|
PureBytes Links
Trading Reference Links
|
Duke,
Simulated Annealing (SA) is a form of optimization. Genetic
algorithms (GA) is another. SA, when designed properly, can
converge much faster than GA. The trick is in finding the
best way to represent the "problem" as a N-dim matrix.
When properly designed, a GA approach may take minutes, hours
or days. On the same problem, SA may take less than 1 second.
Regards,
Mark Jurik
----------
From: Duke Jones - Portfolio Manager
Sent: Wednesday, January 09, 2002 2:29 PM
To: omega-list@xxxxxxxxxx
Subject: Adaptive Simulated Annealing
I have some friends who have been experimenting with systems code
incorporating Adaptive Simulated Annealing techniques and have been quite
impressed with the preliminary results. Admittedly, most of the coding
concepts are beyond my current comprehension but I recently received this
link and it goes a long way toward explaining some of the concepts. I
thought I would pass it on the list.
Perhaps those more astute in C will be able to translate some ideas into
easy language for us.
ADAPTIVE SIMULATED ANNEALING (ASA) ?
URL:http://www.ingber.com/ASA-README.pdf
Regards,
Duke
|