SKM 2023 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

O: Fachverband Oberflächenphysik

O 79: Focus Session: Frontiers of Electronic-Structure Theory VI (joint session O/HL)

O 79.3: Topical Talk

Thursday, March 30, 2023, 11:00–11:30, TRE Ma

New Opportunities for First Principles Simulations of Thousands of Atoms Using Linear Scaling Density Functional Theory — •Laura Ratcliff — University of Bristol, Bristol, United Kingdom

Density-functional theory (DFT) is routinely used to simulate a wide variety of materials and properties, however, standard implementations are cubic scaling with the number of atoms, limiting the system sizes which can be treated. This motivated the development of alternative implementations of DFT, which exploit the nearsightedness principle by using a localised description of the system, leading to algorithms with linear scaling (LS) cost which can treat large systems containing tens of thousands of atoms. One approach, which is implemented in the wavelet-based BigDFT code, uses localised orbitals, also known as support functions, which are optimised to reflect their local chemical environment, and thus constitute an accurate minimal basis set. Beyond reduced computational cost, the localised support function description also facilitates additional developments, such as the ability to automate a fragment-based description. In this talk we will present the formalism behind LS-BigDFT, including some examples of the new types of systems and analyses which are opened up by the ability to treat such large systems. We will also describe recent developments in PyBigDFT, a python-based interface which aims to simplify the usage of LS-BigDFT for complex systems and workflows.