BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Data-Driven Constitutive Modelling for Non-Equilibrium Systems - D
 r. Shenglin Huang\,  Sheffield University 
DTSTART:20260206T140000Z
DTEND:20260206T150000Z
UID:TALK244297@talks.cam.ac.uk
CONTACT:Elizabeth Howard
DESCRIPTION:Non-equilibrium phenomena are ubiquitous across material syste
 ms and of great technological relevance. Examples of such phenomena includ
 e diffusion processes in liquid and gases\, viscoelasticity and plasticity
  in solids\, and rheological behavior of colloidal and granular media. Des
 pite their ubiquity and importance\, the understanding of non-equilibrium 
 phenomena remains in its infancy compared with classical equilibrium therm
 odynamics and statistical mechanics from both theoretical and computationa
 l aspects. As a consequence\, current modeling and simulation strategies\,
  including multiscale paradigms\, are mostly trapped within a compromise b
 etween computational efficiency and physical fidelity.\n\nIn this talk\, I
  present a data-driven framework for constitutive modeling of non-equilibr
 ium systems that integrates continuum mechanics\, statistical physics\, an
 d machine learning. The first part introduces a machine learning architect
 ure called Variational Onsager Neural Networks (VONNs) to learn thermodyna
 mically consistent non-equilibrium evolution PDEs from macroscopic materia
 l responses. The second part focuses on Statistical-Physics-Informed Neura
 l Networks (Stat-PINNs)\, a multiscale learning framework for inferring co
 arse-grained dissipative evolution equations from stochastic particle dyna
 mics by leveraging that fluctuation-dissipation relations. Together\, thes
 e approaches provide a physically grounded and computationally efficient p
 athway for learning constitutive laws of non-equilibrium systems across sc
 ales.
LOCATION:Oatley 1 Meeting Room\, Department of Engineering
END:VEVENT
END:VCALENDAR