The trust region subproblem and semidefinite programming

被引:104
作者
Fortin, C
Wolkowicz, H [1 ]
机构
[1] Univ Waterloo, Dept Combinator & Optimizat, Waterloo, ON N2L 3G1, Canada
[2] McGill Univ, Dept Math & Stat, Montreal, PQ H3A 2T5, Canada
关键词
trust regions; semidefinite programming; duality; unconstrained minimization;
D O I
10.1080/10556780410001647186
中图分类号
TP31 [计算机软件];
学科分类号
081202 ; 0835 ;
摘要
The trust region (TR) subproblem (the minimization of a quadratic objective subject to one quadratic constraint and denoted TRS) has many applications in diverse areas, e.g., function minimization, sequential quadratic programming, regularization, ridge regression, and discrete optimization. In particular, it determines the step in TR algorithms for function minimization. Trust region algorithms are popular for their strong convergence properties. However, a drawback has been the inability to exploit sparsity as well as the difficulty in dealing with the so-called hard case. These concerns have been addressed by recent advances in the theory and algorithmic development. In particular, this has allowed Lanczos techniques to replace Cholesky factorizations. This article provides an in depth study of TRS and its properties as well as a survey of recent advances. We emphasize large scale problems and robustness. This is done using semidefinite programming (SDP) and the modern primal-dual approaches as a unifying framework. The SDP framework arises naturally and solves TRS efficiently. In addition, it shows that TRS is always a well-posed problem, i.e., the optimal value and an optimum can be calculated to a given tolerance. We provide both theoretical and empirical evidence to illustrate the strength of the SDP and duality approach. In particular, this includes new insights and techniques for handling the hard case, as well as numerical results on large test problems.
引用
收藏
页码:41 / 67
页数:27
相关论文
共 40 条
[1]   Solving large-scale sparse semidefinite programs for combinatorial optimization [J].
Benson, SJ ;
Ye, YY ;
Zhang, X .
SIAM JOURNAL ON OPTIMIZATION, 2000, 10 (02) :443-461
[2]   Hidden convexity In some nonconvex quadratically constrained quadratic programming [J].
BenTal, A ;
Teboulle, M .
MATHEMATICAL PROGRAMMING, 1996, 72 (01) :51-63
[3]  
BJORCK A, 1996, NUMERICAL METHODS LE
[4]  
BOURIACHA A, COMMUNICATION
[5]  
Conn A., 2000, SOC IND APPL MATH, DOI [10.1137/1.9780898719857, DOI 10.1137/1.9780898719857]
[6]   TRUNCATED-NEWTON ALGORITHMS FOR LARGE-SCALE UNCONSTRAINED OPTIMIZATION [J].
DEMBO, RS ;
STEIHAUG, T .
MATHEMATICAL PROGRAMMING, 1983, 26 (02) :190-212
[7]  
Demmel JW., 1997, APPL NUMERICAL LINEA
[8]  
Fiacco A.V., 1983, Mathematics in Science and Engineering, V165
[9]  
Fletcher R., 1981, PRACTICAL METHODS OP
[10]   ON STATIONARY VALUES OF A 2ND-DEGREE POLYNOMIAL ON UNIT SPHERE [J].
FORSYTHE, GE ;
GOLUB, GH .
JOURNAL OF THE SOCIETY FOR INDUSTRIAL AND APPLIED MATHEMATICS, 1965, 13 (04) :1050-&