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:20260624T171340Z
LOCATION:Bldg. 8 - Entrance Hall
DTSTART;TZID=Europe/Stockholm:20260630T173000
DTEND;TZID=Europe/Stockholm:20260630T194500
UID:submissions.pasc-conference.org_PASC26_sess135_posC112@linklings.com
SUMMARY:ACMP01 - Adaptive Multidimensional Quadrature on Multi-GPU Systems
DESCRIPTION:Melanie Tonarelli (Università della Svizzera Italiana)\n\nThe 
 present work introduces a distributed adaptive deterministic quadrature fr
 amework for high-dimensional integration on multi-GPU systems, enabling ac
 curate evaluation of integrals arising in applications such as radiative t
 ransfer and probabilistic design. The method is formulated as a hierarchic
 al domain decomposition: the integration domain is recursively subdivided 
 and, at each iteration, all subdomains contributing significantly to the g
 lobal error are refined in parallel. As each GPU progresses independently 
 through the adaptive refinement, workload imbalance naturally arises. To a
 ddress this, a decentralised redistribution scheme based on a cyclic round
 -robin policy is employed. This strategy dynamically rebalances the worklo
 ad across devices through non-blocking, CUDA-aware MPI communication that 
 overlaps with computation. The proposed approach preserves the strict erro
 r guarantees of deterministic quadrature while enabling scalable execution
  on modern HPC platforms. Extensive benchmarks on oscillatory, sharply pea
 ked, and discontinuous test functions demonstrate improved robustness and 
 efficiency compared to a state-of-the-art GPU integration package, particu
 larly at high accuracy and in large dimensions. Using multiple GPUs overco
 mes single-device memory limitations and enables integration of up to 11 d
 imensions, while reducing runtimes for strict accuracy targets.\n\n
END:VEVENT
END:VCALENDAR