This talk describes new developments in numeric atom-centered orbital based all-electron electronic structure theory embodied in the FHI-aims code, for large-scale simulations of molecules and materials with benchmark-quality numerical accuracy, but scalable to large system sizes (thousands of atoms) and scalable from small computer systems to the most important current and future high-performance platforms. The talk will cover recent developments for scalability on GPU hardware and upcoming Intel MIC platforms. A new, open-source software infrastructure "ELSI" provides a seamless interface to traditional O(N<sup>3</sup>) solutions to the self-consistent eigenvalue problem, as well as to newer solutions with a lower scaling exponent. A scalable implementation of periodic hybrid functionals and spin-orbit coupling facilitates materials discovery of new multinary (Cu-Ba-Sn-S/Se and related) semiconductors for light harvesting and new 2D organic-inorganic hybrid perovskite materials. Finally, the talk will touch upon an implementation of nuclear magnetic resonance parameters as a prerequisite to predict the creation of hyperpolarized, long-lived coupled nuclear spin states on molecules for potential future magnetic resonance imaging applications.

Scalable all-electron theory of real-world molecules and materials - examples: Light harvesting and NMR

253rd National Meeting (2017)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.