Title: Projected finite-element orbital population: A scalable computational approach for chemical bonding analysis from DFT calculations

Abstract: We present an efficient scalable computational approach for conducting population analysis from Kohn-Sham density functional theory calculations (DFT) using adaptive spectral finite-element discretization. The proposed method is in the spirit of projected orbital population analysis and provides a unified real-space framework to treat both DFT calculation and projected population analysis using any local atom-centered basis on the same finite-element grid. The work provides an important direction towards extracting chemical bonding information from large-scale Kohn Sham DFT calculations on materials systems involving thousands of atoms while accommodating periodic, semi-periodic or full non-periodic boundary conditions. Towards this, we derive the relevant mathematical expressions and develop computationally efficient and scalable numerical implementation procedures on hybrid CPU-GPU architectures, in order to conduct the projected orbital overlap and orbital Hamilton population analysis in this unified setting within the framework of spectral finite-element discretization of Kohn-Sham DFT. Finally, we benchmark the accuracy of our implementation with Lobster, a widely used population analysis code on representative material systems involving both periodic, and non-periodic DFT calculations and further demonstrate the advantages of our scalable implementation to understand the bonding nature in large-scale silicon nanocomposites containing around 1000 atoms.