High-Throughput Condensed-Phase Hybrid Density Functional Theory for Large-Scale Finite-Gap Systems: The SeA Approach

High-throughput DFT calculations are key to screening existing/novel materials, sampling potential energy surfaces, and generating quantum mechanical data for machine learning. By including a fraction of exact exchange (EXX), hybrid functionals reduce the self-interaction error in semi-local DFT and furnish a more accurate description of the underlying electronic structure, albeit at a high computational cost that often prohibits such high-throughput applications. To address this challenge, we have constructed SeA (SeA=SCDM+exx+ACE), a robust, accurate, and efficient framework for high-throughput condensed-phase hybrid DFT by combining: (1) the non-iterative selected columns of the density matrix (SCDM) orbital localization scheme, (2) a black-box and linear-scaling EXX algorithm (exx), and (3) adaptively compressed exchange (ACE). By considering a diverse set of aqueous configurations, SeA yields ~20x speedup in the rate-determining step in the convolution-based ACE implementation in Quantum ESPRESSO, while reproducing the EXX energy and ionic forces with high fidelity. In doing so, SeA effectively removes the computational bottleneck that prohibits the routine use of hybrid DFT in high-throughput applications, providing an ~6x speedup in the overall cost of the ACE algorithm (and >100x overall speedup when compared to the conventional EXX implementation) for systems similar in size to (H2O)64. As a proof-of-principle high-throughput application, we used SeA to train a DNN potential for ambient (T=300K, p=1Bar) liquid water at the hybrid (PBE0) DFT level based on an actively learned data set of ~8,000 (H2O)64 configurations. Using an out-of-sample test set ((H2O)512 at T=330K, p=1Bar), we confirmed the accuracy of the SeA-trained DNN potential and showcased the capability of SeA by directly computing the ground-truth ionic forces in this challenging system containing >1,500 atoms.