Globalizing Stabilized Sequential Quadratic Programming Method by Smooth Primal-Dual Exact Penalty Function

被引:16
作者
Izmailov, A. F. [1 ,4 ]
Solodov, M. V. [2 ]
Uskov, E. I. [3 ]
机构
[1] Moscow MV Lomonosov State Univ, VMK Fac, OR Dept, Uchebniy Korpus 2,Leninskiye Gory, Moscow 119991, Russia
[2] IMPA, Estr Dona Castorina 110, BR-22460320 Rio De Janeiro, RJ, Brazil
[3] Tambov State Univ, Sovetskaya Str 93, Tambov 392000, Russia
[4] Peoples Friendship Univ Russia, Miklukho Maklaya Str 6, Moscow 117198, Russia
基金
俄罗斯科学基金会; 俄罗斯基础研究基金会;
关键词
Stabilized sequential quadratic programming; Superlinear convergence; Global convergence; Exact penalty function; Second-order sufficiency; Noncritical Lagrange multiplier; SQP METHOD; CONSTRAINED OPTIMIZATION; AUGMENTED LAGRANGIANS; CONVERGENCE; MULTIPLIERS; ATTRACTION; BEHAVIOR;
D O I
10.1007/s10957-016-0889-y
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
An iteration of the stabilized sequential quadratic programming method consists in solving a certain quadratic program in the primal-dual space, regularized in the dual variables. The advantage with respect to the classical sequential quadratic programming is that no constraint qualifications are required for fast local convergence (i.e., the problem can be degenerate). In particular, for equality-constrained problems, the superlinear rate of convergence is guaranteed under the only assumption that the primal-dual starting point is close enough to a stationary point and a noncritical Lagrange multiplier (the latter being weaker than the second-order sufficient optimality condition). However, unlike for the usual sequential quadratic programming method, designing natural globally convergent algorithms based on the stabilized version proved quite a challenge and, currently, there are very few proposals in this direction. For equality-constrained problems, we suggest to use for the task linesearch for the smooth two-parameter exact penalty function, which is the sum of the Lagrangian with squared penalizations of the violation of the constraints and of the violation of the Lagrangian stationarity with respect to primal variables. Reasonable global convergence properties are established. Moreover, we show that the globalized algorithm preserves the superlinear rate of the stabilized sequential quadratic programming method under the weak conditions mentioned above. We also present some numerical experiments on a set of degenerate test problems.
引用
收藏
页码:148 / 178
页数:31
相关论文
共 30 条
[1]   A relaxed constant positive linear dependence constraint qualification and applications [J].
Andreani, Roberto ;
Haeser, Gabriel ;
Laura Schuverdt, Maria ;
Silva, Paulo J. S. .
MATHEMATICAL PROGRAMMING, 2012, 135 (1-2) :255-273
[2]  
[Anonymous], 2019, Reinforcement Learning and Optimal Control
[3]  
[Anonymous], SPRINGER SERIES OPER
[5]  
Bonnans JF, 2006, NUMERICAL OPTIMIZATI, V2nd, DOI 10.1007/978-3-662-05078-1
[6]   NEW CLASS OF AUGMENTED LAGRANGIANS IN NON-LINEAR PROGRAMMING [J].
DIPILLO, G ;
GRIPPO, L .
SIAM JOURNAL ON CONTROL AND OPTIMIZATION, 1979, 17 (05) :618-628
[7]   Benchmarking optimization software with performance profiles [J].
Dolan, ED ;
Moré, JJ .
MATHEMATICAL PROGRAMMING, 2002, 91 (02) :201-213
[8]   An inexact restoration strategy for the globalization of the sSQP method [J].
Fernandez, D. ;
Pilotta, E. A. ;
Torres, G. A. .
COMPUTATIONAL OPTIMIZATION AND APPLICATIONS, 2013, 54 (03) :595-617
[9]   Stabilized sequential quadratic programming for optimization and a stabilized Newton-type method for variational problems [J].
Fernandez, Damian ;
Solodov, Mikhail .
MATHEMATICAL PROGRAMMING, 2010, 125 (01) :47-73