Alessandro Laio
(SISSA)

The ability to distinguish between correlation and causation of variables in complex systems remains an interesting and open area of investigation. In this lecture we will discuss how causality can emerge in systems evolving according to a time-reversible dynamics and satisfying detailed balance, such as a molecular system simulated by molecular dynamics. We will first introduce the traditional approach to infer the presence of a putative causal link in time series, based on the estimate of the transfer entropy. We will then discuss how this approach can be extended to build a causal graph, synthesising causal relationship between many variables. Finally, we will discuss how one can computationally verify if a putative causal link, corresponding to a strongly asymmetric information transfer between two variables, corresponds to a genuine causal link, or is instead an artifact determined by an unobserved variable which causes the two observed variables.
 

Zoom registration link:
https://zoom.us/meeting/register/Bp9J-CW1TgaaxAyVPYhRCQ