BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Deep Learning for the Closure of Partial Differential Equation Mod
 els - Justin Sirignano (University of Oxford)
DTSTART:20211115T143000Z
DTEND:20211115T150000Z
UID:TALK165388@talks.cam.ac.uk
DESCRIPTION:Although the exact physics equations for an application may be
  available\, numerically solving these equations can be computationally in
 tractable. The exact physics typically involve complex phenomena at small 
 scales\, which can require an infeasibly large computational grid to accur
 ately resolve. An example is turbulence\, which is relevant to modeling ai
 rplanes\, biomedical technology\, and power generation. Large-eddy simulat
 ion (LES) is a reduced-order PDE model for the low frequencies of the Navi
 er-Stokes equations for turbulent flows. By modeling only the low frequenc
 ies\, the LES equations can be solved at a low computational cost on a coa
 rse grid. However\, the LES equations introduce an unclosed term which mus
 t be modeled. We develop a deep learning closure model for LES. The "deep 
 learning LES model" is calibrated to high-fidelity data. Training uses adj
 oint PDEs to optimize over the full nonlinearity of the PDE model. The app
 roach is implemented for isotropic turbulence\, turbulent jet flows\, and 
 turbulent wakes.
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR