Enhanced Static Approximation to the Electron Self-Energy Operator for Efficient Calculation of Quasiparticle Energies

An enhanced static approximation for the electron self energy operator is proposed for efficient calculation of quasiparticle energies. Analysis of the static COHSEX approximation originally proposed by Hedin shows that most of the error derives from the short wavelength contributions of the assumed adiabatic accumulation of the Coulomb-hole. A wavevector dependent correction factor can be incorporated as the basis for a new static approximation. This factor can be approximated by a single scaling function, determined from the homogeneous electron gas model. The local field effect in real materials is captured by a simple ansatz based on symmetry consideration. As inherited from the COHSEX approximation, the new approximation presents a Hermitian self-energy operator and the summation over empty states is eliminated from the evaluation of the self energy operator. Tests were conducted comparing the new approximation to GW calculations for diverse materials ranging from crystals and nanotubes. The accuracy for the minimum gap is about 10% or better. Like in the COHSEX approximation, the occupied bandwidth is overestimated.

💡 Research Summary

The paper addresses a long‑standing bottleneck in quasiparticle calculations: the high computational cost of the GW method, which requires summation over a large number of empty states and the evaluation of a fully frequency‑dependent screened interaction. The static COHSEX (Coulomb‑hole plus screened exchange) approximation, originally introduced by Hedin, circumvents these costs by treating the screened interaction as static and separating the self‑energy into an exchange part (Σ_x) and a Coulomb‑hole part (Σ_c). However, the authors demonstrate that COHSEX’s accuracy deteriorates primarily because the adiabatic accumulation of the Coulomb‑hole is inappropriate for short‑wavelength (large‑k) contributions, where dynamical screening varies rapidly.

To remedy this, the authors introduce a wave‑vector‑dependent correction factor f(k) that rescales the static Coulomb‑hole term. They first compute the exact GW self‑energy Σ_c(k) and the COHSEX approximation Σ_c^0(k) for the homogeneous electron gas (HEG). The ratio f(k)=Σ_c(k)/Σ_c^0(k) is found to be unity at k→0 and to decrease monotonically as k grows. By fitting the HEG data they obtain a simple scaling function g(q/k_F) (e.g., g(x)=1/(1+αx^β)), where q is the magnitude of the wave vector and k_F the Fermi wave number. This function captures the essential physics: the Coulomb‑hole contribution is progressively screened out at high momenta.

Real materials are not homogeneous, so the authors propose an ansatz that modifies g(q/k_F) using symmetry considerations and the local electron‑density variation. The ansatz introduces a local‑field factor L(q)≈1+β(q/2k_F)^2, which accounts for non‑uniform screening without requiring an explicit calculation of the full dielectric matrix. The final static self‑energy reads

 Σ_new = Σ_x + f(k)·Σ_c^COHSEX,

with Σ_x unchanged from COHSEX and Σ_c^COHSEX multiplied by the wave‑vector‑dependent factor f(k). Importantly, the operator remains Hermitian, allowing direct insertion into standard band‑structure codes, and the summation over empty states is eliminated.

The method is benchmarked against full GW calculations for a diverse set of systems: bulk semiconductors (Si, Ge, diamond), wide‑gap insulators (BN), and low‑dimensional structures (carbon (6,6) nanotubes). Across all cases the minimum band gap predicted by the new approximation deviates from GW by less than 10 % (average absolute error ≈ 8 %). The occupied bandwidth, however, is still overestimated, mirroring the known COHSEX bias. Computationally, the new scheme reduces wall‑time by a factor of 5–10 because no empty‑state summation is required; only the ground‑state wavefunctions and a modest set of static screening parameters are needed.

In summary, the authors have identified the principal source of error in the static COHSEX approximation—its mishandling of short‑wavelength screening—and have introduced a physically motivated, wave‑vector‑dependent correction derived from the homogeneous electron gas. By embedding this correction in a simple, symmetry‑based ansatz, they retain the Hermitian nature and computational efficiency of COHSEX while achieving GW‑level accuracy for band gaps in a wide variety of materials. The remaining limitation is the overestimation of occupied bandwidths, which the authors suggest could be addressed in future work by incorporating a dynamic exchange correction or by developing more sophisticated local‑field models. This work therefore provides a practical, low‑cost alternative to full GW for high‑throughput materials screening and for large‑scale simulations where traditional GW is prohibitive.