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:20190719T085744Z
LOCATION:HG F 1
DTSTART;TZID=Europe/Stockholm:20190613T181500
DTEND;TZID=Europe/Stockholm:20190613T184500
UID:submissions.pasc-conference.org_PASC19_sess140_msa260@linklings.com
SUMMARY:Hybrid Molecular Simulations with LAMMPS
DESCRIPTION:Minisymposium\nComputer Science and Applied Mathematics, Chemi
 stry and Materials, Physics, Engineering\n\nHybrid Molecular Simulations w
 ith LAMMPS\n\nKnight\n\nThe landscape of computational hardware and progra
 mming models has become quite diverse and with exascale computing on the h
 orizon it is now more important than ever to understand functionality and 
 performance limitations of scientific applications today for tomorrow&rsqu
 o;s potential use cases. There is an clear desire by many to write perform
 ance portable code to accelerate scientific discovery in this time of arch
 itectural diversity (CPUs, GPUs, FPGAs, vendors, etc…), but the opt
 imal path forward for established legacy and community scientific applicat
 ions is not always straightforward. For the LAMMPS molecular simulation co
 de, I’ll first discuss recent contributions to improve functionality
  and enable truly large-scale simulations for some physics models on ~786K
  MPI ranks. Next, I’ll discuss progress on developing computationall
 y efficient algorithms for simulations using a hybrid Monte Carlo/Molecula
 r Dynamics (MC/MD) method in LAMMPS to model interactions of ultra-intense
  x-ray free-electron pulses with matter. Motivated by the science question
 s being asked today and the computational challenges unique to this type o
 f simulation I’ll describe a tailored force-decomposition paralleliz
 ation strategy implemented within LAMMPS. Relative performance and general
  observations from mini-app implementations of the new particle force calc
 ulation in these MC/MD x-ray simulations will be presented spanning multip
 le programming models and hardware choices.
END:VEVENT
END:VCALENDAR