BPCPPDYSOE21 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
CPP: Fachverband Chemische Physik und Polymerphysik
CPP 22: Poster Session III - Charged Soft Matter and Theory and Simulation
CPP 22.36: Poster
Wednesday, March 24, 2021, 16:30–18:30, CPPp
Fast high-accuracy optimization of delocalized Gaussian sets for eigenfunctions — Mohammadreza Eidi, •Benjamin Rabe, and Jan-Michael Rost — Max Planck Institute for the Physics of Complex Systems, Dresden, Germany
Gaussian basis sets are widely used to represent the wave function of atomic and molecular systems due to their great advantage that matrix elements with Gaussians at different locations can often be calculated analytically [1]. However, it is not trivial to optimize for a large set of Gaussians {exp(−αi (x − xi)2 | i = 1, ..., N} the parameters {αi, xi}. Procedures with predefined positions following a geometric series, so called even tempered basis sets (e.g. [2]), have been put forward. Here we introduce an iterative scheme which optimizes positions and widths simultaneously for a given number N of s-type Gaussians. To this end we use only during the optimization process auxiliary Gaussians which have in 3D angular momentum p and d character. This way one gets highly accurate results at low computational cost, even for small numbers of Gaussians. We will demonstrate how the method works with explicit 1D examples for various potentials representing atomic and molecular scenarios. [1] Mitroy, Jim, et al., Rev. Mod. Phys. 85.2 (2013): 693. [2] Cherkes, I., Klaiman, S., and Moiseyev, N., Int. J. Quantum Chem. 109, 2996 (2009).