求解大规模线性离散不适定问题的Arnoldi—Fractional Tikhonov正则化算法
2016-03-22张慧
张慧
摘要:随着科技发展,不适定问题出现在地球物理等多种领域。正则化方法是求解此类问题近似解的有效算法。该文将Fractional Tikhonov正则化算法应用于投影算法,提出求解大规模线性离散不适定问题的Arnoldi-Fractional Tikhonov正则化算法。并进一步提出广义Arnoldi-Fractional Tikhonov正则化算法。最后,论文对所提出的算法编写程序进行数值试验比较。结果表明新算法是有效且具有优势的。
关键词:不适定问题;正则化方法;Fractional Tikhonov正则化方法;Arnoldi-Fractional Tikhonov正则化算法
中图分类号:TP311 文献标识码:A 文章编号:1009-3044(2016)02-0236-03
Abstract:With the development of the technology ,the ill-posed problems widely arise in many areas such as geophysics and so on. In this thesis, An Arnoldi-Fractional Tikhonov regularization method for large scale linear discrete ill-posed problems is presented via applying the Fractional Tikhonov regularization to the projection algorithm. Further more, the generalized Arnoldi-Fractional Tikhonov method are proposed in the follows. At last, this thesis conducts numerous classical numerical experiments on the improved methods proposed above. Numerical experiments and comparisons indicate that the new improved regularization methods are feasible and efficient.
Key words: ill-posed problems; regularization methods; the Fractional Tikhonov regularization method; the Arnoldi-Fractional Tikhonov regularization method
1 引言及主要结论
由表3的结果知,GAFT的计算结果精度比GAT的计算结果高,但是参数[α]的选取会直接影响到求解精度,并且不同矩阵所取的最合适的[α]不一样。
根据以上三组数值试验的结果,可以看出:在Fractional Tikhonov正则化方法基础上给出的Arnoldi-Fractional Tikhonov和Generalized-Arnoldi- Fractional Tikhonov正则化方法在求解大规模线性离散不适定问题时是具有一定优势的,计算结果的精度更高一些。
5 结束语
对于大规模不适定问题的求解难度在于其系数矩阵的奇异值分解计算量过大。首先将其投影到小规模子空间上,再对投影问题的求解采用Fractional Tikhonov正则化方法,并推广到广义情形。数值试验表明新方法是具有一定优势的。但其中参数[α]的值影响解的精度,如何确定合适的[α]是下一步需要解决的问题。
参考文献:
[1] 黄光远,刘小军.数学物理反问题[M].济南:山东科学技术出版社,1993.
[2] Kirsch A.An Introduction to the Mathematical Theory of Inverse Problems[M].New York:Springer- Verlag, 1996.
[3] 王彦飞.反演问题的计算方法及其应用[M].北京:高等教育出版社,2007.
[4] Tikhonov A N.Solution of incorrectly formulated problems and the regularization method[J]. Soviet. Math. Dokl.,1963(4):1035-1038.
[5] Hochstenbach M E, Reichel. L. Fractional Tikhonov regularization for linear discrete ill-posed problems[J]. BIT, 2011(51):197-215.
[6] Lewis B,Reichel L.Arnoldi-Tikhonov regularization methods[J].J.Comput.Appl.Math., 2009,226: 92-102.
[7] Wilkinson J H. The Algebraic Eigenvalue problem[J].Oxford University Press., 1965:104-108.
[8] Reichel L,Sgallari F,Ye Q.Tikhonov regularization based on generalized Krylov subspace methods[J]. Applied Numerical Mathematics.,2012(62):1215-1228.
[9] 肖庭延,于渗根,王彦飞.反问题数值解法[M].北京:科学出版社,2003.
[10] Morikuni K, Reichel L,Hayami K. FGMRES for linear discrete ill-posed problems[J]. Appl. Numer. Math.,2014(75): 175-187.