Discontinuous Galerkin method with Voronoi partitioning for Quantum Simulation of Chemistry

TL;DR

This paper extends the discontinuous Galerkin method with Voronoi partitioning to quantum chemistry simulations, enabling efficient modeling of molecules and crystals of arbitrary geometry with reduced computational costs.

Contribution

It introduces a Voronoi-based partitioning strategy combined with DG for flexible, lower-cost quantum simulations of complex molecular and crystalline systems.

Findings

Effective for quasi-1D, 2D, and 3D systems

Reduces two-electron integral computations

Applicable to crystalline graphene

Abstract

Molecular orbitals based on the linear combination of Gaussian type orbitals are arguably the most employed discretization in quantum chemistry simulations, both on quantum and classical devices. To circumvent a potentially dense two-body interaction tensor and obtain lower asymptotic costs for quantum simulations of chemistry, the discontinuous Galerkin (DG) procedure using a rectangular partitioning strategy was recently piloted [McClean et al, New J. Phys. 22, 093015, 2020]. The DG approach interpolates in a controllable way between a compact description of the two-body interaction tensor through molecular orbitals and a diagonal characterization through primitive basis sets, such as a planewave dual basis set. The DG procedure gives rise to a block-diagonal representation of the two-body interaction with reduced number of two-electron repulsion integrals, which in turn reduces the cost of quantum simulations. In the present work we extend this approach to be applicable to molecular and crystalline systems of arbitrary geometry. We take advantage of the flexibility of the planewave dual basis set, and combine the discontinuous Galerkin procedure with a general partitioning strategy based on the Voronoi decomposition. We numerically investigate the performance, at the mean-field and correlated levels, with quasi-1D, 2D and 3D partitions using hydrogen chains, H$_4$, CH$_4$ as examples, respectively. We also apply the method to graphene as a prototypical example of crystalline systems.