Mixed Ramp-Gaussian Basis Sets for Core-Dependent Properties: STO-RG and STO-R2G for Li-Ne

Citations Over Time

Abstract

The traditional Gaussian basis sets used in modern quantum chemistry lack an electron-nuclear cusp, and hence struggle to accurately describe core electron properties. A recently introduced novel type of basis set, mixed ramp-Gaussians, introduce a new primitive function called a ramp function which addresses this issue. This paper introduces three new mixed ramp-Gaussian basis sets - STO-R, STO-RG and STO-R2G, made from a linear combination of ramp and Gaussian primitive functions - which are derived from the single-core-zeta Slater basis sets for the elements Li to Ne. This derivation is done in an analogous fashion to the famous STO-nG basis sets. The STO-RG basis functions are found to outperform the STO-3G basis functions and STO-R2G outperforms STO-6G, both in terms of wavefunction fit and other key quantities such as the one-electron energy and the electron-nuclear cusp. The second part of this paper performs preliminary investigations into how standard all-Gaussian basis sets can be converted to ramp-Gaussian basis sets through modifying the core basis functions. Using a test case of the 6-31G basis set for carbon, we determined that the second Gaussian primitive is less important when fitting a ramp-Gaussian core basis function directly to an all-Gaussian core basis function than when fitting to a Slater basis function. Further, we identified the basis sets that are single-core-zeta and thus should be most straightforward to convert to mixed ramp-Gaussian basis sets in the future.

Related Papers

• → Mixed Ramp-Gaussian Basis Sets for Core-Dependent Properties: STO-RG and STO-R2G for Li-Ne(2020)4 cited
• → The interacting correlated fragments method with non-orthogonal orbitals(1999)4 cited
• → Gaussian Distributions(2017)1 cited
• → Reproducing Gaussian densities and linear Gaussian detection(2000)10 cited
• → Gaussian Fields(2019)