GLOBAL CONVERGENCE OF GENERAL DERIVATIVE-FREE TRUST-REGION ALGORITHMS TO FIRST- AND SECOND-ORDER CRITICAL POINTS

被引:144
作者
Conn, Andrew R. [1 ]
Scheinberg, Katya [1 ]
Vicente, Luis N. [2 ]
机构
[1] IBM Corp, Thomas J Watson Res Ctr, Dept Math Sci, Yorktown Hts, NY 10598 USA
[2] Univ Coimbra, Dept Math, CMUC, P-3001454 Coimbra, Portugal
关键词
trust-region methods; derivative-free optimization; nonlinear optimization; global convergence; FREE OPTIMIZATION; INTERPOLATION; GEOMETRY; SETS;
D O I
10.1137/060673424
中图分类号
O29 [应用数学];
学科分类号
070104 ;
摘要
In this paper we prove global convergence for first- and second-order stationary points of a class of derivative-free trust-region methods for unconstrained optimization. These methods are based on the sequential minimization of quadratic (or linear) models built from evaluating the objective function at sample sets. The derivative-free models are required to satisfy Taylor-type bounds, but, apart from that, the analysis is independent of the sampling techniques. A number of new issues are addressed, including global convergence when acceptance of iterates is based on simple decrease of the objective function, trust-region radius maintenance at the criticality step, and global convergence for second-order critical points.
引用
收藏
页码:387 / 415
页数:29
相关论文
共 14 条
[1]   Optimizing partially separable functions without derivatives [J].
Colson, B ;
Toint, PL .
OPTIMIZATION METHODS & SOFTWARE, 2005, 20 (4-5) :493-508
[2]   Geometry of interpolation sets in derivative free optimization [J].
Conn, A. R. ;
Scheinberg, K. ;
Vicente, Luis N. .
MATHEMATICAL PROGRAMMING, 2008, 111 (1-2) :141-172
[3]  
Conn A. R., 1997, On the convergence of derivative-free methods for unconstrained optimization, P83
[4]   Geometry of sample sets in derivative-free optimization: polynomial regression and underdetermined interpolation [J].
Conn, Andrew R. ;
Scheinberg, Katya ;
Vicente, Luis N. .
IMA JOURNAL OF NUMERICAL ANALYSIS, 2008, 28 (04) :721-748
[5]  
CONN AR, 2000, MPS SIAM O
[6]  
CONN AR, 2009, MPS SIAM O
[7]   Wedge trust region methods for derivative free optimization [J].
Marazzi, M ;
Nocedal, J .
MATHEMATICAL PROGRAMMING, 2002, 91 (02) :289-305
[8]  
Nocedal J, 2006, SPRINGER SER OPER RE, P1, DOI 10.1007/978-0-387-40065-5
[9]  
Powell J., CONFLICT TRENDS, P4
[10]  
Powell M. J. D., 1998, Acta Numerica, V7, P287, DOI 10.1017/S0962492900002841