Berlin 2012 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

DY: Fachverband Dynamik und Statistische Physik

DY 22: Posters I

DY 22.3: Poster

Wednesday, March 28, 2012, 17:00–19:00, Poster A

An entropy based model-free definition of causal coupling strength for multivariate time series — •Jakob Runge^1,2, Jobst Heitzig¹, and Jürgen Kurths¹ — ¹Potsdam Institute for Climate Impact Research, P.O. Box 60 12 03, D--14412 Potsdam, Germany — ²Department of Physics, Humboldt University, Newtonstr. 15, D--12489 Berlin, Germany

We discuss the problem of defining a causal coupling strength between components of a multivariate time series if no underlying model is assumed. This is in contrast to methods like cross-correlation or granger causality. The method is based on conditional mutual information and utilizes the concept of graphical models. We investigate how our measure relates to model systems where a coupling strength is known and discuss its limitations.