中国物理B ›› 2013, Vol. 22 ›› Issue (3): 38902-038902.doi: 10.1088/1674-1056/22/3/038902

• INTERDISCIPLINARY PHYSICS AND RELATED AREAS OF SCIENCE AND TECHNOLOGY • 上一篇    下一篇

Measuring causality by directional symbolic mutual information approach

陈贵, 谢磊, 褚健   

  1. Institute of Cyber-System and Control, Zhejiang University, Hangzhou 310027, China
  • 收稿日期:2012-09-05 修回日期:2012-10-24 出版日期:2013-02-01 发布日期:2013-02-01
  • 基金资助:
    Project supported by the National Natural Science Foundation of China (Grant No. 60904039).

Measuring causality by directional symbolic mutual information approach

Chen Gui (陈贵), Xie Lei (谢磊), Chu Jian (褚健)   

  1. Institute of Cyber-System and Control, Zhejiang University, Hangzhou 310027, China
  • Received:2012-09-05 Revised:2012-10-24 Online:2013-02-01 Published:2013-02-01
  • Contact: Chen Gui E-mail:gui.sam.chen@gmail.com
  • Supported by:
    Project supported by the National Natural Science Foundation of China (Grant No. 60904039).

摘要: We propose a novel measure to assess causality through the comparison of symbolic mutual information between the future of one random quantity and the past of the other. This provides a new perspective different from the conventional conceptions. Based on this point of view, a new causality index is derived that uses the definition of directional symbolic mutual information. This measure presents properties different from the time delayed mutual information since the symbolization captures the dynamic features of the analyzed time series. In addition to characterizing the direction and the amplitude of the information flow, it can also detect coupling delays. This method has the property of robustness, conceptual simplicity, and fast computational speed.

关键词: causality measure, Bandt and Pompe method, mutual information, transfer entropy

Abstract: We propose a novel measure to assess causality through the comparison of symbolic mutual information between the future of one random quantity and the past of the other. This provides a new perspective different from the conventional conceptions. Based on this point of view, a new causality index is derived that uses the definition of directional symbolic mutual information. This measure presents properties different from the time delayed mutual information since the symbolization captures the dynamic features of the analyzed time series. In addition to characterizing the direction and the amplitude of the information flow, it can also detect coupling delays. This method has the property of robustness, conceptual simplicity, and fast computational speed.

Key words: causality measure, Bandt and Pompe method, mutual information, transfer entropy

中图分类号:  (Entropy and other measures of information)

  • 89.70.Cf
02.50.-r (Probability theory, stochastic processes, and statistics) 05.45.Tp (Time series analysis) 87.19.lo (Information theory)