Speaker
Description
Green's function based methods, such as the Random Phase Approximation (RPA) and the GW methods, provide an accurate prediction of the electronic structure of molecules and solid materials. The space-time formulation of the RPA and GW methods has recently re-emerged to overcome the unfavorable quartic scaling with system size of the canonical implementations [1]. Moreover, the combination of contour deformation and analytical continuation techniques has been used to reduce the highly quintic scaling with system size of the GW method for core level excitations [2]. In this talk, I will give an overview of the recent development of low-scaling algorithms for RPA and GW in the all-electron numerical atomic orbitals program FHI-aims [3]. I will also mention how these developments have used and contributed to the NOMAD project.
[1] H. N. Rojas, R.W. Godby and R.J. Needs, Phys. Rev. Lett. 1995, 74, 1827-1830
[2] R. L. Panadés-Barrueta and D. Golze. J. Chem. Theory Comput. 2023, 19, 5450–5464
[3] V. Blum, et al, Comput. Phys. Commun. 180, 2175, (2009)