A CLASS OF NONMONOTONE STABILIZATION METHODS IN UNCONSTRAINED OPTIMIZATION

被引：122

作者：

GRIPPO, L ^{[1
]}

LAMPARIELLO, F ^{[1
]}

LUCIDI, S ^{[1
]}

机构：

[1] CNR, IST ANAL SIST & INFORMAT, I-00185 ROME, ITALY

来源：

NUMERISCHE MATHEMATIK | 1991年 / 59卷 / 08期

关键词：

D O I：

10.1007/BF01385810

中图分类号：

O29 [应用数学];

学科分类号：

070104 ;

摘要：

This paper deals with the solution of smooth unconstrained minimization problems by Newton-type methods whose global convergence is enforced by means of a nonmonotone stabilization strategy. In particular, a stabilization scheme is analyzed, which includes different kinds of relaxation of the descent requirements. An extensive numerical experimentation is reported.

引用

页码：779 / 805

页数：27

共 24 条

[1]

BERTSEKAS DP, 1980, CONSTRAINED OPTIMIZA

[2] A GLOBALLY AND QUADRATICALLY CONVERGENT ALGORITHM FOR GENERAL NON-LINEAR PROGRAMMING-PROBLEMS [J].