Real-space Method for Evaluating Correlation Energy within Random-phase Approximation
Author(s)
Zhang, Boqin
Advisor(s)
Editor(s)
Collections
Supplementary to:
Permanent Link
Abstract
Correlation interaction between electrons is important for simulating molecules and materials. Random-phase approximation (RPA) under adiabatic-connection fluctuation-dissipation theorem (ACFDT) is an accurate approach to evaluate the correlation energy. However, the expensive cost confines its range of applications. We present a real-space code for evaluating the RPA correlation energy for both isolated and extended systems based on real space grid method. Our code makes a low-rank decomposition on the polarizability operator, and the correlation energy is calculated by eigendecomposition or Spectral Quadrature (SQ). The basis vectors for projection are obtained by subspace iteration, in which the irreducible polarizability and trial vectors are multiplied by solving Sternheimer equations from Density Functional Perturbation Theorem (DFPT). A multi-level parallelization framework is built to distribute the workload efficiently. With the time complexity and parallel scalability outperforming the direct approaches, our real-space RPA code makes large-scale applications possible to explore the influence of electronic correlation interaction on material properties and chemical reactions.
Sponsor
Date
2025-04-28
Extent
Resource Type
Text
Resource Subtype
Dissertation