Information interference in stochastic systems with correlated noises

by Andrea Marchetti

Thursday 2 Apr 2026, 16:45 → 17:00 Europe/Rome

1/1-3 - Aula B (Dipartimento di Fisica e Astronomia - Edificio Marzolo)

Real-world systems are shaped by both internal interactions and fluctuating environments. When reconstructing network interactions from experimental data, unobserved environmental noise is often mistaken for true internal coupling. While information theory can help disentangle these effects, physical complexities often trigger "information interference", where environmental and internal contributions become inextricably linked. Here, we investigate an overlooked source of such interference in stochastic systems: correlated noises. We show that, in linearised models, non-diagonal diffusivity matrices induce effective couplings that can either amplify or mask deterministic interactions. We further investigate the combined effects of correlated noise and switching environments, identifying regimes in which internal and environmental influences can — or cannot — be disentangled. Finally, we apply our framework to hydrodynamically coupled systems and chemical reaction networks, illustrating its practical relevance.