Spectral scheme for atomic structure calculations in density functional theory

Sayan Bhowmik,John E. Pask,Andrew J. Medford,Phanish Suryanarayana
2024-06-07
Abstract:We present a spectral scheme for atomic structure calculations in pseudopotential Kohn-Sham density functional theory. In particular, after applying an exponential transformation of the radial coordinates, we employ global polynomial interpolation on a Chebyshev grid, with derivative operators approximated using the Chebyshev differentiation matrix, and integrations using Clenshaw-Curtis quadrature. We demonstrate the accuracy and efficiency of the scheme through spin-polarized and unpolarized calculations for representative atoms, while considering local, semilocal, and hybrid exchange-correlation functionals. In particular, we find that $\mathcal{O}$(200) grid points are sufficient to achieve an accuracy of 1 microhartree in the eigenvalues for optimized norm conserving Vanderbilt pseudopotentials spanning the periodic table from atomic number $Z = 1$ to $83$.
Computational Physics,Materials Science
What problem does this paper attempt to address?
This paper attempts to address the problem of improving the accuracy and efficiency of atomic structure calculations in Density Functional Theory (DFT). Specifically, the authors propose a scheme based on spectral methods for atomic structure calculations in pseudopotential Kohn-Sham DFT. By introducing an exponential transformation of the radial coordinate and using global polynomial interpolation on a Chebyshev grid, this method can efficiently and accurately solve electronic structure problems. ### Main Issues and Solutions 1. **Computational Efficiency and Accuracy**: - **Problem**: Traditional atomic structure calculation methods can achieve relatively accurate results, but the computational cost is high when dealing with large-scale systems. - **Solution**: The authors propose a spectral method that maps the radial coordinate to a finite interval through exponential transformation and performs polynomial interpolation on a Chebyshev grid. This method can achieve high precision with fewer grid points, significantly improving computational efficiency. 2. **Handling Different Types of Exchange-Correlation Functionals**: - **Problem**: Existing atomic structure calculation methods usually can only handle local or semi-local exchange-correlation functionals and cannot effectively deal with hybrid functionals. - **Solution**: The proposed spectral method is not only applicable to local and semi-local exchange-correlation functionals but also capable of handling hybrid functionals, thus enhancing the method's generality and applicability. ### Specific Implementation - **Radial Coordinate Transformation**: The radial coordinate \( r \) is mapped to a finite interval \( y \) through exponential transformation, making the grid points denser near the atomic nucleus and gradually sparser away from the nucleus. - **Chebyshev Grid**: In the transformed coordinate system, a Chebyshev grid is used for discretization, employing the Chebyshev differential matrix to approximate derivative operators and using the Clenshaw-Curtis quadrature method for integration. - **Boundary Condition Handling**: By appropriately modifying the Chebyshev differential matrix, Dirichlet boundary conditions are ensured. ### Validation and Results - **Accuracy Validation**: The proposed method's accuracy and convergence under different exchange-correlation functionals were validated by comparison with reference codes ONCVPSP and SPARC. - **Efficiency Evaluation**: Results show that for optimized van der Waals pseudopotentials (ONCV), only about 200 grid points are needed to achieve an accuracy of 1 microhartree. In summary, this paper proposes an efficient spectral method that can significantly reduce the computational cost of atomic structure calculations while maintaining high accuracy, applicable to various exchange-correlation functionals.