Curvilinear Stabilization Techniques for Truncated Newton Methods in Large Scale Unconstrained Optimization

The aim of this paper is to define a new class of minimization algorithms for solving large scale unconstrained problems. In particular we describe a stabilization framework, based on a curvilinear linesearch, which uses a combination of a Newton-type direction and a negative curvature direction. The motivation for using negative curvature direction is that of taking into account local nonconvexity of the objective function. On the basis of this framework, we propose an algorithm which uses the Lanczos method for determining at each iteration both a Newton-type direction and an effective negative curvature direction. The results of extensive numerical testing are reported together with a comparison with the LANCELOT package. These results show that the algorithm is very competitive, which seems to indicate that the proposed approach is promising.

[1]  James M. Ortega,et al.  Iterative solution of nonlinear equations in several variables , 2014, Computer science and applied mathematics.

[2]  M. Saunders,et al.  Solution of Sparse Indefinite Systems of Linear Equations , 1975 .

[3]  Garth P. McCormick,et al.  A modification of Armijo's step-size rule for negative curvature , 1977, Math. Program..

[4]  Danny C. Sorensen,et al.  On the use of directions of negative curvature in a modified newton method , 1979, Math. Program..

[5]  B. Parlett.  The Symmetric Eigenvalue Problem , 1981 .

[6]  Philippe L. Toint,et al.  Towards an efficient sparsity exploiting newton method for minimization , 1981 .

[7]  Richard H. Byrd,et al.  A Family of Trust Region Based Algorithms for Unconstrained Minimization with Strong Global Convergence Properties. , 1985 .

[8]  R. Dembo,et al.  INEXACT NEWTON METHODS , 1982 .

[9]  Trond Steihaug,et al.  Truncated-newtono algorithms for large-scale unconstrained optimization , 1983, Math. Program..

[10]  G. McCormick.  Nonlinear Programming: Theory, Algorithms and Applications , 1983 .

[11]  S. Nash.  Newton-Type Minimization via the Lanczos Method , 1984 .

[12]  J. Cullum,et al.  Lanczos algorithms for large symmetric eigenvalue computations , 1985 .

[13]  S. Nash.  Preconditioning of Truncated-Newton Methods , 1985 .

[14]  L. Grippo,et al.  A nonmonotone line search technique for Newton's method , 1986 .

[15]  R. Fletcher.  Practical Methods of Optimization , 1988 .

[16]  L. Grippo,et al.  A truncated Newton method with nonmonotone line search for unconstrained optimization , 1989 .

[17]  L. Grippo,et al.  A class of nonmonotone stabilization methods in unconstrained optimization , 1991 .

[18]  P. Toint,et al.  Lancelot: A FORTRAN Package for Large-Scale Nonlinear Optimization (Release A) , 1992 .

[19]  Danny C. Sorensen,et al.  Implicit Application of Polynomial Filters in a k-Step Arnoldi Method , 1992, SIAM J. Matrix Anal. Appl..

[20]  Tamar Schlick,et al.  TNPACK—A truncated Newton minimization package for large-scale problems: I. Algorithm and usage , 1992, TOMS.

[21]  T. Chan,et al.  Computing a search direction for large scale linearly constrained nonlinear optimization calculations , 1993 .

[22]  N. Deng,et al.  Nonmonotonic trust region algorithm , 1993 .

[23]  R. D. Murphy,et al.  Iterative solution of nonlinear equations , 1994 .

[24]  Dimitri P. Bertsekas,et al.  Nonlinear Programming , 1997 .

[25]  Nicholas I. M. Gould,et al.  CUTE: constrained and unconstrained testing environment , 1995, TOMS.

[26]  Michael C. Ferris,et al.  Nonmonotone curvilinear line search methods for unconstrained optimization , 1996, Comput. Optim. Appl..

[27]  Philippe L. Toint,et al.  An Assessment of Nonmonotone Linesearch Techniques for Unconstrained Optimization , 1996, SIAM J. Sci. Comput..

[28]  Philippe L. Toint,et al.  Non-monotone trust-region algorithms for nonlinear optimization subject to convex constraints , 1997, Math. Program..