BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Denver
X-LIC-LOCATION:America/Denver
BEGIN:DAYLIGHT
TZOFFSETFROM:-0700
TZOFFSETTO:-0600
TZNAME:MDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0600
TZOFFSETTO:-0700
TZNAME:MST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260422T000713Z
LOCATION:301-302-303
DTSTART;TZID=America/Denver:20231115T113000
DTEND;TZID=America/Denver:20231115T120000
UID:submissions.supercomputing.org_SC23_sess176_pap292@linklings.com
SUMMARY:NNQS-Transformer: An Efficient and Scalable Neural Network Quantum
  States Approach for Ab Initio Quantum Chemistry
DESCRIPTION:Yangjun Wu (Institute of Computing Technology, Chinese Academy
  of Sciences); Chu Guo (Hunan Normal University); Yi Fan (University of Sc
 ience and Technology of China); Pengyu Zhou (Institute of Computing Techno
 logy, Chinese Academy of Sciences); and Honghui Shang (University of Scien
 ce and Technology of China)\n\nNeural network quantum state (NNQS) has eme
 rged as a promising candidate for quantum many-body problems, but its prac
 tical applications are often hindered by the high cost of sampling and loc
 al energy calculation.  We develop a high-performance NNQS method for ab i
 nitio electronic structure calculations. The major innovations include: \n
 \n   (1)  A transformer based architecture as the quantum wave function an
 satz; \n\n   (2)  A data-centric parallelization scheme for the variationa
 l Monte Carlo (VMC) algorithm which preserves data locality and well adapt
 s for different computing architectures; \n\n   (3)  A parallel batch samp
 ling strategy which reduces the sampling cost and achieves good load balan
 ce; \n\n   (4)  A parallel local energy evaluation scheme which is both me
 mory and computationally efficient; \n\n   (5)  Study of real chemical sys
 tems demonstrates both the superior accuracy of our method compared to sta
 te-of-the-art and the strong and weak scalability for large molecular syst
 ems with up to 120 spin orbitals.\n\nTag: Applications, Modeling and Simul
 ation\n\nRegistration Category: Tech Program Reg Pass\n\nReproducibility B
 adges: Artifact Available, Artifact Functional\n\nSession Chair: Hoon Ryu 
 (KISTI)\n\n
END:VEVENT
END:VCALENDAR