Premium
A new parallel method for verified global optimization
Author(s) -
Ibraev S.
Publication year - 2002
Publication title -
pamm
Language(s) - English
Resource type - Journals
ISSN - 1617-7061
DOI - 10.1002/1617-7061(200203)1:1<470::aid-pamm470>3.0.co;2-b
Subject(s) - computer science , reduction (mathematics) , parallel computing , process (computing) , distributed computing , global optimization , mathematical optimization , algorithm , mathematics , operating system , geometry
We present a new parallel method for verified global optimization, using challenge leadership for the dynamic load balancing. The new approach combines advantages of two previous models: the centralized mediator model (see [1]) and the processor farm (see [2]). It has the following properties: centralization of the process; reduction of the number of box exchanges, communications used to send boxes from one processor to another; handling of the box that most probably contains the global minimizer. Numerical results show the efficiency of this method.