New results on verified global optimization

被引：13

作者：

Berner, S

机构：

[1] Bergische Universität, Gesamthochschule Wuppertal, D-42097 Wuppertal

来源：

COMPUTING | 1996年 / 57卷 / 04期

关键词：

global optimization; interval arithmetic; branch and bound principle; best-first strategy; multisection;

D O I：

10.1007/BF02252252

中图分类号：

TP301 [理论、方法];

学科分类号：

081202 ;

摘要：

We study the influence of different strategies for selecting the next box for subdivision in an interval branch and bound method for global optimization. A theoretical result establishes the optimality of a 'best-first' strategy. Moreover, we define a multisection algorithm for subdivision of boxes. Different possibilities for the choice of bisection directions and the number of bisections are compared by numerical results showing that multisection is usually superior to bisection. In addition, we identify crucial properties of a division strategy which yield convergence of an algorithm with multisection.

引用

页码：323 / 343

页数：21

共 20 条

[1] Alefeld G., 1983, INTRO INTERVAL COMPU
[2] BAUMANN E, 1988, OPTIMAL CTR FORMS, V28, P80
[3] BERNER S, 1995, THESIS U GH WUPPERTA
[4] BERNER S, 1996, ZAMM NUMERICAL ANAL
[5] CSENDES T, 1997, IN PRESS SIAM J NUME
[6] GLOBAL OPTIMIZATION USING INTERVAL-ANALYSIS - THE MULTIDIMENSIONAL CASE
HANSEN, E
[J]. NUMERISCHE MATHEMATIK, 1980, 34 (03) : 247 - 270
[7] Hansen Eldon R., 1992, Global optimization using interval analysis
[8] GLOBAL OPTIMIZATION USING INTERVAL ANALYSIS - ONE-DIMENSIONAL CASE
HANSEN, ER
[J]. JOURNAL OF OPTIMIZATION THEORY AND APPLICATIONS, 1979, 29 (03) : 331 - 344
[9] HENRIKSEN T, 1992, 9210 TECHN U DENM I
[10] JANSSON C, 1994, TOPICS VALIDATED COM, P381

← 1 2 →