A New Kind of Science: The NKS Forum (http://forum.wolframscience.com/index.php)
- NKS Way of Thinking (http://forum.wolframscience.com/forumdisplay.php?forumid=5)

Posted by Jonathan Kramer on 10-03-2003 03:53 AM:

Optimal algorithm combinations

Here is a question:

Within the same 24 hour period that Wolfram's interview aired on C-Span, there was another program on "WARDTV" about wine. This program reported that the world's favorite wine: "Cabernet Sauvignon" was found to be genetically composed of two grapes, the distinguished classic "Pinot Noir" and another grape few of us have ever heard of, one that makes a mediocre wine at best. Yet a little of that variety appears necessary to create the masterpiece. For me this was vaguely reminiscent of that horrible smelling fish sauce they use in Thailand, a dash of which is necessary to make the finest curry all it can be.

So much for intro. My question:

Given two algorithms one seeks to combine in order to produce what one considers a possibly optimized result, what strategy should one use (of course ruling out trial and error) to generate the optimum homeostatic proportions of the two algorithms?

This question may be more important than it appears on first blush.

Posted by RNewsom on 02-10-2004 04:31 PM:

Here's another: I have a bakery making commercial bread. There's a small demand for unsliced loaves, and after consulting my accountants and a few demographic studies, it's determined that I can sell unsliced loaves, even at a slightly greater price. (I must divert some loaves, separate them, label them differently, thus the extra charge) My demand for unsliced loaves slowly increases. At what percentage point of increasing sales of unsliced loaves do I begin suddenly charging more for sliced loaves and less for the unsliced?

Posted by Kovas Boguta on 02-10-2004 05:40 PM:

You must be hungry

mm.

__________________
Everything is an expression.