Lab Home | Phone | Search
	Center for Nonlinear Studies
	Home  ▶ People   Current   Executive Committee   Postdocs   Visitors   Students   Research   Publications  ▶ Conferences   Workshops   Sponsorship  ▶ Talks   Seminars   Postdoc Seminars Archive   Quantum Lunch   Quantum Lunch Archive   P/T Colloquia   Archive   Ulam Scholar     Postdoc Nominations  ▶ Student Requests   Student Program  ▶ Visitor Requests   Description   Past Visitors  ▶ Services   General     History of CNLS     Maps, Directions   CNLS Office   T-Division   LANL				Thursday, August 24, 20232:00 PM - 2:45 PMHybrid - CNLS Conference Room (TA-3, Bldg 1690) and WebEx Seminar Physics constrained neural networks for learning a lattice Boltzmann collisional operator Alessandro GabbanaEindhoven University of Technology, The Netherlands In this talk we discuss the possibility of learning collision operators for the Lattice Boltzmann Method (LBM) using a deep learning approach. We present results in which a neural network is successfully trained as a surrogate of the single relaxation time BGK operator. We compare the accuracy achieved in the simulation of a few selected benchmarks, employing several approaches for the architecture of the neural network. We show that only by embedding in the neural network physics properties, such as conservation laws and symmetries, it is possible to correctly reproduce the time dynamic of simple fluid flows. Moreover, we present a framework which allows coupling LBM with Direct simulation Monte Carlo (DSMC), paving the way for learning more complex collisional operators, aiming at providing a reliable and computationally efficient description of flows in rarefied conditions. Host: Vitaliy Gyrya (T-5) and Daniel Livescu (CCS-2)