Automatic, high-order, and adaptive algorithms for Brillouin zone integration

Jason Kaye,Sophie Beck,Alex Barnett,Lorenzo Van Muñoz,Olivier Parcollet

DOI: https://doi.org/10.21468/SciPostPhys.15.2.062

2023-07-07

Abstract:We present efficient methods for Brillouin zone integration with a non-zero but possibly very small broadening factor $\eta$, focusing on cases in which downfolded Hamiltonians can be evaluated efficiently using Wannier interpolation. We describe robust, high-order accurate algorithms automating convergence to a user-specified error tolerance $\varepsilon$, emphasizing an efficient computational scaling with respect to $\eta$. After analyzing the standard equispaced integration method, applicable in the case of large broadening, we describe a simple iterated adaptive integration algorithm effective in the small $\eta$ regime. Its computational cost scales as $\mathcal{O}(\log^3(\eta^{-1}))$ as $\eta \to 0^+$ in three dimensions, as opposed to $\mathcal{O}(\eta^{-3})$ for equispaced integration. We argue that, by contrast, tree-based adaptive integration methods scale only as $\mathcal{O}(\log(\eta^{-1})/\eta^{2})$ for typical Brillouin zone integrals. In addition to its favorable scaling, the iterated adaptive algorithm is straightforward to implement, particularly for integration on the irreducible Brillouin zone, for which it avoids the tetrahedral meshes required for tree-based schemes. We illustrate the algorithms by calculating the spectral function of SrVO$_3$ with broadening on the meV scale.

Strongly Correlated Electrons,Materials Science,Numerical Analysis

What problem does this paper attempt to address?

The paper is primarily dedicated to addressing the problem of Brillouin zone (BZ) integration in solid-state electronic structure calculations, particularly when dealing with a non-zero but potentially very small broadening factor \(\eta\). Specifically, the study focuses on how to efficiently and automatically compute these integrals when \(\eta\) is small, achieving a user-specified error tolerance \(\varepsilon\).

Key contributions of the paper include:

1. **Iterated Adaptive Integration (IAI)**: This is a simple yet highly accurate method that relies solely on one-dimensional adaptive integration algorithms. This method has a computational complexity of \(O(\log^d(\eta^{-1}))\) as \(\eta \rightarrow 0^+\), offering significant advantages over uniform integration methods when \(\eta\) is small.

2. **Performance Analysis of Tree-based Adaptive Integration (TAI)**: The authors point out that although the TAI method is common, its performance is inferior to the IAI method when \(\eta\) is small. The computational complexity of TAI is \(O(\log(\eta^{-1})/\eta^{d-1})\).

3. **In-depth Discussion of the Periodic Trapezoidal Rule (PTR)**: While PTR is highly efficient for larger \(\eta\), it is not suitable for small \(\eta\). The paper demonstrates how to automate the use of PTR and proposes a hybrid method that combines PTR (for large \(\eta\)) and IAI (for small \(\eta\)) to cover all scenarios.

Additionally, the paper proposes an efficient frequency-adaptive interpolation method for automatically generating effective grids that can resolve local features such as Van Hove singularities, thereby minimizing the number of BZ integrals needed to obtain the spectral function \(A(\omega)\).

Using these methods, researchers are able to compute the spectral function of SrVO_3 with several digits of precision, even when the broadening rate \(\eta\) is as low as 1 meV. These results are particularly useful for many body Green's function methods, where BZ integrals have system and temperature-dependent scattering rates.