A spectral updating for the method of moving asymptotes

Abstract

A modified version of the method of moving asymptotes is proposed based on the spectral parameter used in the updating of a key parameter of the model. The second-order information present in the spectral parameter is thus included in the model functions that define the rational approximations. Numerical experiments indicate that the idea is promising in the sense that the cost–benefit of computing the spectral parameter is worth it for reducing the total effort of the algorithm when compared with the original version.

Keywords:

• constrained optimization
• method of moving asymptotes
• spectral parameter
• large scale problems
• global convergence

AMS Subject Classification :

• 49M37
• 65K05
• 90C30

Acknowledgements

This work is partially supported by FAPESP (06/52846-7) and CNPq (303465/2007-7, 309380/2006-5). The authors are thankful to the comments and suggestions of the referees, which helped to improve the first version of the manuscript.