In this paper we present a hybrid path-following algorithm that generates inexact Newton steps suited for solving large scale and/or degenerate nonlinear programs. The algorithm uses as a central region a relaxed notion of the central path, called quasicentral path, a generalized augmented Lagrangian function, weighted proximity measures, and a linesearch within a trust region strategy. We apply a semi-iterative method for obtaining inexact Newton steps by using the conjugate gradient algorithm as an iterative procedure. We present a numerical comparison, and some promising results are reported.

Key words: Interior-point methods, trust region methods, linesearch technique, nonlinear programming, and conjugate gradient.

En este artículo nosotros presentamos un algoritmo híbrido de seguimiento de camino que genera pasos inexactos de Newton para resolver problemas de gran escala o degenerados para programación no lineal. El algoritmo usa como una region de centralidad una noción mas débil que el bien conocido camino central, llamada camino quasi-central, una generalización de la función aumentada de Lagrange, medidas de aproximación pesadas, y una dirección de búsqueda dentro de una region de verdad. Nosotros aplicamos un método semi-iterativo para obtener direcciones inexactas del método de Newton usando el algoritmo del gradiente conjugado y presentamos una comparación numérica con resultados prometedores.

Palabras clave: Métodos de punto interior, Métodos de región verdadera, técnica de búsqueda de línea, programación nolinear y gradiente conjugado.

Texto completo disponible en PDF

References

1 M. Argáez, An inexact projected Newton method for large scale linear programming, Technical Report, Department of Mathematical Sciences, The University of Texas at El Paso, 2006.         [ Links ]

2 M. Argáez, Exact and inexact Newton linesearch interior-point algorithms for nonlinear programming problems, Ph.D Thesis, Department of Computational and Applied Mathematics, Rice University, Houston, Texas, 1997.         [ Links ]

3 M. Argáez, H. Klie, M. Ramé & R. A. Tapia, An interior-point Krylov- Orthogonal projection method for nonlinear programming, Tec. Rep. 97-16, Department of Computational and Applied Mathematics, Rice University, Houston, Texas, 1997.         [ Links ]

4 M. Argáez & R. A. Tapia, On the global convergence of a modified augmented Lagrangian linesearch interior-point Newton method for nonlinear programming, J. Optim. Theory Appl., 114 (2002), 1-25.         [ Links ]

5 M. Argáez, R. A. Tapia & L. Velázquez, Numerical Comparisons of path- following strategies for a primal-dual Interior-Point method for nonlinear programming. J. Optim. Theory Appl., 114 (2002) 4, 255-272.         [ Links ]

6 M. Argáez, L. Velázquez & M.C. Villalobos, On the development of a trust region interior-point method for large scale nonlinear programs, ACM Proc. in Diversity of Comput. Conf., (2003), 14-16.         [ Links ]

7 M. Argáez, O. Mendez & L. Velázquez, 2005. The notion of the quasicentral path for linear programming, Technical Report, Department of Mathematical Sciences, The University of Texas at El Paso, El Paso, Texas, 2006.         [ Links ]

8 R. Byrd, J.C. Gilbert & J. Nocedal, A trust region method based on interior point techniques for nonlinear programming, Math. Program. A, 89 (2000), 149-185.         [ Links ]

9 J.E. Dennis, Jr., & R.B. Schnabel, Numerical methods for unconstrained optimization and nonlinear equations, 2nd ed., Society for Industrial and Applied Mathematics, Philadelphia PA, 1996.         [ Links ]

10 M. El-Alem, A global convergence theory for a class of trust region algorithms for constrained optimization, Ph.D Thesis, Rice University, Houston, Texas, (1998).         [ Links ]

11 A.S. El-Bakry, R.A. Tapia, T. Tsuchiya & Y. Zhang, On the formulation of the primal-dual interior-point method for nonlinear programming, J. Optim. Theory Appl., 89 (1996), 507-541.         [ Links ]

12 N.I.M. Gould, M. E. Hribar, & J. Nocedal, On the solution of equality constrained quadratic programming problems arising in optimization, SIAM J. Sci. Comput., 23 (2001), 1375-1394.         [ Links ]

13 M.R. Hestenes, Multiplier and gradient methods, J. Optim. Theory Appl., 4 (1969), 303-320.         [ Links ]

14 D. Kincaid, & W. Cheney, Numerical analysis: mathematics of scientific computing, 3rd ed., Brooks-Cole, Boston MA, (2002).         [ Links ]

15 D.G. Luenberger, Linear and nonlinear programming, 2nd ed., Addison-Wesley, Reading MA, (1984).         [ Links ]

16 J. Nocedal & S.J. Wright, Numerical optimization, Springer-Verlag, (1999).         [ Links ]

17 R. Sáenz, An inexact Newton interior-point algorithm for nonnegative constrained minimization, Master Thesis, Department of Mathematical Sciences, The University of Texas at El Paso, El Paso, Texas (2003).         [ Links ]

18 D.f. Shanno & R. J. Vanderbei, Interior-point methods for nonconvex nonlinear programming: Orderings and higher-order methods, Math. Prog., 87 (2000), 303-316.         [ Links ]

19 P. Tseng & Y. Ye, On some interior-point algorithms for nonconvex quadratic optimization, Math. Prog. 93 (2002), 217-225.         [ Links ]

20 Y. Zhang, Solving large-scale linear programs by interior-point methods under the MATLAB environment, Optimization Methods and Software, 10 (1998), 1-31.         [ Links ]