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:20230831T095746Z
LOCATION:Sertig
DTSTART;TZID=Europe/Stockholm:20230627T170000
DTEND;TZID=Europe/Stockholm:20230627T173000
UID:submissions.pasc-conference.org_PASC23_sess170_msa245@linklings.com
SUMMARY:Faster Climate Sciences - from DSLs to AI
DESCRIPTION:Minisymposium\n\nTorsten Hoefler and Huang Langwen (ETH Zurich
 )\n\nAs machine learning applications continue to succeed, the industry is
  shifting towards optimizing AI workloads. In this paper, we make three pr
 edictions about how this will impact high-performance computing (HPC) tech
 nology. We then identify challenges and opportunities for future weather a
 nd climate simulations based on these predictions. We explore various impl
 ementation options, from purely data-driven modeling to hybrid approaches 
 that combine data-driven and model-driven simulations. Finally, we propose
  a strategy for massive data compression using a deep learning model overf
 itted to climate data. These approaches will enable weather and climate sc
 ientists to confidently embrace the future of AI-based accelerator archite
 ctures.\n\nDomain: Climate, Weather and Earth Sciences\n\nSession Chair: K
 arthik Kashinath (NVIDIA Inc., Lawrence Berkeley National Laboratory)
END:VEVENT
END:VCALENDAR