论文部分内容阅读
本文根据Ｈｕｒｓｔ系数估计值ｋ的特征，用蒙特卡洛方法建立了ｋ的经验分布表，通过观测序列的ｋ值与表中相应的值进行比较以判别序列的长期相关性．此外，本文还借助于ｂｏｏｔｓｔｒａｐ法提出了长期相关性识别的非参数法．使用这两种方法所得出的结论是一致的，其有效性通过实例得到了验证．
In this paper, based on the characteristics of Hurst coefficient estimation k, the empirical distribution table of k is established by Monte Carlo method. The k value of the observation sequence is compared with the corresponding value in the table to determine the long-term correlation of the sequence. In addition, this paper also proposed a non-parametric method of long-term correlation recognition by means of bootstrap method. The conclusions reached by these two methods are consistent and their validity is verified by examples.