Our Facebook account

I. A. Obzherin¹, F. N. Yasinsky¹, V. V. Sotsky²

Author affiliations
¹Ivanovo State Power University,
Rabfakovskaya str., 34, 153003 Ivanovo, Russia. 
E-mail: ivan.obzherin@gmail.com
²Nanomaterials Research Institute, Ivanovo State University,
Ermak str., 39, 153025 Ivanovo, Russia. 
E-mail: sotsky2005@yandex.ru

Abstract
The most time consuming part of computations with molecular orbitals is calculation of numerous multiple integrals. It is possible to enhance computation speed using approximate problem solving methods that can be used in parallel computing. The described approach is based on Hartree – Fock method, Monte-Carlo method and NVIDIA CUDA parallel computing tools. The calculations with Hartree – Fock method are described, as well as approaches to enhancing computational speed using Monte-Carlo method combined with dispersion reduction techniques. The parallel computation of the aforementioned methods and techniques is outlined. The described approach allows to create software for assessing calculations of molecular structures without the help of complex software packages and to enhance the computation speed using parallel computing abilities of wide-spread and relatively cheap hardware.

Keywords: molecular orbitals, Hartree – Fock method, Monte-Carlo method, parallel computing, CUDA