Optimizing with Many Parameters
Posted: Sun Nov 23, 2003 10:36 am
MODERATOR'S NOTE: Moved from viewtopic.php?t=664 to new topic since this wasn't really about "Synthetic Indicator's".
I am naive enough to have thought that I would simply set step values for all of the indexable fields and let 'er rip. I'd come back in a few hours or days and see the world's largest simultaneous linear equation would have completed who knows how many tests and -- voila -- there is the best of the best.
I see it doesn't quite work that way and now I realize I don't really know how to test with a system in which there are a lot of steppable variables. Do you pick one, vary it over a range, write down the best value?
Set that variable to the best value and pick another one, vary it over a range, and write it down.
Then set the second variable to the best value and so forth?
I would appreciate thoughts from the persons on this forum, who are obviously quite experienced and knowledgeable.
Thanks!
I am naive enough to have thought that I would simply set step values for all of the indexable fields and let 'er rip. I'd come back in a few hours or days and see the world's largest simultaneous linear equation would have completed who knows how many tests and -- voila -- there is the best of the best.
I see it doesn't quite work that way and now I realize I don't really know how to test with a system in which there are a lot of steppable variables. Do you pick one, vary it over a range, write down the best value?
Set that variable to the best value and pick another one, vary it over a range, and write it down.
Then set the second variable to the best value and so forth?
I would appreciate thoughts from the persons on this forum, who are obviously quite experienced and knowledgeable.
Thanks!