Event Details
Global Optimization by Interval Constraints
Presenter: M.H. van Emden, Department of Computer Science
Supervisor:
Date: Fri, November 16, 2001
Time: 10:30:00 - 12:00:00
Place: EOW 430
ABSTRACT
Abstract:
We consider numerical optimization problems where the objective function has an unknown and possibly large number of local minima. Widely known methods are either stochastic or act by finding a local minimum. None can be relied on to find the global optimum.
In this tutorial lecture I review applications of global optimization and outline the main features of the widely known methods. I then explain how interval methods can bound the global optimum on both sides with bounds that are as close as machine limitations allow. I give a tutorial introduction to the two interval methods: interval arithmetic and interval constraints.
This lecture is a rehearsal of a tutorial to be given at the CP2001 conference.