Static Hermitian Self-Energy

• Static Hermitian self-energy is defined as a frequency-independent operator replacing the dynamical self-energy, transforming a nonlinear Dyson equation into a linear Hermitian eigenvalue problem.
• Methodologies like static COHSEX and GW-based hierarchies correct Coulomb-hole terms and enhance numerical robustness while preserving quasiparticle physics.
• The concept extends beyond electronic structure to include geometric potentials and static-source self-energy in lattice gauge theory, illustrating its versatility in static energy corrections.

Searching arXiv for the specified papers and closely related uses of “static Hermitian self-energy” to ground the article in the cited literature. Search query: "static Hermitian self-energy arXiv (Kang et al., 2010, Loos et al., 9 Apr 2026, Bali et al., 2011)"

“Static Hermitian self-energy” denotes, in its most standard contemporary usage, a self-energy operator that is both frequency independent and Hermitian, so that quasiparticle energies are obtained from an ordinary Hermitian eigenvalue problem rather than from a nonlinear dynamical Dyson equation. In electronic-structure theory this usage is explicit: “static” means that the operator does not depend explicitly on energy or frequency, while “Hermitian” means that it is a real symmetric operator in the quasiparticle basis, with real eigenvalues and orthonormal quasiparticle wavefunctions (Kang et al., 2010). The literature does not use the expression uniformly, however. Closely related language also appears for curvature-induced residual terms on constrained manifolds, and, in a descriptive rather than terminological sense, for the self-energy of a static color source in lattice gauge theory (Shikakhwa et al., 2021, Bali et al., 2011).

1. Conceptual scope and defining properties

In many-body perturbation theory, the natural reference point is the dynamical self-energy

$$\Sigma(\mathbf r,\mathbf r';E)=\frac{i}{2\pi}\int dE' \, e^{-i\delta^+ E'}\, G(\mathbf r,\mathbf r';E-E')\,W(\mathbf r,\mathbf r';E'),$$

which is nonlocal and energy dependent. A static Hermitian self-energy replaces that object by a frequency-independent effective operator. The immediate consequence is structural rather than merely numerical: the quasiparticle problem becomes a linear Hermitian eigenvalue problem instead of a nonlinear root-finding problem (Kang et al., 2010, Loos et al., 9 Apr 2026).

The Hermitian condition is not an incidental aesthetic choice. In the arXiv literature summarized here it is tied to real spectra, orthonormal effective eigenstates, and numerical robustness. Staticity likewise has a concrete meaning: no explicit energy dependence survives in the final operator. This sharply distinguishes static Hermitian constructions from full $GW$, coupled-cluster Green’s-function formalisms, and other spectral theories in which poles, resolvents, and branch-specific denominators are essential (Coveney et al., 9 Mar 2025, Loos et al., 9 Apr 2026).

A broader reading of the phrase is also present in other subfields. On curved surfaces, Hermitizing the surface and normal momentum operators produces a finite curvature-dependent term that survives after the normal Hermitian kinetic energy is removed; the corresponding paper frames the geometric potential as a kind of static Hermitian self-energy (Shikakhwa et al., 2021). In lattice gauge theory, the static-source self-energy is the additive mass renormalization of an infinitely heavy color source; although the paper does not use the phrase “static Hermitian self-energy,” it studies the standard static-source self-energy extracted from a temporal Polyakov line, which is the same quantity in the static limit if that label is used descriptively (Bali et al., 2011).

2. Static Hermitian self-energy in quasiparticle electronic structure

The clearest explicit formulation appears in the effort to improve the static COHSEX approximation while retaining its computational advantages. In static COHSEX, the self-energy is decomposed into screened exchange and Coulomb-hole parts and then frozen at zero frequency:

$$\Sigma_{SEX}^{static}(\mathbf r,\mathbf r') = -\sum_{n,\mathbf k}^{occ} \phi_{n\mathbf k}(\mathbf r)\phi^*_{n\mathbf k}(\mathbf r') \,W(\mathbf r,\mathbf r';E=0),$$

$$\Sigma_{COH}^{static}(\mathbf r,\mathbf r') = \frac{1}{2}\delta(\mathbf r-\mathbf r')\,W_p(\mathbf r,\mathbf r';E=0),$$

with $W_p=W-v$. This form is energy independent and Hermitian, and it eliminates the need for explicit empty-state summations in the self-energy evaluation (Kang et al., 2010).

The central defect of static COHSEX is traced to the Coulomb-hole term. The screened-exchange contribution is already fairly close to full $GW$, whereas the Coulomb-hole term is too large in magnitude because COHSEX assumes an instantaneous, adiabatic accumulation of the Coulomb hole. The error is wavevector dependent: COHSEX works well at long wavelength, while at short wavelength the static COH term is wrong by about a factor of 2. For the homogeneous electron gas, the diagnostic ratio

$f(q,E_k)=\frac{W_p^{full}(q,E_k)}{W_p(q,E=0)}$

approaches $1$ for small $q$ and $0.5$ for large $GW$0 (Kang et al., 2010).

The improved static Hermitian approximation keeps the static structure but corrects the Coulomb-hole term by a wavevector-dependent factor derived from the homogeneous electron gas. In that setting,

$GW$1

where $GW$2 is a universal scaling function fit in Padé form. For real materials, $GW$3 is replaced by

$GW$4

and local-field effects are incorporated through the symmetry-based ansatz

$GW$5

The resulting operator remains Hermitian because the correction factor is a real scalar function of wavevector magnitude and the COH term retains a local real form in coordinate space (Kang et al., 2010).

The practical motivation is efficiency without a full loss of quasiparticle accuracy. Tests reported for crystals and nanotubes show that the accuracy for the minimum gap is about $GW$6 or better, while occupied bandwidths remain overestimated, as in COHSEX, because the Coulomb-hole term is still local and lacks dispersion (Kang et al., 2010). This establishes the canonical meaning of a static Hermitian self-energy in modern electronic-structure work: a frequency-independent surrogate for $GW$7 that preserves Hermiticity and much of the quasiparticle physics while avoiding the full dynamical problem.

3. From dynamical $GW$8 to a purely static Hermitian operator

A later development constructs a systematic hierarchy of $GW$9-based approximations by progressively removing dynamical content from the self-energy rather than replacing it in a single step. The starting point is the nonlinear Dyson equation

$$\Sigma_{SEX}^{static}(\mathbf r,\mathbf r') = -\sum_{n,\mathbf k}^{occ} \phi_{n\mathbf k}(\mathbf r)\phi^*_{n\mathbf k}(\mathbf r') \,W(\mathbf r,\mathbf r';E=0),$$0

and its equivalent auxiliary-space supermatrix representation

$$\Sigma_{SEX}^{static}(\mathbf r,\mathbf r') = -\sum_{n,\mathbf k}^{occ} \phi_{n\mathbf k}(\mathbf r)\phi^*_{n\mathbf k}(\mathbf r') \,W(\mathbf r,\mathbf r';E=0),$$1

In that representation, the $$\Sigma_{SEX}^{static}(\mathbf r,\mathbf r') = -\sum_{n,\mathbf k}^{occ} \phi_{n\mathbf k}(\mathbf r)\phi^*_{n\mathbf k}(\mathbf r') \,W(\mathbf r,\mathbf r';E=0),$$2 block describes hole satellites and the $$\Sigma_{SEX}^{static}(\mathbf r,\mathbf r') = -\sum_{n,\mathbf k}^{occ} \phi_{n\mathbf k}(\mathbf r)\phi^*_{n\mathbf k}(\mathbf r') \,W(\mathbf r,\mathbf r';E=0),$$3 block describes particle satellites, while exact elimination of those auxiliary configurations yields the full dynamical $$\Sigma_{SEX}^{static}(\mathbf r,\mathbf r') = -\sum_{n,\mathbf k}^{occ} \phi_{n\mathbf k}(\mathbf r)\phi^*_{n\mathbf k}(\mathbf r') \,W(\mathbf r,\mathbf r';E=0),$$4 self-energy (Loos et al., 9 Apr 2026).

The hierarchy begins by downfolding only one branch. For example, if the $$\Sigma_{SEX}^{static}(\mathbf r,\mathbf r') = -\sum_{n,\mathbf k}^{occ} \phi_{n\mathbf k}(\mathbf r)\phi^*_{n\mathbf k}(\mathbf r') \,W(\mathbf r,\mathbf r';E=0),$$5 sector is projected out but the $$\Sigma_{SEX}^{static}(\mathbf r,\mathbf r') = -\sum_{n,\mathbf k}^{occ} \phi_{n\mathbf k}(\mathbf r)\phi^*_{n\mathbf k}(\mathbf r') \,W(\mathbf r,\mathbf r';E=0),$$6 sector is retained explicitly, the reduced problem contains $$\Sigma_{SEX}^{static}(\mathbf r,\mathbf r') = -\sum_{n,\mathbf k}^{occ} \phi_{n\mathbf k}(\mathbf r)\phi^*_{n\mathbf k}(\mathbf r') \,W(\mathbf r,\mathbf r';E=0),$$7. The key approximation is then to evaluate that branch at the symmetric frequency

$$\Sigma_{SEX}^{static}(\mathbf r,\mathbf r') = -\sum_{n,\mathbf k}^{occ} \phi_{n\mathbf k}(\mathbf r)\phi^*_{n\mathbf k}(\mathbf r') \,W(\mathbf r,\mathbf r';E=0),$$8

which defines the static Hermitian matrix element

$$\Sigma_{SEX}^{static}(\mathbf r,\mathbf r') = -\sum_{n,\mathbf k}^{occ} \phi_{n\mathbf k}(\mathbf r)\phi^*_{n\mathbf k}(\mathbf r') \,W(\mathbf r,\mathbf r';E=0),$$9

This yields a “half-and-half” scheme in which one branch remains dynamical and the other becomes static (Loos et al., 9 Apr 2026).

When both branches are treated in this way, the self-energy becomes purely static:

$$\Sigma_{COH}^{static}(\mathbf r,\mathbf r') = \frac{1}{2}\delta(\mathbf r-\mathbf r')\,W_p(\mathbf r,\mathbf r';E=0),$$0

and the Dyson equation reduces to the Hermitian eigenvalue problem

$$\Sigma_{COH}^{static}(\mathbf r,\mathbf r') = \frac{1}{2}\delta(\mathbf r-\mathbf r')\,W_p(\mathbf r,\mathbf r';E=0),$$1

This is the paper’s “novel static Hermitian self-energy”: a frequency-independent effective one-body operator obtained as the static endpoint of a controlled $$\Sigma_{COH}^{static}(\mathbf r,\mathbf r') = \frac{1}{2}\delta(\mathbf r-\mathbf r')\,W_p(\mathbf r,\mathbf r';E=0),$$2-based hierarchy, not as an ad hoc model (Loos et al., 9 Apr 2026).

The construction is formally distinct from qs$$\Sigma_{COH}^{static}(\mathbf r,\mathbf r') = \frac{1}{2}\delta(\mathbf r-\mathbf r')\,W_p(\mathbf r,\mathbf r';E=0),$$3. In qs$$\Sigma_{COH}^{static}(\mathbf r,\mathbf r') = \frac{1}{2}\delta(\mathbf r-\mathbf r')\,W_p(\mathbf r,\mathbf r';E=0),$$4, the static operator is defined by symmetrizing dynamical self-energy matrix elements,

$$\Sigma_{COH}^{static}(\mathbf r,\mathbf r') = \frac{1}{2}\delta(\mathbf r-\mathbf r')\,W_p(\mathbf r,\mathbf r';E=0),$$5

whereas the new hierarchy symmetrizes the frequency argument first and evaluates the self-energy once at the midpoint $$\Sigma_{COH}^{static}(\mathbf r,\mathbf r') = \frac{1}{2}\delta(\mathbf r-\mathbf r')\,W_p(\mathbf r,\mathbf r';E=0),$$6. The paper reports that the two static constructions are nevertheless numerically very close: the mean absolute and mean signed deviations between the new static Hermitian self-energy and qs$$\Sigma_{COH}^{static}(\mathbf r,\mathbf r') = \frac{1}{2}\delta(\mathbf r-\mathbf r')\,W_p(\mathbf r,\mathbf r';E=0),$$7 are about $$\Sigma_{COH}^{static}(\mathbf r,\mathbf r') = \frac{1}{2}\delta(\mathbf r-\mathbf r')\,W_p(\mathbf r,\mathbf r';E=0),$$8 eV and $$\Sigma_{COH}^{static}(\mathbf r,\mathbf r') = \frac{1}{2}\delta(\mathbf r-\mathbf r')\,W_p(\mathbf r,\mathbf r';E=0),$$9 eV, respectively, with a maximum deviation of $W_p=W-v$0 eV (Loos et al., 9 Apr 2026).

The same work emphasizes why static Hermitian projections can be attractive even when they are less complete than full dynamical $W_p=W-v$1. By eliminating explicit frequency dependence from the final eigenvalue problem, the purely static construction avoids unstable root-finding and near-singular behavior, is more Davidson-friendly, and reduces algorithmic complexity. The partially static “half-and-half” schemes retain more dynamical information and can yield valence ionization-potential errors of only a few hundredths of an eV relative to fully dynamical $W_p=W-v$2, whereas the purely static scheme is less accurate with respect to full $W_p=W-v$3 but can sometimes be slightly closer to high-level reference data because of error cancellation (Loos et al., 9 Apr 2026).

4. What a static Hermitian self-energy is not

The conceptual limits of the term become especially clear in non-Hermitian Green’s-function theories. In the coupled-cluster similarity-transformed framework, the single-particle self-energy is fundamentally non-Hermitian and frequency dependent. The formal Dyson structure is

$W_p=W-v$4

and the self-energy is explicitly decomposed as

$W_p=W-v$5

Here $W_p=W-v$6 is static, but the forward and backward terms are dynamical resolvent contributions. The paper states that the coupled-cluster self-energy is fundamentally non-Hermitian, generally frequency dependent, and built from effective interactions that themselves depend on the Green’s function (Coveney et al., 9 Mar 2025).

This does not mean that static limits are absent. The static component can be written as

$W_p=W-v$7

and, in the electronic limit where the similarity transformation is removed,

$W_p=W-v$8

The same work also states that when $W_p=W-v$9, the coupled-cluster Dyson equation reduces to the electronic Dyson equation and the coupled-cluster self-energy reduces to the two-body electronic self-energy (Coveney et al., 9 Mar 2025).

A plausible implication is that “static Hermitian self-energy” names a controlled approximation or limiting case rather than the generic endpoint of all self-energy formalisms. In the coupled-cluster setting, the exact theory is biorthogonal, non-Hermitian, and spectral. Static Hermitian self-energies arise only after projection, reduction, or removal of the non-Hermitian similarity transformation. This is precisely why the phrase is natural in some $GW$0-based approximations and unnatural in full coupled-cluster Green’s-function theory (Coveney et al., 9 Mar 2025).

5. Geometric and static-spacetime reinterpretations

Outside many-body quasiparticle theory, the phrase acquires a geometrical meaning. For a spin-zero particle confined to a curved surface, the central construction begins from the three-dimensional momentum operator in adapted curvilinear coordinates,

$GW$1

decomposed into surface and normal pieces,

$GW$2

Neither component is Hermitian by itself with respect to the $GW$3 inner product, so the paper rewrites the momentum as a sum of separately Hermitian operators,

$GW$4

with the Hermitizing correction

$GW$5

Near the surface,

$GW$6

and

$GW$7

These terms generate the geometric potential (Shikakhwa et al., 2021).

For a free neutral particle, the Hermitian normal kinetic term $GW$8 can be removed cleanly in the thin-layer limit because it is itself Hermitian. The surviving surface Hamiltonian is

$GW$9

with geometric potential

$f(q,E_k)=\frac{W_p^{full}(q,E_k)}{W_p(q,E=0)}$0

The paper explicitly frames this geometric potential as a kind of static Hermitian self-energy: it is the finite curvature-dependent residue left after one enforces Hermiticity on surface and normal dynamics and then constrains the particle to the surface (Shikakhwa et al., 2021).

A related but distinct static-spacetime self-energy problem is the renormalized energy of a pointlike electric dipole in a static $f(q,E_k)=\frac{W_p^{full}(q,E_k)}{W_p(q,E=0)}$1-dimensional curved spacetime. There the self-energy functional is classically invariant under the local conformal rescaling

$f(q,E_k)=\frac{W_p^{full}(q,E_k)}{W_p(q,E=0)}$2

but the regularization defining

$f(q,E_k)=\frac{W_p^{full}(q,E_k)}{W_p(q,E=0)}$3

breaks that invariance and produces a self-energy anomaly. The final anomaly is

$f(q,E_k)=\frac{W_p^{full}(q,E_k)}{W_p(q,E=0)}$4

and in an ultrastatic spacetime $f(q,E_k)=\frac{W_p^{full}(q,E_k)}{W_p(q,E=0)}$5 this reduces to

$f(q,E_k)=\frac{W_p^{full}(q,E_k)}{W_p(q,E=0)}$6

This is not presented as a Hermitian one-body operator, but it is a static, geometry-dependent renormalized self-energy correction, and therefore occupies a nearby conceptual region (Frolov et al., 2013).

6. Static-source self-energy in lattice gauge theory

In lattice gauge theory, the relevant object is the static self-energy of a color source, equivalently the additive mass shift entering the heavy-quark pole mass. It is defined through the temporal Polyakov line

$f(q,E_k)=\frac{W_p^{full}(q,E_k)}{W_p(q,E=0)}$7

and extracted from

$f(q,E_k)=\frac{W_p^{full}(q,E_k)}{W_p(q,E=0)}$8

The perturbative series is

$f(q,E_k)=\frac{W_p^{full}(q,E_k)}{W_p(q,E=0)}$9

The paper computes these coefficients in lattice $1$0 gauge theory up to order $1$1 using Numerical Stochastic Perturbation Theory, with Wilson gauge action, a Langevin update, an $1$2 integrator, periodic boundary conditions in time, and twisted boundary conditions in the three spatial directions (Bali et al., 2011).

The conceptual reason this quantity is a preferred renormalon probe is its $1$3 operator-product-expansion structure. The paper contrasts this with the plaquette, whose first nontrivial operator is dimension $1$4, and states that for the static source the renormalon behavior should become visible at much lower orders, roughly “four times faster.” The predicted asymptotic behavior is

$1$5

with the ratio $1$6 approaching $1$7 up to $1$8 corrections. The numerical study reports that the onset of factorial growth is observed around $1$9, and that the order-by-order contributions reach their minimum around the predicted $q$0 for representative values of the coupling, consistent with

$q$1

The large-order behavior is found to be the same for smeared and unsmeared data, and the octet data are compatible with the same asymptotic pattern (Bali et al., 2011).

Finite-volume control is essential. The observable is fitted with a structure containing a leading $q$2 term,

$q$3

and subleading $q$4 and $q$5 corrections. The leading $q$6 effect is interpreted as an interaction with mirror charges on lattice replicas and is governed by the same infrared renormalon as the self-energy itself, making the separation of $q$7 from $q$8 necessary for isolating the $q$9 renormalon (Bali et al., 2011).

The extracted renormalon normalization is reported as $0.5$0 and $0.5$1 for smeared and unsmeared triplet data, translating to about $0.5$2 in the $0.5$3 scheme, in very good agreement with the earlier estimate $0.5$4. The practical conclusion drawn is that the pole mass cannot be determined more accurately than about $0.5$5 because of the $0.5$6 renormalon ambiguity (Bali et al., 2011).

Taken together, these usages show that “static Hermitian self-energy” is not a single universal object but a family resemblance. In its strictest and most explicit sense, it is a frequency-independent Hermitian effective self-energy operator for quasiparticle calculations. In broader usage, it can describe the finite static residual left by Hermitizing constrained dynamics, or, descriptively, the additive self-energy of an infinitely heavy source. What unifies these cases is the replacement of a more complicated dynamical, geometric, or asymptotic structure by a static energy correction with a well-defined effective meaning.