News
From July 10th to 14th, Professor X. San Liang was invited to attend the EPS 2023 Sigma-Phi Statistical Physics Conference, Chania, Greece, followed by a visit to Technical University of Crete. At the conference meetings, he delivered two presentations.
In his first talk entitled "Causality as a Real Physical
Notion ab initio, and Quantitative Causality Analysis in Climate and
Environmental Sciences," Professor Liang emphasized the pivotal role of
causal inference in scientific research. He introduced his information flow and
quantitative causality analysis theory rigorously derived from first
principles. He showcased the applications in various disciplines such as
computer science, neuroscience, and quantum mechanics. In this talk, he not only put information flow and causality in the nutshell within the framework of statistical physics, establishing the its invariance upon arbitrary nonlinear coordinate transformation, and hence its intrinsic property reflecting the nature of physical world, but also turn the philosophical statement "correlation
does not imply causation, but causation implies correlation" into an explicit mathematical expressio. These theoretical results have been
put to application across disciplines, providing deep insights from
different perspectives, arousing wide interest from the audience.
In another presentation, entitled "Measuring the Importance of Individual Units to the Structural Integrity of a Complex Network," Professor Liang discussed another significant contribution of his work – the measurement of individual nodal contributions in a complex network. He presented two key findings: (1) For a given complex system, an individual node's contribution is quantified by its cumulative information flow across the entire network. Generally, this cumulative information flow is not equivalent to the sum of its information flows to all the nodes else, highlighting possible emergent collective properties of the network. (2) Counterintuitively, a node's importance is not identical to its degree of connectivity. In certain cases, the most critical nodes in a network are not the highly connected "hub" nodes. Professor Liang illustrated this theory using an ideal network composed of Stuart-Landau oscillators, and a real world example——the US stock market. Apart from applications in the financial domain, this research could also be employed for diagnosing neural disorders, controlling infectious diseases, identifying potential causes of urban traffic bottlenecks, pinpointing underlying reasons for power grid failures (such as the 2003 blackout event that affected most of North America), and constructing robust computer networks.