🔬 physics

Smoothed truncated Coulomb potential for periodic Gaussian-basis Hartree--Fock exchange

This paper introduces the smoothed truncated Coulomb (sTC) potential, a systematically improvable approximation that enables efficient and accurate periodic Hartree--Fock exchange calculations with Gaussian basis sets for both pseudopotential and all-electron systems by smoothing the truncation boundary to facilitate a dual-space integral evaluation algorithm.

Original authors: Gengzhi Yang, Aamy Bakry, Hong-Zhou Ye

Published 2026-09-02

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Gengzhi Yang, Aamy Bakry, Hong-Zhou Ye

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine trying to understand how a crowd of people moves through a city by watching just a single, tiny block. If you only look at that one block, you miss the flow of traffic from the rest of the city, and your picture of the movement will be distorted. In the world of quantum chemistry, scientists face a similar challenge when they try to model how electrons behave inside solid materials like diamonds or metals. To do this, they use a powerful mathematical framework called Hartree–Fock theory, which calculates how electrons repel each other. However, because computers cannot handle an infinite amount of space, scientists must simulate these materials using a repeating, finite box. This creates an artificial edge where the simulation stops, and the electrons on one side of the box interact with their own copies on the other side in a way that doesn't happen in the real, infinite world. These artificial interactions create errors that make it difficult to predict the true properties of the material, such as how hard it is or how it conducts electricity.

For years, researchers have tried to fix this by cutting off the long-range interactions at the edge of their simulation box, a technique known as truncating the Coulomb potential. This works beautifully in some computer methods, but it hits a wall when scientists try to use it with the specific mathematical tools, called Gaussian basis sets, that are standard for describing atoms in molecules and solids. The problem is that the sharp edge created by this cutoff is mathematically jagged, making it incredibly difficult and slow to calculate the forces between electrons, especially when trying to include every single electron in the atom rather than just the outer ones. Without a way to smooth out this jagged edge, scientists are forced to choose between using a method that is accurate but limited to simplified models, or a method that includes all the electrons but suffers from large errors.

A team of researchers has now introduced a new approach that bridges this gap, allowing for highly accurate calculations that include every electron without the usual computational headaches. They developed a method they call the smoothed truncated Coulomb potential. Instead of creating a sharp, jagged cutoff where the interaction suddenly stops, their technique gently blurs the boundary. Think of it like smoothing the rough edge of a piece of sandpaper until it becomes a gentle slope; this small change makes the mathematics much easier to handle while keeping the physics accurate. By adjusting a single control knob, the researchers can make this slope as steep or as gentle as needed, allowing them to systematically improve the accuracy of their results until they match the ideal, error-free scenario.

The researchers tested this new method on a wide variety of materials, ranging from insulators like diamond and silicon to metals like lithium and aluminum, and even complex molecular crystals. They found that by using their smoothed approach, they could reproduce the results of the ideal, error-free calculations with remarkable precision. For insulating materials, they only needed to set their control knob to a specific moderate value to get results that were virtually indistinguishable from the perfect theoretical limit. For metals, which are generally harder to model because their electrons move more freely, they needed to tighten the setting slightly, but the method still worked efficiently. Crucially, this new technique allowed them to perform these high-precision calculations on all-electron systems, which were previously too difficult to handle with this level of accuracy.

The impact of this work is seen in how quickly the results converge to the true answer as the simulation size increases. In the past, scientists using older methods had to simulate massive, computationally expensive grids of points to get a reliable answer, and even then, the results often lagged behind the truth. With the new smoothed method, the errors drop away much faster. For example, in calculations for molecular crystals, the new method reached a level of accuracy that is considered the gold standard for chemical predictions using a grid of points that was significantly smaller than what was previously required. This means that scientists can now obtain reliable predictions for the structure and behavior of materials much faster and with less computing power.

The researchers also demonstrated that this method works consistently across different types of materials, whether they are layered, metallic, or made of molecules. They showed that the results for the spacing between atoms and the stiffness of the material matched the best available theoretical values. Furthermore, the method proved robust enough to handle the complex, irregular shapes of the boundaries found in real crystals, not just simple spheres. By solving the problem of the jagged edge, the team has provided a practical tool that allows the scientific community to move beyond simplified models and study the full complexity of real materials with a level of detail that was previously out of reach. This advancement ensures that the theoretical models used to design new materials and understand their properties are now grounded in a more accurate and efficient foundation.

Technical Summary: Smoothed Truncated Coulomb Potential for Periodic Gaussian-Basis Hartree–Fock Exchange

Problem Statement
Periodic Hartree–Fock (HF) and hybrid density functional theory (DFT) calculations suffer from finite-size errors (FSEs) arising from the long-range nature of the Coulomb interaction. While the Truncated Coulomb (TC) method effectively mitigates these errors by removing spurious exchange interactions between periodic images, its application within Gaussian-basis sets presents significant computational challenges. Specifically, evaluating electron repulsion integrals (ERIs) over the TC kernel is difficult for non-spherical truncation domains, such as the Wigner–Seitz (WS) cell. Existing implementations for spherical truncation (Sph-TC) exist, but generalizing to WS-TC in Gaussian codes has been limited. Current approaches, such as Fast Fourier Transform Density Fitting (FFTDF) used in PySCF, require expanding pair densities in a plane-wave basis. This necessitates an impractically large plane-wave cutoff to resolve compact core densities in all-electron calculations, effectively restricting WS-TC to pseudopotential treatments. There is a lack of an efficient local-orbital implementation of WS truncation that remains practical for all-electron systems.

Methodology
The authors introduce the Smoothed Truncated Coulomb (sTC) potential, a systematically improvable approximation to a parent TC potential. The core methodology involves:

Potential Construction: The sTC potential is defined by convolving the complementary part of the TC potential with a normalized Gaussian function and subtracting the result from the full Coulomb potential. This smooths the sharp truncation boundary over a characteristic length scale controlled by a parameter $\omega$ .
Dimensionless Control Parameter: Accuracy is governed by the Inverse Relative Smoothing Width (IRSW), defined as $\eta = \omega R_{in}$ , where $R_{in}$ is the inradius of the truncation domain. As $\eta \to \infty$ , the sTC potential converges to the parent TC potential.
Range Separation: The sTC potential is decomposed into short-range (SR) and long-range (LR) components.
- The SR component decays rapidly in reciprocal space and is evaluated via a real-space lattice sum using standard complementary-error-function (erfc)-attenuated Coulomb integrals, which are widely available in molecular integral libraries.
- The LR component retains the smoothed TC kernel and is evaluated in reciprocal space. Crucially, the Gaussian damping factor suppresses high-momentum components, allowing the use of a modest plane-wave basis even for all-electron calculations, as compact core contributions are strongly damped.
Dual-Space Algorithm: The authors implement a Range-Separated Density Fitting (RSDF) algorithm in PySCF. This approach avoids the need to resolve compact core densities in a large plane-wave basis by evaluating the LR contribution analytically in reciprocal space and the SR contribution in real space. This enables efficient periodic HF calculations for both pseudopotential and all-electron systems.

Key Contributions

Formulation: Development of the sTC potential as a smooth, systematically convergent approximation to TC potentials, applicable to general truncation domains including the WS cell.
Algorithm: Implementation of a dual-space RSDF algorithm that eliminates the plane-wave basis size bottleneck for all-electron WS-TC calculations.
Software: Integration of these methods into PySCF, providing the first efficient Gaussian-basis implementation of WS-TC-quality HF exchange for all-electron systems.
Parameterization: Systematic benchmarking to identify practical values of $\eta$ that balance accuracy and computational cost.

Results
The authors benchmarked the sTC method across insulating, semiconducting, layered, metallic, and molecular-crystal systems:

Convergence to TC: For gapped systems (diamond, silicon, h-BN), an IRSW of $\eta \ge 3$ reproduces WS-TC exchange energies, band structures, and structural properties with high fidelity. The difference between sTC and TC is negligible compared to the remaining finite-size error. For metallic systems (lithium, aluminum), a slightly more conservative $\eta = 4$ is recommended to achieve similar agreement, though $\eta=3$ still offers significant improvement over standard methods.
All-Electron Performance: The all-electron sTC calculations exhibit FSE trends parallel to their pseudopotential counterparts. The method successfully extends TC-quality convergence to all-electron settings, where direct TC calculations are computationally prohibitive.
Comparison to Ewald: Across all systems, sTC-based HF calculations substantially improve thermodynamic-limit (TDL) convergence rates compared to the probe-charge Ewald method. For molecular crystals, sTC with $\eta \ge 3$ achieves chemical accuracy (errors < 1 kcal/mol) with sparse k-meshes (e.g., $2 \times 2 \times 2$ or even $\Gamma$ -point for some cases), whereas the Ewald method requires denser sampling and extrapolation.
Structural and Spectral Properties: The method accurately reproduces lattice constants, bulk moduli, and band gaps. The smoothing has minimal impact on spectral properties at the recommended $\eta$ values.

Significance
The paper claims that the sTC potential provides a practical and systematically improvable route to high-accuracy periodic HF exchange calculations using Gaussian basis sets. Its primary significance lies in enabling all-electron WS-TC calculations, a capability previously lacking due to the computational cost of resolving core densities in reciprocal space. By decoupling the accuracy of the truncation approximation from the computational cost of the integral evaluation, the sTC approach allows researchers to achieve rapid thermodynamic-limit convergence for a wide range of materials (including metals and molecular crystals) without resorting to large plane-wave bases or complex numerical schemes. The work establishes $\eta = 3$ for gapped systems and $\eta = 4$ for metals as robust, practical choices for achieving TC-quality results in standard Gaussian-basis codes.

Technical Summary: Smoothed Truncated Coulomb Potential for Periodic Gaussian-Basis Hartree–Fock Exchange

More like this