Parameterize the Center of the Trust Region for Solving Quadratic Interpolation Models

Abstract:

The derivative free trust region algorithm was considered for solving the unconstrained optimization problems. This paper introduces a novel methodology that modified the center of the trust region in order to improve the search region. The main idea is parameterizing the center of the trust region based on the ideas of multi-directional search and simplex search algorithms. The scope of the new region was so expanded by introducing a parameter as to we can find a better descent directions. Experimental results reveal that the new method is more effective than the classic trust region method on the testing problems.

• Periodical: Applied Mechanics and Materials (Volumes 52-54)
• Pages: 920-925
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.52-54.920
• Online since: March 2011
• Authors: Qing Hua Zhou, Yan Geng, Ya Rui Zhang, Feng Xia Xu
• Keywords: Derivative Free Method, Quadratic Interpolation, Trust Region, Unconstrained Optimization

References

[1] M.J.D. Powell: On the Lagrange functions of quadratic models that are defined by interpolation, Optim. Methods Softw, 16, p.289–309 (2001).

[2] M.J.D. Powell: UOBYQA: Unconstrained optimization by quadratic approximation, Math. Program., Ser. B 92: 555-582 (2002).

DOI: 10.1007/s101070100290

[3] A.R. Conn, K. Scheinberg, and L.N. Vicente: Global convergence of general derivative–free trust-region algorithms to first-and second-order critical points, SIAM J. OPTIM., Vol. 20, No. 1, p.387–415 (2009).

DOI: 10.1137/060673424

[4] A.R. Conn, K. Scheinberg, and L. N. Vicente: Geometry of interpolation sets in derivative free optimization, Math. Program., 111, p.141–172 (2008).

DOI: 10.1007/s10107-006-0073-5

[5] Q.H. Zhou: On the use of simplex methods in constructing quadratic models, Science in China Series A: Mathematics, 50(7), pp.913-924 (2007).

DOI: 10.1007/s11425-007-0054-z

[6] Q.H. Zhou: A New Algorithm on quadratic interpolation models, Chinese journal of Engineering Mathematics 23(6), pp.1075-1087 (2006).

[7] M.J.D. Powell: On the use of quadratic models in unconstrained minimization without derivatives, Optimization Methods and Software, 19, pp.399-411 (2004).

DOI: 10.1080/10556780410001661450

[8] T. G. Kolda, R. M. Lewis, and V. Torczon: Optimization by direct search: New perspectives on some classical and modern methods, SIAM Rev., 45, p.385–482 (2003).

DOI: 10.1137/s003614450242889

[9] V. Torczon: On the convergence of pattern search algorithms, SIAM J. Optim., 7, p.1–25(1997).

[10] C. Audet and J. E. Denni: Analysis of generalized pattern searches, SIAM J. Optim., 13 , p.889–903 (2003).

DOI: 10.1137/s1052623400378742

[11] J.A. Nelder and R. Mead: A simplex method for function minimization, The computer Journal, 7(4): 308-313 (1965).

DOI: 10.1093/comjnl/7.4.308

[12] V.J. Torczon: Multi-directional Search: A direct search algorithm for parallel machines, Ph. D Thesis, Rice University, Houston, Texas, USA (1989).

[13] J.J. Moré, B.S. Garbow, and K.E. Hillstorm: Testing unconstrained optimization software, ACM Transactions on Mathamatical Software, 7, pp.17-41 (1981).

DOI: 10.1145/355934.355936