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:20241120T082410Z
LOCATION:HG F 30 Audi Max
DTSTART;TZID=Europe/Stockholm:20240604T100100
DTEND;TZID=Europe/Stockholm:20240604T100200
UID:submissions.pasc-conference.org_PASC24_sess158_pos107@linklings.com
SUMMARY:P50 - Towards Linear-Scaling Density Functional Theory on Real Spa
 ce Grids
DESCRIPTION:Poster\n\nPaul F. Baumeister and Shigeru Tsukamoto (Forschungs
 zentrum Jülich)\n\nDesign structures of semiconductor circuits have shrunk
  to the lengthscale of a few nanometers. Despite that, systems so far have
  been too large to predict the electronic structure of realistic nano devi
 ces with an atom model description as accurate as density functional theor
 y (DFT). DFT eigenvalue problems leads to an unaffordable cubic scaling be
 haviour of the total workload, no matter how smart the diagonalization alg
 orithm is. Density matrix-based DFT algorithms allow for linear-scaling an
 d hence millions of atoms, however, they require a band-gapped system, i.e
 . conducting leads are problematic. Green function based DFT allows for bo
 th, metallic systems and linear-scaling due to truncation. In this work, G
 reen function DFT is expanded to real-space grids to be able to achieve an
  accuracy comparable to that of a plane-wave basis. The core of this algor
 ithm is a GPU-accelerated implicit Hamiltonian operator that is applied re
 peatedly to find the grid-resolved Green function by using an iterative re
 sidual minimization technique. Key to high performance are reduced GPU-mem
 ory bandwidth requirements of the implicit Hamiltonian.\nAn important ingr
 edient are factorizable projector functions for the pseudopotential that a
 re computed on the fly. We show details of the CUDA-C++ implementation and
  first performance numbers.\n\nSession Chair: Iva Kavcic (Met Office)
END:VEVENT
END:VCALENDAR