A Line-Search Algorithm Inspired by the Adaptive Cubic Regularization Framework and Complexity Analysis

被引:10
作者
Bergou, El Houcine [1 ]
Diouane, Youssef [2 ]
Gratton, Serge [3 ]
机构
[1] Univ Paris Saclay, INRA, MaIAGE, F-78350 Jouy En Josas, France
[2] Univ Toulouse, ISAE SUPAERO, F-31055 Toulouse 4, France
[3] Univ Toulouse, INP ENSEEIHT, F-31071 Toulouse 7, France
关键词
Nonlinear optimization; Unconstrained optimization; Line-search methods; Adaptive regularized framework using cubics; Trust-region methods; Worst-case complexity; TRUST-REGION; UNCONSTRAINED OPTIMIZATION; NONLINEAR OPTIMIZATION; PERFORMANCE; NORM;
D O I
10.1007/s10957-018-1341-2
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
Adaptive regularized framework using cubics has emerged as an alternative to line-search and trust-region algorithms for smooth nonconvex optimization, with an optimal complexity among second-order methods. In this paper, we propose and analyze the use of an iteration dependent scaled norm in the adaptive regularized framework using cubics. Within such a scaled norm, the obtained method behaves as a line-search algorithm along the quasi-Newton direction with a special backtracking strategy. Under appropriate assumptions, the new algorithm enjoys the same convergence and complexity properties as adaptive regularized algorithm using cubics. The complexity for finding an approximate first-order stationary point can be improved to be optimal whenever a second-order version of the proposed algorithm is regarded. In a similar way, using the same scaled norm to define the trust-region neighborhood, we show that the trust-region algorithm behaves as a line-search algorithm. The good potential of the obtained algorithms is shown on a set of large-scale optimization problems.
引用
收藏
页码:885 / 913
页数:29
相关论文
共 20 条
  • [11] Benchmarking optimization software with performance profiles
    Dolan, ED
    Moré, JJ
    [J]. MATHEMATICAL PROGRAMMING, 2002, 91 (02) : 201 - 213
  • [12] CUTEst: a Constrained and Unconstrained Testing Environment with safe threads for mathematical optimization
    Gould, Nicholas I. M.
    Orban, Dominique
    Toint, Philippe L.
    [J]. COMPUTATIONAL OPTIMIZATION AND APPLICATIONS, 2015, 60 (03) : 545 - 557
  • [13] CUTEr and SifDec: a constrained and unconstrained testing environment, revisited
    Gould, NIM
    Orban, D
    Toint, PL
    [J]. ACM TRANSACTIONS ON MATHEMATICAL SOFTWARE, 2003, 29 (04): : 373 - 394
  • [14] Recursive trust-region methods for multiscale nonlinear optimization
    Gratton, Serge
    Sartenaer, Annick
    Toint, Philippe L.
    [J]. SIAM JOURNAL ON OPTIMIZATION, 2008, 19 (01) : 414 - 444
  • [15] Griewank A., 1981, Technical Report NA/12
  • [16] Cubic-regularization counterpart of a variable-norm trust-region method for unconstrained minimization
    Martinez, J. M.
    Raydan, M.
    [J]. JOURNAL OF GLOBAL OPTIMIZATION, 2017, 68 (02) : 367 - 385
  • [17] Nesterov Y., 2018, Introductory lectures on convex optimization: A basic course
  • [18] Cubic regularization of Newton method and its global performance
    Nesterov, Yurii
    Polyak, B. T.
    [J]. MATHEMATICAL PROGRAMMING, 2006, 108 (01) : 177 - 205
  • [19] SOLUTION OF SPARSE INDEFINITE SYSTEMS OF LINEAR EQUATIONS
    PAIGE, CC
    SAUNDERS, MA
    [J]. SIAM JOURNAL ON NUMERICAL ANALYSIS, 1975, 12 (04) : 617 - 629
  • [20] Recent advances in trust region algorithms
    Yuan, Ya-xiang
    [J]. MATHEMATICAL PROGRAMMING, 2015, 151 (01) : 249 - 281