BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Chicago
X-LIC-LOCATION:America/Chicago
BEGIN:DAYLIGHT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
TZNAME:CDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
TZNAME:CST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20230124T171525Z
LOCATION:C144-145
DTSTART;TZID=America/Chicago:20221115T110000
DTEND;TZID=America/Chicago:20221115T113000
UID:submissions.supercomputing.org_SC22_sess190_gb102@linklings.com
SUMMARY:Reshaping Geostatistical Modeling and Prediction for Extreme-Scale
Environmental Applications
DESCRIPTION:ACM Gordon Bell Finalist, Awards Presentation\n\nReshaping Geo
statistical Modeling and Prediction for Extreme-Scale Environmental Applic
ations\n\nCao, Abdulah, Alomairy, Pei, Nag...\n\nWe extend the capability
of space-time geostatistical modeling using algebraic approximations, illu
strating application-expected accuracy worthy of double precision from maj
ority low-precision computations and low-rank matrix approximations. We ex
ploit the mathematical structure of the dense covariance matrix whose inve
rse action and determinant are repeatedly required in Gaussian log-likelih
ood optimization. Geostatistics augments first-principles modeling approa
ches for the prediction of environmental phenomena given the availability
of measurements at a large number of locations; however, traditional Chole
sky-based approaches grow cubically in complexity, gating practical extens
ion to continental and global datasets now available. We combine the linea
r algebraic contributions of mixed-precision and low-rank computations wit
hin a tile-based Cholesky solver with on-demand casting of precisions and
dynamic runtime support from PaRSEC to orchestrate tasks and data movement
. Our adaptive approach scales on various systems and leverages the Fujits
u A64FX nodes of Fugaku to achieve up to 12X performance speedup against t
he highly optimized dense Cholesky implementation.\n\nSession Format: Reco
rded\n\nTag: Awards\n\nRegistration Category: Tech Program Reg Pass
END:VEVENT
END:VCALENDAR