Armando Bazzani, "The modelling problem of complex systems evolution: from big data analysis to stochastic dynamical systems"

Thursday 13 Jun 2024, 15:00 → 16:00 Europe/Rome

Rectorate/Building-Auditorium (GSSI)

The complexity concept has been considered in Physics since big data sets started to be collected on economic, biological and social systems and the researchers had to cope with the problem of highlighting the relevant hidden information. One of the main results of

Complex System Physics is the development of complex network theory to define features that characterize the complex nature of the considered system. The application of the Statistical Physics methods allowed to study the equilibrium states of complex systems extending the thermodynamics formalism, but the attempt to formulate a non-equilibrium Statistical Physics is still an open problem. The intrinsic incomplete knowledge of

a complex system implies the necessity of using effective stochastic models (not derived from the physical laws) and the characterization of non-equilibrium stationary states using the entropy concept does not seem to suggest universal behaviours. Moreover, the recent success of the deep learning neural networks in the features detection using big data set has opened a research line that substitutes the classical dynamical systems models

in many applications beyond the complex systems, like the meteorological forecasting or material science.

In this seminar we illustrate as the Statistical Physics can explain the stationary distributions of human mobility computed using large data sets provided by the Information and Communication Technologies. We discuss on the possible application of simple stochastic models to understand the congestion formation on a road network, and we show as delay differential equation can be used to compute the existence of periodic

stationary solution for dynamical systems on graphs.

 

Zoom Link:
https://us02web.zoom.us/j/86086663267?pwd=ZzRBNTI0bUIzTU9ReHVtdlhXblljZz09

ID riunione: 860 8666 3267
Codice d’accesso: 965677