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:20250822T115805Z
LOCATION:Campussaal - Plenary Room
DTSTART;TZID=Europe/Stockholm:20250617T103000
DTEND;TZID=Europe/Stockholm:20250617T110000
UID:submissions.pasc-conference.org_PASC25_sess150_pos137@linklings.com
SUMMARY:P35 - Performance Portability Across Different Mathematical Models
 , Hardware, and Simulation Scenarios in Molecular Dynamics
DESCRIPTION:Samuel James Newcome, Fabio Alexander Gratl, Manish Kumar Mish
 ra, Markus Mühlhäußer, Jonas Schumacher, and Hans-Joachim Bungartz (Techni
 cal University of Munich)\n\nDue to the importance of Molecular Dynamics s
 imulations within fields such as thermodynamics, numerous methods have bee
 n developed to speedup the force calculations, which typically dominate th
 e runtime. None of these methods are, however, optimal for every molecular
  model, on every hardware, and for every distribution of molecules. For no
 n-HPC-expert simulation developers, choosing and then implementing the bes
 t method for their particular simulation is a challenging task. Furthermor
 e, different regions of the simulation can have different molecule distrib
 utions with different optimal methods, and these can change as the distrib
 ution changes. A solution to this problem, AutoPas, is an open-source, C++
 17, particle simulation library, which can be used to build a particle sim
 ulator that automatically selects and tunes the optimal algorithmic config
 uration from its internal library. It can optimise for either time or ener
 gy. This poster focuses on efforts to improve its performance with complex
  multi-site molecular models, where the amount of data required per molecu
 le may vary, as well as algorithm selection methods that use data-driven a
 nd expert-knowledge-based approaches to reduce the overhead of its selecti
 on process. This poster highlights the practicalities of such approaches b
 eing used by non-HPC expert users.\n\nSession Chair: David Moxey (King's C
 ollege London)\n\n
END:VEVENT
END:VCALENDAR