A New Switching Algorithm Based on Modified Quasi-Newton Updates

Author

Jiang, Xiaowei ; Lu, Yunlong ; Yang, Yueting

Author_Institution

Dept. of Math., Beihua Univ., Jilin, China

Volume

2

fYear

2010

fDate

28-31 May 2010

Firstpage

42

Lastpage

45

Abstract

We propose a new switching method for unconstrained minimization. The method is based on the quasi-Newton updates of Hiroshi Yabe, Hideho Ogasawara and Masayuki Yoshino. The eigenvalues and condition numbers of the modified SR1 update and the modified BFGS update are evaluated and used in the switching rule. When the condition number of the former is superior to that of the latter, the step is the modified SR1 step in the proposed quasi-Newton method, otherwise it is the modified BFGS step. The numerical results are reported and analyzed to show the superiority of the proposed method.

Keywords

Newton method; minimisation; modified quasi-Newton updates; switching method; unconstrained minimization; Convergence; Differential equations; Eigenvalues and eigenfunctions; Mathematics; Minimization methods; Optimization methods; Symmetric matrices; BFGS update; quasi-Newton; switching method; symmetric rank-one update; unconstrained optimization;

fLanguage

English

Publisher

ieee

Conference_Titel

Computational Science and Optimization (CSO), 2010 Third International Joint Conference on

Conference_Location

Huangshan, Anhui

Print_ISBN

978-1-4244-6812-6

Electronic_ISBN

978-1-4244-6813-3

Type

conf

DOI

10.1109/CSO.2010.107

Filename

5533138