Re: can anybody please point me to a good simulated annealing C/C++ code?

On Wed, 25 Jul 2007 07:17:41 +0000, James Dow Allen wrote:

On Jul 23, 10:16 pm, Lionel B <m...@xxxxxxxxxxx> wrote:
spaces. I used to research genetic algorithms and can honestly say
[dons flameproof suit...] that I never encountered a search problem for
which the most lovingly crafted GA could outperform a well-tuned
simulated annealer. But be warned, tuning SA parameters for a
particular problem is (in common with many other "general-purpose"
search algorithms such as GAs) something of a "black art".

I've developed a very simple alternative to SA; principle advantage
being that it largely self-tunes.

Sounds interesting. Do you have a link?

The problem with SA you mention makes
it difficult for me to test whether my method is easier/better than SA.

.... *for a particular problem*, presumably.

It would be nice to have some standard test suite to make a
comparison....

I'm not convinced the idea of "standard test suite" is particularly
meaningful. I guess you could throw some well-known NP-hard problems
(like the TSP) at it. Sure you'll find a zillion refs on Google.

--
Lionel B
.