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:20210916T132446Z
LOCATION:Henry Dunant
DTSTART;TZID=Europe/Stockholm:20210708T153000
DTEND;TZID=Europe/Stockholm:20210708T160000
UID:submissions.pasc-conference.org_PASC21_sess205_pap120@linklings.com
SUMMARY:Fast Scalable Implicit Solver with Convergence of Equation-Based M
 odeling and Data-Driven Learning: Earthquake City Simulation on Low-Order 
 Unstructured Finite Element
DESCRIPTION:Paper\n\nFast Scalable Implicit Solver with Convergence of Equ
 ation-Based Modeling and Data-Driven Learning: Earthquake City Simulation 
 on Low-Order Unstructured Finite Element\n\nIchimura, Fujita, Koyama, Kiku
 chi, Kusakabe...\n\nWe developed a new approach in converging equation-bas
 ed modeling and data-driven learning on high-performance computing resourc
 es to accelerate physics-based earthquake simulations. Here, data-driven l
 earning based on data generated while conducting equation-based modeling w
 as used to accelerate the convergence process of an implicit low-order uns
 tructured finite-element solver. This process involved a suitable combinat
 ion of data-driven learning for estimating high-frequency components and c
 oarsened equation-based models for estimating low-frequency components of 
 the problem. The developed solver achieved a 12.8-fold speedup over the st
 ate-of-art solver with a 96.4% size-up scalability up to 24,576 nodes (98,
 304 MPI processes x 12 OpenMP threads = 1,179,648 CPU cores) of Fugaku wit
 h 126,581,788,413 degrees-of-freedom, leading to solving a huge city earth
 quake shaking analysis in a 10.1-fold shorter time than the previous state
 -of-the-art solver. Furthermore, to show that the developed method attains
  high performance on variety of systems with small implementation costs, w
 e ported the developed method to recent GPU systems by use of directive ba
 sed methods (OpenACC). The equation based modeling and the data-driven lea
 rning are of utterly different characteristics, and hence they are rarely 
 combined. The developed approach of combining them is effective, and remar
 kable results mentioned above are achieved.\n\nDomain: Solid Earth Dynamic
 s, Engineering
END:VEVENT
END:VCALENDAR