Student: Subhrajit Sinha
Advisor: Umesh Vaidya
Title: Information Transfer and Causality in Dynamical Systems
Abstract: Causality and influence characterization is a problem of immense importance in many different fields of research, like economics, social science, neuroscience, finance, etc. However, there is no universally accepted definition of causality, and people use different definitions, according to the application. For example, in economics, Granger causality is used to infer causality. Other measures use information theoretic ideas to define a measure of causality, like Transfer Entropy, Directed Information and Liang Kleeman Information Transfer. But all these definitions have severe drawbacks and give erroneous results, even for very simple systems. In this work, we provide a new definition of causality and show how this measure overcomes the deficiencies of the previous measures.