BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20241120T082409Z
LOCATION:HG E 3
DTSTART;TZID=Europe/Stockholm:20240605T120000
DTEND;TZID=Europe/Stockholm:20240605T123000
UID:submissions.pasc-conference.org_PASC24_sess132_msa228@linklings.com
SUMMARY:GPU4GEO: Frontier GPU Multiphysics Solvers Using Julia
DESCRIPTION:Minisymposium\n\nAlbert de Montserrat and Ivan Utkin (ETH Zuri
 ch), Ludovic Räss (University of Lausanne), and Boris Kaus (Johannes Guten
 berg University Mainz)\n\nThe GPU4GEO project aims at developing new High-
 Performance Computing (HPC) tools for modelling geodynamics and ice sheet 
 dynamics written in the Julia language. This initiative is a response to t
 he practical demands of HPC, particularly the need for optimal performance
  in supercomputing environments that rely on GPU accelerators. We will pre
 sent our flagship applications JustRelax.jl (geodynamics) and FastIce.jl (
 ice flow). These applications offer a high-level API for massively paralle
 l thermo-mechanical Stokes solvers based on the highly-scalable pseudo-tra
 nsient iterative method. We will further discuss the developed software up
 on which these applications are built: (i) portability to multi-GPU system
 s (ParallelStencil.jl and ImplicitGlobalGrid.jl); (ii) solver-agnostic mat
 erial physics computations (GeoParams.j); and (iii) particles-in-cell adve
 ction (JustPIC.jl).\nWe tackle the increasing demand for merging data-driv
 en workflows with physics-based modelling, utilising Julia’s native suppor
 t for differentiable programming. Leveraging automatic differentiation (AD
 ), we efficiently compute model sensitivities, offering a unified framewor
 k for both inverse modelling and physics-informed machine learning. We wil
 l demonstrate Julia’s powerful AD application via Enzyme.jl for computing 
 adjoint sensitivities in our solvers, and present benchmarks showcasing mu
 lti-GPU performance and scalability.\n\nDomain: Climate, Weather, and Eart
 h Sciences, Physics, Computational Methods and Applied Mathematics\n\nSess
 ion Chairs: Ludovic Raess (University of Lausanne, ETH Zurich); Samuel Oml
 in (ETH Zurich / CSCS); and Michael Schlottke-Lakemper (University of Augs
 burg)
END:VEVENT
END:VCALENDAR