Re: [Swarm-Modelling] Is there such a thing as approximate optimization?

[Andre Costa]

Any approach that is not theoretically guaranteed to find an optimal solution is generally referred to as a heuristic method for estimating or approximating an optimal solution (for this to make sense, one needs to have posed an optimisation problem, even if you can't solve it exactly).

Constrained optimisation is any optimisation problem with constraints, and does not imply anything about how one estimates or approximates a solution to the problem.

Andre.



On 06/11/2006, at 11:54 AM, Steve Railsback wrote:

> How should I refer to a decision-making approach in which agents make a rough-but-realistic approximation of the outcome of each decision alternative and then select the alternative with the best approximated outcome?
>
> Is there already a term for this approach? Can you call it approximate optimization? Or optimal approximation?
>
> The term "constrained optimization" is often used; is there a precedent for applying it to this approach?
>
> Steve Railsback
>
> _______________________________________________
> Modelling mailing list
> address@hidden
> http://www.swarm.org/mailman/listinfo/modelling
-------------------------------------------------
Andre Costa
Post-Doctoral Research Fellow
Centre of Excellence for Mathematics and Statistics of Complex Systems
University of Melbourne
Victoria, 3010
Australia

Email: address@hidden
Telephone: (03) 8344 1619
Fax: 	     (03) 9347 8165
www.ms.unimelb.edu.au/~acosta/main.htm
------------------------------------------------