Abstract
A new method for nonlinear minimax problems is presented. The method is of the trust region type and based on sequential linear programming. It is a first order method that only uses first derivatives and does not approximate Hessians. The new method is well suited for large sparse problems as it only requires that software for sparse linear programming and a sparse symmetric positive definite equation solver are available. On each iteration a special linear/quadratic model of the function is minimized, but contrary to the usual practice in trust region methods the quadratic model is only defined on a one dimensional path from the current iterate to the boundary of the trust region. Conjugate gradients are used to define this path. One iteration involves one LP subproblem and requires three function evaluations and one gradient evaluation. Promising numerical results obtained with the method are presented. In fact, we find that the number of iterations required is comparable to that of state-of-the-art quasi-Newton codes.
Similar content being viewed by others
References
D.H. Andersson and M.R. Osborne, Discrete nonlinear approximation in polyhedral norms: a Levenberg-like algorithm, Numer. Math. 28 (1977) 157–170.
C. Charalambous and A.R. Conn, An efficient method to solve the minimax problem directly, SIAM J. Numer. Anal. 15 (1978) 162–187.
A.R. Conn, An efficient second order method to solve the (constrained) minimax problem, Report CCRR 79-5, Dept. of Combinatorics and Optimization, Univ. of Waterloo, Canada (1979).
A.R. Conn, N.I.M. Gould and Ph.L. Toint,LANCELOT: A Fortran Package for Large Scale Nonlinear Optimization (Release A) (Springer, 1992).
A.R. Conn and Yuying Li, An efficient algorithm for nonlinear minimax problems, University of Waterloo report CS-88-41 (1989).
L. Cromme, Strong uniqueness, Numer. Math. 29 (1978) 179–194.
R. Fletcher, A model algorithm for composite nondifferentiable optimization problems, Math. Prog. Study 17 (1982) 67–76.
R. Fletcher,Practical Methods of Optimization, 2nd ed. (Wiley, 1987).
R. Fletcher, Low storage methods for unconstrained optimization, Report NA117, Dept. of Math. Sciences, University of Dundee (1988).
R. Fletcher and E. Sainz de la Maza, Nonlinear programming and nonsmooth optimization by successive linear programming, Math. Prog. 43 (1989).
J. Hald, MMLA1Q, a FORTRAN subroutine for linearly constrained minimax optimization, Report no. NI-81-01, Inst. for Numer. Anal., Technical University of Denmark (1981).
J. Hald and K. Madsen, Combined LP and quasi-Newton methods for minimax optimization, Math. Prog. 20 (1981).
S.P. Han, Variable metric methods for minimizing a class of nondifferentiable functions, Math. Prog. 20 (1981) 1–13.
R. Hettich, A Newton-method for nonlinear Chebyshev approximation, in:Approximation Theory, eds. R. Schaback and K. Scherer, Lect. Notes in Math. 556 (Springer, Berlin, 1976) pp. 222–236.
K. Jōnasson and K. Madsen, Corrected sequential linear programming for sparse minimax optimization, Report NI-92-06, Inst. for Numer. Anal., Technical University of Denmark (1992).
K. Madsen, An algorithm for minimax solution of over-determined systems of non-linear equations, J. Inst. Math. Appl. 16 (1975) 321–328.
K. Madsen, O. Tingleff, P.Chr. Hansen and W. Owczarc, Robust subroutines for non-linear optimization, Report NI-90-06, Inst. for Numer. Anal., Technical University of Denmark (1990).
W. Murray and M.L. Overton, A projected Lagrangian algorithm for nonlinear minimax optimization, SIAM J. Sci. Stat. Comp. 1 (1980) 345–370.
M.R. Osborne and G.A. Watson, An algorithm for minimax optimization in the non-linear case, Comp. J. 12 (1969) 63–68.
G. Di Pillo, L. Grippo and S. Lucidi, A smooth method for the finite minimax problem, Report Rap.05.92, Dipartimento di Informatica e Sistematica, Università di Roma (1992).
G.A. Watson, The minimax solution of an overdetermined system of non-linear equations, J. Inst. Math. Appl. 23 (1979) 167–180.
Author information
Authors and Affiliations
Additional information
Research supported by The Nordic Council of Ministers, The Icelandic Science Council, The University of Iceland Research Fund and The Danish Natural Science Research Council.
Rights and permissions
About this article
Cite this article
Jónasson, K. A projected conjugate gradient method for sparse minimax problems. Numer Algor 5, 309–323 (1993). https://doi.org/10.1007/BF02108465
Issue Date:
DOI: https://doi.org/10.1007/BF02108465