We introduce new computational and algorithmic paradigms that can significantly improve the efficiency of existing methodologies used in first-principle DFT ground state and TDDFT excited states calculations of complex molecular systems and nanostructures. Our modeling framework  benefits from well-suited combinations of numerical methods that can exploit the linear scaling capabilities of real-space mesh techniques, and provide parallel optimization at various stages of the modeling process. The significant challenges posed by these large-scale atomistic calculations have been addressed with the development of the FEAST eigensolver (www.feast-solver.org), a new state-of-the-art HPC numerical library for solving the eigenvalue problem (Hermitian and non-Hermitian) which has been recently integrated into the INTEL-MKL library. We will then outline how FEAST can be operated beyond the "black-box" solver for addressing: (i) the all-electron (full potential) electronic structure problem; (ii) the self-consistent (SCF) problem for DFT; and (iii) the real-time propagation approach for TDDFT. Finally, we will review the broader impacts of these new FEAST-based computational strategies by presenting various simulation results with applications ranging from computational electronic spectroscopy of molecules, to plasmonic excitations in carbon-based nanostructures.

Scalable algorithms for real-space and real-time first-principle calculations

254th National Meeting (2017)

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