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:20210916T132456Z
LOCATION:
DTSTART;TZID=Europe/Stockholm:20210706T173000
DTEND;TZID=Europe/Stockholm:20210706T190000
UID:submissions.pasc-conference.org_PASC21_sess182_post180@linklings.com
SUMMARY:P40 - MaMiCo: Fast Non-Local Means Filtering for Molecular-Continu
 um HPC Flow Simulation
DESCRIPTION:Poster\n\nP40 - MaMiCo: Fast Non-Local Means Filtering for Mol
 ecular-Continuum HPC Flow Simulation\n\nJarmatz, Neumann\n\nThe Non-Local 
 Means (NLM) algorithm is an advanced denoising method that was developed f
 or image processing and is used in the context of 2D and 3D medical imagin
 g. In the area of multiscale computational fluid dynamics, coupled molecul
 ar-continuum flow simulations are often impaired by thermal noise fluctuat
 ions. Here we propose a modified variant of NLM filtering to enhance the q
 uality of transient flow data sampled from a molecular dynamics (MD) syste
 m in that coupling. Our approach outperforms conventional filtering method
 s. Due to the inherent non-local nature of the NLM method, one critical is
 sue in this context can be the challenging computational demand in some ca
 ses, as well as the scalability on HPC systems. We thus  present an o
 ptimized implementation of our method, utilizing the Macro-Micro-Coupling 
 tool (MaMiCo). MaMiCo is a flexible framework written in C++ that can be u
 sed to couple arbitrary MD and continuum codes. Our NLM filter for MaMiCo 
 is designed for massively-parallel distributed memory computations. It com
 bines an excellent gain, in terms of signal-to-noise ratio of fluctuating 
 data, with a performance that constitutes just an insignificant overhead o
 ver unfiltered MD simulation.
END:VEVENT
END:VCALENDAR
