Abstract
We propose a new truncated Newton method for large scale unconstrained optimization, where a Conjugate Gradient (CG)-based technique is adopted to solve Newton’s equation. In the current iteration, the Krylov method computes a pair of search directions: the first approximates the Newton step of the quadratic convex model, while the second is a suitable negative curvature direction. A test based on the quadratic model of the objective function is used to select the most promising between the two search directions. Both the latter selection rule and the CG stopping criterion for approximately solving Newton’s equation, strongly rely on conjugacy conditions. An appropriate linesearch technique is adopted for each search direction: a nonmonotone stabilization is used with the approximate Newton step, while an Armijo type linesearch is used for the negative curvature direction. The proposed algorithm is both globally and superlinearly convergent to stationary points satisfying second order necessary conditions. We carry out a significant numerical experience in order to test our proposal.
Similar content being viewed by others
References
Dembo R.S., Steihaug T.: Truncated-Newton algorithms for large-scale unconstrained optimization. Math. Program. 26, 190–212 (1983)
Dolan, E.D., More, J.J.: Benchmarking Optimization Software with Performance Profiles. Mathematics and Computer Science Division. Preprint ANL/MCS-P861-1200
Fasano, G., Lucidi, S.: A nonmonotone truncated Newton–Krylov method exploiting negative curvature directions, for large scale unconstrained optimization: complete results. Technical Report INSEAN 2008-035/rt, 2008
Fasano G., Roma M.: Iterative computation of negative curvature directions in large scale optimization. Comput. Optim. Appl. 38(1), 81–104 (2007)
Ferris M.C., Lucidi S., Roma M.: Nonmonotone curvilinear linesearch methods for unconstrained optimization. Comput. Optim. Appl. 6, 117–136 (1996)
Forsgren A.: On the Behavior of the Conjugate-Gradient Method on Ill-conditioned Problems. Technical Report TRITA-MAT-2006-OS1, Department of Mathematics, Royal Institute of Technology
Goldfarb D.: Curvilinear path steplength algorithms for minimization which use directions of negative curvature. Math. Program. 18, 31–40 (1980)
Gould N.I.M., Lucidi S., Roma M., Toint Ph.L.: Exploiting negative curvature directions in linesearch methods for unconstrained optimization. Optim. Methods Softw. 14, 75–98 (2000)
Gould N.I.M., Orban D., Toint Ph.L.: CUTEr: Constrained and unconstrained testing environment, revised. Trans. ACM Math. Softw. 29(4), 373–394 (2003)
Gould N.I.M., Orban D., Toint Ph.L.: GALAHAD, a library of thread-safe Fortran 90 packages for large-scale nonlinear optimization. ACM Trans. Math. Softw. 29(4), 353–372 (2003)
Gould N.I.M., Sainvitu C., Toint Ph.L.: A Filter-Trust-Region method for unconstrained optimization. SIAM J. Optim. 16(2), 341–357 (2006)
Grippo L., Lampariello F., Lucidi S.: A truncated Newton method with nonmonotone linesearch for unconstrained optimization. J. Optim. Theory Appl. 60, 401–419 (1989)
Grippo L., Lampariello F., Lucidi S.: A class of nonmonotone stabilization methods in unconstrained optimization. Numer. Math. 59, 779–805 (1991)
Lucidi S., Rochetich F., Roma M.: Curvilinear stabilization techniques for truncated Newton methods in large scale unconstrained optimization. SIAM J. Optim. 8, 916–939 (1998)
Lucidi S., Roma M.: Numerical experiences with new truncated Newton methods in large scale unconstrained optimization. Comput. Optim. Appl. 7, 71–87 (1997)
McCormick G.P.: A modification of Armijo’s stepsize rule for negative curvature. Math. Program. 13, 111–115 (1977)
More’ J.J., Sorensen D.C.: On the use of directions of negative curvature in a modified Newton method. Math. Program. 16, 1–20 (1979)
Mukai H., Polak E.: A second-order method for unconstrained optimization. J. Optim. Theory Appl. 26, 501–513 (1978)
Nash S.G., Sofer A.: Assessing a search direction within a truncated Newton method. Oper. Res. Lett. 9, 219–221 (1990)
Olivares A., Moguerza J.M., Prieto F.J.: Nonconvex optimization using negative curvature within a modified linesearch. Euro. J. Oper. Res. 189, 706–722 (2008)
Shultz G.A., Schnabel R.B., Byrd R.H.: A family of trust-region-based algorithms for unconstrained minimization. SIAM J. Numer. Anal. 22, 47–67 (1985)
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Fasano, G., Lucidi, S. A nonmonotone truncated Newton–Krylov method exploiting negative curvature directions, for large scale unconstrained optimization. Optim Lett 3, 521–535 (2009). https://doi.org/10.1007/s11590-009-0132-y
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11590-009-0132-y