Abstract: By climbing the five rungs of Jacob’s ladder, the Density Functional Theory (DFT) hierarchy approaches the “heaven” of chemical accuracy needed for accurate and reliable modeling of molecules and complex materials. In particular, fourth-rung hybrid functionals can provide semi-quantitative accuracy and have therefore been used to generate data for machine-learning (ML) applications for studying important gas-phase systems and reactive processes. However, such hybrid DFT based data remains scarce for large-scale condensed-phase systems due to the prohibitive computational cost associated with evaluating the exact-exchange (EXX) interaction in periodic systems. In this work, we used a recently developed high-throughput hybrid DFT data generator, SeA, for large-scale finite-gap condensed-phase systems containing thousands of atoms. The SeA (SeA = SCDM+exx+ACE) data generator seamlessly integrates three recent theoretical developments, including orbital localization via the non-iterative selected columns of the density matrix (SCDM) method, a black-box linear-scaling EXX solver (exx), and the adaptively compressed exchange (ACE) operator. By harnessing three levels of computational savings, SeA performs hybrid DFT based calculations at an overall time-to-solution comparable to second-rung GGA functionals (i.e., the computational workhorse for condensed-phase systems) and is capable of treating a wide range of condensed-phase systems—without the need for system-dependent parameters—including molecular crystals, aqueous solutions, interfaces, and highly porous materials.
About the speaker
Dr. Hsin-Yu Ko received his Ph.D. in Theoretical Chemistry from Princeton University in 2019. After graduation, Dr. Ko was a Postdoctoral Research Associate at Princeton University under Dr. Car and then moved to Cornell University under Dr. DiStasio. Dr. Ko joined the UNT faculty in 2024 as Assistant Professor. His research is based on quantum and statistical mechanics, developing computational algorithms and software using high-performance computing and machine-learning techniques.
