离散时间序列的网络模体分析

The study of network motifs induced from discrete time series

时间序列可以被转换成网络的形式,复杂网络理论也因此可以用于刻画时间序列的时域和相空间特性.本文针对可视图算法和相空间重构算法这两种时间序列的转换算法,研究了它们的伴生网络在倍周期分岔和混沌等各种类型时间序列的模体分布特征,分析了这两种算法各自的优点.

Complex network theory is used to characterize the temporal and phase space features of a time series when it is transferred into a network. In this paper, we study the motif ranks of complex networks induced from different cat egories of time series with periodic bifurcations and chaos, which are generated with two algorithms： the Visiblity Graph （VG） algorithm and the Phase-space Reconstruction （PR） algorithm. The advantages of both algorithms are analyzed.