Detecting causality from observations of dynamic variables.
Rosebrock, Daniel.
2015
-
Abstract: ``Granger causality" is a methodology, first proposed in 1969 by Clive Granger, for detecting causal relationships between two stochastically fluctuating quantities. I will present several equivalent ways of describing this methodology. In Granger's original work, it was assumed that the stochastic quantities are discretized Ornstein-Uhlenbeck processes, that is, AR(1) processes. Thus ... read morethe deterministic piece of the dynamics is simply relaxation into a stable steady state. If the deterministic dynamics are more complicated, Granger causality is not applicable, but the idea of one of the equivalent versions that I present has a relatively straightforward generalization. I present numerical experiments testing the application of this idea to noisy oscillators that are coupled in one direction or the other, with the goal of deducing the direction of the coupling from observed solutions.
Thesis (M.S.)--Tufts University, 2015.
Submitted to the Dept. of Mathematics.
Advisor: Christoph Borgers.
Committee: Bruce Boghosian, and Zbigniew Nitecki.
Keywords: Mathematics, Neurosciences, and Nuclear chemistry.read less - ID:
- cr56nc365
- Component ID:
- tufts:21516
- To Cite:
- TARC Citation Guide EndNote