Premium
Cover Image, Volume 35, Issue 28
Publication year - 2014
Publication title -
journal of computational chemistry
Language(s) - English
Resource type - Reports
SCImago Journal Rank - 0.907
H-Index - 188
eISSN - 1096-987X
pISSN - 0192-8651
DOI - 10.1002/jcc.23749
Subject(s) - cover (algebra) , computer science , bottleneck , matrix (chemical analysis) , key (lock) , image (mathematics) , theoretical computer science , algorithm , computer graphics (images) , computational science , artificial intelligence , operating system , engineering , chemistry , mechanical engineering , chromatography , embedded system
A key bottleneck in the Hartree‐Fock method is exchange matrix calculation. On page 2056 (DOI: 10.1002/jcc.23720 ) Taweetham Limpanuparb, Josh Milthorpe, and Alistair Rendell present an algorithm for computing long‐range Coulomb and exchange interactions implemented on the modern parallel programming language X10. Using the Ewald resolution, the calculation of the exchange matrix can be efficiently distributed to multiple computing nodes. The new algorithm is tested on ‘Raijin’, the largest supercomputer in Australia operated by the National Computational Infrastructure. The study shows performance and accuracy results of large molecule/high quality basis running on up to 1024 cores of Raijin.