两种平滑优化算法在一维情况下的比较研究被引量：3

Comparisons of Two Smoothing Algorithms for Optimization in One-Dimensional Space

下载PDF

导出

摘要平滑方法是全局优化的一种启发式方法 ,在分子构像优化问题中得到大量应用。扩散方程法和邻域平均法是平滑方法的两种实现方式。在一维情况下给出了邻域平均法的性质 ,通过数值实验比较了两种方法的性能。对类似Griewank函数的测试函数 ,发现邻域平均法的优化结果在总体上优于扩散方程法。 The smoothing method is a heuristic method for global optimization,which is much applicable in the research area of molecular conformations.The diffusion equation method(DEM)and the neighborhood averaging method(NAM)are two kinds of implementations of the smoothing method.The properties of the neighborhood averaging method and the efficiency of the two implementations in one-dimensional space are given.For Griewank-like functions,the optimization results of NAM as a whole excel DEM.

作者崔鹏张立昂吉利久高立

机构地区北京大学信息科学技术学院北京大学数学学院

出处《北京大学学报（自然科学版）》 CAS CSCD 北大核心 2003年第5期681-689,共9页 Acta Scientiarum Naturalium Universitatis Pekinensis

基金 973项目"信息技术中的应用理论和高性能软件"资助项目 (G19980 30 4 0 3)

关键词平滑方法全局优化扩散方程法邻域平均法 Griewank函数 smoothing method global optimization diffusion equation method neighborhood averaging method Griewank function

分类号 TP301.6 [自动化与计算机技术—计算机系统结构]

引文网络
相关文献

参考文献13

1袁亚湘孙文瑜.最优化理论与方法[M].北京:科学出版社,2001..
2Yudin D B. Quantitative Analysis of Complex Systems I. Engineering Cybernetics, 1965,3 : 1 - 9.
3Rubinstein R. Monte Carlo Optimization, Simulation and Sensitivity of the Queueing Networks. New York:Wiley, 1986.194 - 206.
4Piela L, Kostrowicki J, Scheraga H A. The Multiple-minima Problem in the Conformational Analysis of Molecules:Deformation of the Potential Energy Hypersurface by the Diffusion Equation Method. J Phys Chem, 1989,93 : 3 339 -3346.
5Ma J, Hsu D, Straub J E. Approximate Solution of the Classical Liouville Equation Using Gaussian Phase Packet Dynamics:Application to Enhanced Equilibrium Averaging and Global Optimization. J Chem Phys, 1993,99:4 024-4 035.
6Coleman T,Wu Z. Parallel Continuation-based Global Optimization for Molecular Conformation and Protein Folding.Journal of Global Optimization, 1996,8:49 - 65.
7Andricioaei I, Straub J E. Global Optimization Using Bad Derivatives: Derivative-free Method for Molecular Energy Minimization .J Comput Chem, 1998,19 : 1 445 - 1 455.
8Pliner V. Metric Unidimensionl Scaling and Global Optimization. Journal of Classification, 1996,13 : 3 - 18.
9Yulle A L, Poggio T A. Scaling Theorems for Zero Crossings. IEEE Tran on Patt Anal and Mach Inte, 1986,8:15 -25.
10Babaud J,Witkin A P, Baudin M, et al. Uniqueness of the Gaussian Kernel for Scale-space Filtering. IEEE Tran on Patt Anal,and Math Inte,1986,8:26-33.