From: aoki@CS.Berkeley.EDU (Paul M. Aoki) Subject: Re: "Invisible hand" algorithms? Date: Thu, 04 Feb 1999 17:13:24 -0800 Newsgroups: [missing] To: Dave Rusin Dave Rusin writes: > So, what exactly do they algorithms do? I mean, simulated annealing imagines > a particle hopping around on the graph as the temperature cools (so that > seeks a lower energy state nearby but is willing to jump around and test > new options, except as it cools it jumps less.) How would you describe > a technique mimicking economic activity? depends on the problem. the more obvious way (say, for load balancing): each allocation problem represents the equivalent of some transaction (in the monetary sense). transactions can be concurrent. doing work results in payment. each site/host/agent represents an entity trying to maximize income. they may advertise prices (retail); they may bid on offered work (contractor). i think we have different pictures of what constitutes a "problem". an economic approach requires iteration and feedback. (this is implicit in the phrase "invisible hand".) i don't know about work on using economics in (e.g.) conventional optimization problems; most of what i've seen has been in the context of systems. -- Paul M. Aoki | Computer Science Division, EECS #1776 aoki@CS.Berkeley.EDU | University of California | Berkeley, CA 94720-1776 USA ============================================================================== From: Dave Rusin Subject: Re: "Invisible hand" algorithms? Date: Thu, 4 Feb 1999 22:23:47 -0600 (CST) Newsgroups: [missing] To: aoki@CS.Berkeley.EDU >i think we have different pictures of what constitutes a "problem". I'm happy to take the broad perspective; if you've got a hammer, I'm sure we can find a nail to match. Thanks for your response. You're right, this kind of approach looks well suited to multiple-objective satisfiability problems even if no global objective is clearly stated. I'll have to try this some time when I have some convex optimization problem. dave