Kybernetika 49 no. 2, 236-257, 2013

A new regular multiplier embedding

Gemayqzel Bouza Allende and Jürgen Guddat

Abstract:

Embedding approaches can be used for solving non linear programs \emph{P}. The idea is to define a one-parametric problem such that for some value of the parameter the corresponding problem is equivalent to \emph{P}. A particular case is the multipliers embedding, where the solutions of the corresponding parametric problem can be interpreted as the points computed by the multipliers method on \emph{P}. However, in the known cases, either path-following methods can not be applied or the necessary conditions for its convergence are fulfilled under very restrictive hypothesis. In this paper, we present a new multipliers embedding such that the objective function and the constraints of $P(t)$ are $C^3$ differentiable functions. We prove that the parametric problem satisfies the \emph{JJT}-regularity generically, a necessary condition for the success of the path-following method.

Keywords:

Jongen-Jonker-Twilt regularity, multipliers method, embedding

Classification:

90C31, 49M30

References:

  1. M. Afonso, J. Bioucas-Dias and M. Figueiredo: Fast image recovery using variable splitting and constrained optimization. IEEE Trans. Signal Process. 19 (2010), 2345-2356.   CrossRef
  2. R. Andreani, E. G. Birgin, J. M. Martínez and M. L. Schuverdt: On augmented Lagrangian methods with general lower-level constraints. SIAM J. Optim. 28 (2008), 1286-1309.   CrossRef
  3. C Avelino and L. N. Vicente: Updating the multipliers associated with inequality constraints in an augmented Lagrangian multiplier method. J. Optim. Theory Appl. 199 (2003), 215-233.   CrossRef
  4. M. S. Bazaraa, H. D. Sherali and C. M. Shetty: Non Linear Programming Theory and Algorithms. John Willey and Sons, 1993.   CrossRef
  5. D. P. Bertsekas: Constrained Optimization and Lagrange Multiplier Methods. Academic Press, New York 1982.   CrossRef
  6. E. G. Birgin and J. M. Martínez: Augmented lagrangian method with nonmonotone penalty parameters for constrained. Optimization Online, \url{http://www.optimization-online.org/DB_FILE/2010/06/2662.pdf} 2010.   CrossRef
  7. G. Bouza: A new embedding for the augmented Lagrangean method. Investigación Oper. 22 (2001), 145-153.   CrossRef
  8. G. Bouza and J. Guddat: A note on embeddings for the augmented Lagrange method. Yugosl. J. Oper. Res. 20 (2010), 183-196.   CrossRef
  9. D. Dentcheva, R. Gollmer, J. Guddat and J. R{ü}ckmann: Pathfollowing methods in non linear optimization, multipliers embedding. ZOR 41 (1995), 127-152.   CrossRef
  10. Z. Dostal, A. Friedlander and A. Santos: Augmented Lagrangians with adaptative precision control for quadratic programming problems with equality constrains. Comput. Optim. Appl. 14 (1999), 37-53.   CrossRef
  11. R. Gollmer, U. Kausmann, D. Nowack, K. Wendler and J. Bacallao Estrada: Computerprogramm PAFO. Humboldt-Universitaet, Institut fuer Mathematik 2004.   CrossRef
  12. W. G{ó}mez: On generic quadratic penalty embeddings for non linear optimization problems. Optimization 50 (2001), 279-295.   CrossRef
  13. W. Gómez, J. Guddat, H. Th. Jongen, J. J. R{ü}ckmann and C. Solano: Curvas criticas y saltos en la optimizacion no lineal. \url{http://www.emis.de/monographs/curvas/index.html} 2000.   CrossRef
  14. J. Guddat, F. Guerra and H. Th. Jongen: Parametric Optimization: Singularities, Pathfollowing and Jumps. Teubner and John Wiley, Chichester 1990.   CrossRef
  15. M. Hirsch: Differential Topology. Springer Verlag, New York 1976.   CrossRef
  16. A. N. Iusem: Augmented Lagrangean methods and proximal point methods for convex optimization. Investigación Oper. 8 (1999), 11-49.   CrossRef
  17. H. Th. Jongen, P. Jonker and F. Twilt: Critical sets in parametric optimization. Math. Programming 34 (1986), 333-353.   CrossRef
  18. H. Th. Jongen, P. Jonker and F. Twilt: On one-parametrer families of optimization problems: Equality constrains. J. Optim. Theory Appl. 48 (1986), 141-161.   CrossRef
  19. D. Li and X. L. Sun: Local convexification of the Lagrangian function in non-convex poptimization. J. Optim. Theory Appl. 104 (2000), 109-120.   CrossRef
  20. Z. Li and M. G. Ierapetritou: Production planning and scheduling integration through augmented lagrangian optimization. Comput. and Chemical Engrg. 34 (2010), 996-1006.   CrossRef
  21. D. G. Luenberger and Yinyu Ye: Linear and Nonlinear Programming. Third edition. Internat. Ser. Oper. Res. Management Sci. Springer, New York 2008.   CrossRef
  22. R. Schmidt: Eine modifizierte standard Einbettung zur Behandlung von Gleichungs und Ungleichungs Restriktionen. Master's Thesis, Humboldt Universitaet zu Berlin 2000.   CrossRef