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:20260611T145141Z
LOCATION:Bldg. 8 - Room B 101
DTSTART;TZID=Europe/Stockholm:20260701T090000
DTEND;TZID=Europe/Stockholm:20260701T093000
UID:submissions.pasc-conference.org_PASC26_sess155_msa187@linklings.com
SUMMARY:Oceananigans.jl: GPU-Accelerated Ocean Modeling through Julia-Enab
 led HPC
DESCRIPTION:Simone Silvestri (Politecnico di Torino, Massachusetts Institu
 te of Technology); Gregory Wagner (Aeolus Labs, Massachusetts Institute of
  Technology); Navid Constantinou (University of Melbourne); Valentin Chura
 vy (University of Augsburg); and Raffaele Ferrari (Massachusetts Institute
  of Technology)\n\nAccurate ocean modeling is critical for climate project
 ions, yet turbulent motions governing ocean transport remain unresolved in
  most climate models, approximated instead through empirical parameterizat
 ions, a persistent source of bias. Resolving these dynamics globally has b
 een computationally out of reach, but GPU computing and high-level scienti
 fic programming are changing this. Oceananigans.jl is an open-source ocean
  model written entirely in Julia, with a hydrostatic dynamical core featur
 ing GPU-optimized algorithms and novel WENO-based advection schemes for oc
 eanic turbulence. Julia's multiple dispatch and composable abstractions en
 able expressive code that compiles to efficient GPU kernels, while KernelA
 bstractions.jl provides portability across NVIDIA, AMD, and Intel GPUs. On
  64 NVIDIA A100 GPUs (16 nodes), Oceananigans simulates a decade of near-g
 lobal ocean dynamics per wall-clock day at 8 km resolution, sufficient to 
 resolve turbulent features that current climate models parameterize. At ~1
 0 km it reaches 10 simulated years per day on 68 GPUs, and at an unprecede
 nted 488 m resolution, 15 simulated days per day on 768 GPUs. These benchm
 arks use a fraction of current supercomputing resources, suggesting that l
 arge ensembles of turbulence-resolving simulations are now feasible. By re
 moving empirical ocean parameterizations, Oceananigans opens a path toward
  sharper climate projections with quantified uncertainty.\n\nDomain: Clima
 te, Weather, and Earth Sciences, Physics, Computational Methods and Applie
 d Mathematics\n\nSession Chairs: Ludovic Raess (University of Lausanne, ET
 H Zurich) and Samuel Omlin (ETH Zurich / CSCS)\n\n
END:VEVENT
END:VCALENDAR