splus functions for
simulated annealing and amoeba, functions presented in the book Numerical
Recipes in C. I also asked about hints for the Splus-C interface where a C
program is involved, and for any specific information about interfacing with
Numerical Recipes routines.

I didn't hear of any new algorithms or functions; but Longhow Lam of
CANdiensten did point me to an interesting and simple example at
http://www.candiensten.nl/splus/Training/Download_tekst.sht#Ccode , an
example that I had not seen before. In this example he repeats the
Numerical Recipes C programs for the simulated annealing solution of the
traveling salesman problem and shows how the DLL for this C program can be
called from Splus (we're dealing just with Version 4 here, I believe). The
way the details were presented for this relatively straightforward problem
was appealing to me; and the example provides a nice template for other
applications of linking with C DLL's as well. Not only that, his example
worked nicely on my PC with little effort on my part!

Others reminded of information I've seen in past postings, so I'll breifly
repeat that here.
1. Brian Dawkins points to his "recipes" library that is available on
statlib for handling Numerical Recipes programs.
2. Daniel Heitjan pointed me to an anonymous ftp to, cd to
/pub/dheitjan/power and a file called optfcn.q, which includes amoeba and
several other translations of Numerical Recipes algorithms into pure Splus
functions. His fun.amoeba function, for example, contains a pure Splus
version of amoeba.
3. No one mentioned this one; but there is a Fortran program on statlib
called simann by Bill Goffe for simulated annealing. I haven't tried his
program; but one of the strengths of it is that there is a lot of nice
documention explaining what the program is doing.

I'll also summarize a few related comments.
1. Matt Calder has used several Numerical Recipes functions from Splus.
Via Windows he uses them in a DLL.
2. Olga Ryndina asks for the replies I get.
3. Bill Venables expressed an interest in some newer and better simulated
annealing functions, and suggested that he would like to try simulated
annealing for automatic construction of experimental designs.

Thanks for the responses.

Allen Humbolt
Quantitative Analyst
Koch Industries, Inc.

This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news