Title

Explicit Analytical OSV-MP2 Energy   Gradients For Molecular Dynamical Properties

Reporter

Prof. YANG Jun

Reporter’s   institution

Department of Chemistry, The University of   Hong Kong

report time

2018-12-10   10:00

Report location

Conference room on the 9th floor（9004）of Hefei National   Laboratory for Physical Sciences at the Microscale

Organizer

Hefei National Laboratory for Physical   Sciences at the Microscale，International Center   for Chemical Theory (ICCT), School of Chemistry and Materials Science

Report introduction

Abstract：

In several limits of approximation, an   efficient low-scaling hierarchy of coupled cluster (CC) electronic structure methods   has been established, such as the alternatives within the framework of   projected atomic orbitals, orbital-specific-virtual (OSV),   pair-nature-orbital (PNO) and so on. These methods have been demonstrated to   often achieve chemical accuracy with significantly reduced CC computational   cost by orders of magnitude, and applied to treat systems containing hundreds   of atoms as well as condensed materials. The success of these high-level   low-scaling electronic structure methods in computing molecular static   properties opens up a new perspective for us to think in new ways to compute   molecular dynamical properties by making similar efficiency and accuracy to   those of static calculations. Along this line, we are developing an algorithm   to compute the explicit analytical gradients towards the OSV-based hierarchy   of energy functionals, starting with the second-order Møller Plesset   perturbation (OSV-MP2) method. Here we consider the explicit relaxation of   perturbed OSV orbitals that are required to be the eigenvectors of the   perturbed MP2 diagonal amplitudes. In this presentation, we will have a   discussion on (1) the algorithm and implementation with an analysis of formal   computing complexity, (2) formal issues and properties of both OSV and hybrid   PNO orbital response vectors, (3) the numerical assessment of OSV-MP2   structure optimization for a test set of molecules, and (4) a   proof-of-principle demonstration of the low-scaling ab-initio molecular   dynamics simulation driven by OSV-MP2 forces.

Biosketch：

 Jun Yang is   an assistant professor of Department of Chemistry of   The University of Hong Kong. He received his Ph.D. from Institut   für Theoretische Chemie, Universität zu Köln in 2007, under the direction of   Prof. Michael Dolg. After doing postdoctoral research with Prof. Garnet Chan   at Cornell University and at Princeton University, he became a lecturer of   Department of Chemistry of Princeton University in 2013. He then moved to   Hong Kong and joined the HKU faculty in 2016. Jun Yang has published many   important papers, including 1 in Science. He has also received numerous   awards, inlcudings the Early Career Scheme Award from The Research Grants   Council of Hong Kong, the IBM-Löwdin Award from the 50th Sanibel Symposium,   and the DFG SPP1145 Research Fellow from Deutsche   Forschungsgemeinschaft, Germany.