摘要
考虑由扰动数据重构原函数的导数问题.基于L-广义解正则化理论,提出了一个新的磨光方法的框架.给出一个具体的求解前3阶导数的算法,其中正则化策略选择了一种改进的TSVD(truncated singular value decomposition)方法(典则TSVD方法).数值结果进一步验证了理论结果及新方法的有效性.
The problem of reconstructing numerical derivatives from noisy data is considered. A new framework of mollification methods based on L-generalized solution regularization methods was proposed. A concrete algorithm for the first three derivatives was presented, in which a modification of （called cTSVD （ canonical truncated singular value decomposition） ） is chosen as the needed regularization technique. The numerical examples given verify the theoretical results and show the efficiency of the new method.
出处
《应用数学和力学》
CSCD
北大核心
2008年第6期696-704,共9页
Applied Mathematics and Mechanics
