Low Order-Value Optimization and applications

被引:19
作者
Andreani, R. [1 ]
Martinez, J. M. [1 ]
Martinez, L. [2 ,3 ]
Yano, F. S. [1 ,4 ]
机构
[1] Univ Estadual Campinas, IMECC UNICAMP, Dept Appl Math, BR-13081970 Campinas, SP, Brazil
[2] Univ Estadual Campinas, Inst Chem, BR-13081970 Campinas, SP, Brazil
[3] Inst Pasteur, Paris, France
[4] Itau Bank, Sao Paulo, Brazil
基金
巴西圣保罗研究基金会;
关键词
Order-Value Optimization; Algorithms; Convergence; Robust estimation of parameters; Hidden patterns; AUGMENTED LAGRANGIAN-METHODS; LINEAR-DEPENDENCE CONDITION; PROTEIN; ALGORITHMS; ALIGNMENT;
D O I
10.1007/s10898-008-9280-3
中图分类号
C93 [管理学]; O22 [运筹学];
学科分类号
070105 ; 12 ; 1201 ; 1202 ; 120202 ;
摘要
Given r real functions F (1)(x),...,F (r) (x) and an integer p between 1 and r, the Low Order-Value Optimization problem (LOVO) consists of minimizing the sum of the functions that take the p smaller values. If (y (1),...,y (r) ) is a vector of data and T(x, t (i) ) is the predicted value of the observation i with the parameters , it is natural to define F (i) (x) = (T(x, t (i) ) - y (i) )2 (the quadratic error in observation i under the parameters x). When p = r this LOVO problem coincides with the classical nonlinear least-squares problem. However, the interesting situation is when p is smaller than r. In that case, the solution of LOVO allows one to discard the influence of an estimated number of outliers. Thus, the LOVO problem is an interesting tool for robust estimation of parameters of nonlinear models. When p << r the LOVO problem may be used to find hidden structures in data sets. One of the most successful applications includes the Protein Alignment problem. Fully documented algorithms for this application are available at www.ime.unicamp.br/martinez/lovoalign. In this paper optimality conditions are discussed, algorithms for solving the LOVO problem are introduced and convergence theorems are proved. Finally, numerical experiments are presented.
引用
收藏
页码:1 / 22
页数:22
相关论文
共 33 条
[1]   ON AUGMENTED LAGRANGIAN METHODS WITH GENERAL LOWER-LEVEL CONSTRAINTS [J].
Andreani, R. ;
Birgin, E. G. ;
Martinez, J. M. ;
Schuverdt, M. L. .
SIAM JOURNAL ON OPTIMIZATION, 2008, 18 (04) :1286-1309
[2]   Augmented Lagrangian methods under the constant positive linear dependence constraint qualification [J].
Andreani, R. ;
Birgin, E. G. ;
Martinez, J. M. ;
Schuverdt, M. L. .
MATHEMATICAL PROGRAMMING, 2008, 111 (1-2) :5-32
[3]   On the relation between constant positive linear dependence condition and quasinormality constraint qualification [J].
Andreani, R ;
Martinez, JM ;
Schuverdt, M .
JOURNAL OF OPTIMIZATION THEORY AND APPLICATIONS, 2005, 125 (02) :473-485
[4]   Nonlinear-programming reformulation of the order-value optimization problem [J].
Andreani, R ;
Dunder, C ;
Martínez, JM .
MATHEMATICAL METHODS OF OPERATIONS RESEARCH, 2005, 61 (03) :365-384
[5]   Order-Value Optimization:: Formulation and solution by means of a primal cauchy method [J].
Andreani, R ;
Dunder, C ;
Martínez, JM .
MATHEMATICAL METHODS OF OPERATIONS RESEARCH, 2003, 58 (03) :387-399
[6]  
Andreani R, 2006, NONCON OPTIM ITS APP, V84, P379
[7]  
Andreani R., 2006, Pac J Optim, V2, P11
[8]  
ANDREANI R, 2005, 051013 MCDO STAT U C
[9]   Continuous optimization methods for structure alignments [J].
Andreani, Roberto ;
Martinez, Jose Mario ;
Martinez, Leandro ;
Yano, Flavio .
MATHEMATICAL PROGRAMMING, 2008, 112 (01) :93-124
[10]   Mesh adaptive direct search algorithms for constrained optimization [J].
Audet, C ;
Dennis, JE .
SIAM JOURNAL ON OPTIMIZATION, 2006, 17 (01) :188-217