---
title: Real Space Energy Gap (RSEG) in Stark Insulators
url: https://www.emergentmind.com/topics/real-space-energy-gap-rseg
type: topic
---

# Real Space Energy Gap (RSEG) in Stark Insulators

Real Space Energy Gap (RSEG) denotes, in its formal usage, an insulating gap generated by hybridization between localized states that are separated in real space, rather than by a conventional avoided crossing of Bloch bands at fixed crystal momentum. The term is introduced explicitly in the theory of a non-interacting fractional topological Stark insulator, where a linear potential gradient acting on topological flat bands produces Stark localization and a new real-space hybridization gap that becomes the relevant protection mechanism for the low-energy phase [2507.21501]. Earlier and parallel literatures contain closely related real-space gap constructions—phase-sensitive STM inference of a charge-density-wave gap, DOS-based gap extraction in real-space electronic-structure methods, geometry-controlled excitation gaps, and real-space topological diagnostics protected by an ordinary spectral gap—but these are not equivalent to the formal RSEG concept [1307.2282][1708.05564][1612.07713][1507.00015][2310.14204].

## 1. Formal definition in Stark-localized topological flat bands

In its precise modern sense, the RSEG is defined for a \((1+1)\)-dimensional system with one physical spatial dimension \(x\) and one synthetic adiabatic dimension \(\lambda\), described by a topological lattice Hamiltonian \(H_0(\lambda)\) supplemented by a linear Stark potential \(F\hat x\) [2507.21501]. The relevant low-energy manifold consists of \(n_b\) topological flat bands with total Chern number \(C\), isolated from higher bands by a large gap \(\Delta_l\), with \(\Delta_l \gg W_b\), where \(W_b\) is the total width of the low-energy flat-band manifold.

The Stark field reorganizes extended Bloch-band physics into ladders of localized states \(|w_j^{(n)}\rangle\) with energies
\[
\overline{E}_n + jFa,
\]
and localization lengths
\[
\xi^{(1,2)} \simeq \frac{W_b^{(1,2)}}{|F|}.
\]
In the weak-field regime, this is ordinary Stark localization. The RSEG appears only when states descending from different bands become both energetically resonant and spatially close enough to hybridize.

For the two-band case, the strong-field condition is
\[
|F| > \frac{\overline{E}_2-\overline{E}_1}{a},
\]
under which Stark states from the lower and upper bands become degenerate within a unit cell, with spatial separation
\[
\Delta x_{1,2} = \frac{\overline{E}_2-\overline{E}_1}{|F|} \le a.
\]
The real-space energy gap is then defined as the hybridization splitting
\[
\Delta^{\mathrm{RS}} = 2\left|\langle w_{j_1}^{(1)}|F\hat{x}|w_{j_2}^{(2)}\rangle\right|.
\]
This definition is the distinguishing feature of RSEG: the gap is controlled by overlap of Stark-localized wavefunctions and by their real-space separation, not by momentum-space band dispersion [2507.21501].

A central consequence is that the effective insulating structure of the system can cease to be controlled by the original internal band gap \(\Delta_s\) between the low-energy flat bands. Instead, once the field is sufficiently strong, protection is transferred to \(\Delta^{\mathrm{RS}}\), and the relevant single-particle basis becomes a set of real-space-hybridized multiband Stark states \(|{\cal W}_{j,\mu}\rangle\).

## 2. Weak-field, coexistence, and strong-field regimes

The Stark construction yields three distinct regimes. In the weak-potential regime,
\[
|F| < \Delta_s/L,
\]
the two band-derived Stark ladders remain separated across a finite system of length \(L\). Filling all Stark states derived from the lower band gives an integer topological band insulator adiabatically connected to the \(F=0\) phase, with protection by
\[
\Delta_s - |F|L.
\]
This regime is explicitly not controlled by the RSEG [2507.21501].

Between the weak- and strong-field limits lies an intermediate coexistence regime,
\[
\Delta_s/L < |F| < \frac{\overline{E}_2-\overline{E}_1}{a},
\]
in which bulk Stark states from different bands may hybridize in the interior, but unpaired integer-type sectors survive near the boundaries. The paper states that in this regime there is no globally well-defined bulk gap and no globally defined topology [2507.21501]. This point is conceptually important because it separates ordinary band-insulator protection from genuine RSEG protection.

In the strong-field regime, all bulk Stark states are paired and hybridized into multiband Stark states within each unit cell. The resulting insulating many-body state is formed by occupying one species \(\mu\) of the multiband Stark manifold,
\[
\left|\Psi_{\mu}(x_{\alpha})\right\rangle
=
\left|{\cal W}_{1,\mu},{\cal W}_{2,\mu},\dots\right\rangle.
\]
This state is not a conventional Bloch-band insulator. Its protection comes from \(\Delta^{\mathrm{RS}}\), and the phase transition is associated with closing of the RSEG rather than necessary closure of the original momentum-space bulk gap [2507.21501].

Near small \(|F|\), the supplementary analysis gives the asymptotic scaling
\[
\Delta^{\mathrm{RS}}
\sim
2|F|
\exp\left(-\frac{\Delta x_{1,2}}{\xi^{(1,2)}}\right)
=
O(|F|),
\]
with Stark wavefunctions decaying exponentially in real space. Because both the resonance distance \(\Delta x_{1,2}\) and the localization lengths scale as \(|F|^{-1}\), the exponential factor becomes effectively trivial and the RSEG scales linearly with \(|F|\) [2507.21501].

The explicit model used in the main text is a \(2/9\) commensurate generalized Aubry-André model with modulated hopping,
\[
H_0(\lambda)=\sum_{m}\left[ J+\delta J\cos\left(\lambda-\frac{4\pi}{9}m\right) \right]c_{m+1}^{\dagger}c_{m}+\mathrm{h.c.},
\]
with
\[
J=-1,\qquad \delta J=-0.98.
\]
Its lowest two bands carry
\[
C_1=-4,\qquad C_2=5,\qquad C=1,
\]
with separation from the third band
\[
\Delta_l=1.94
\]
and total bandwidth
\[
W_b=0.011.
\]
These numbers are the concrete demonstration that the flat-band manifold can be both topological and narrow enough for Stark localization to dominate [2507.21501].

## 3. Multiband Stark states, Wannier structure, and fractional pumping

The RSEG is not only a spectral quantity; it reorganizes the topology of the low-energy Hilbert space. In the projected flat-band subspace, the interpolation Hamiltonian
\[
\tilde{H}(\lambda,\alpha)=P(F\hat{x}+\alpha H_0(\lambda))P
\]
connects the finite-field multiband Stark problem to the projected position operator
\[
P\hat{x}P,
\]
whose eigenstates are the maximally localized Wannier functions (MLWFs) in the \(\alpha \to 0\) limit [2507.21501]. This makes the RSEG construction explicitly compatible with Wilson-loop and Wannier-center topology.

The Wilson loop is defined by
\[
W_{k_x^{(0)}+2\pi\leftarrow k_x^{(0)}}(\lambda)
=
\mathcal{P}\exp \left[\mathrm{i}\int_{k_x^{(0)}}^{k_x^{(0)}+2\pi}\mathbf{A}_x\,\mathrm{d}k_x\right],
\]
with non-Abelian Berry connection
\[
\mathbf{A}_x^{mn}
=
\left\langle u^m_{k_x}(\lambda) \right| \mathrm{i}\partial_{k_x} \left| u^n_{k_x}(\lambda) \right \rangle.
\]
The corresponding Wannier centers \(\phi_\mu(\lambda)\) satisfy
\[
P\hat{x}P\left|{\cal W}^{\mathrm{MLWF}}_{j,\mu}\right\rangle
=
(\phi_{\mu}+j)\left|{\cal W}^{\mathrm{MLWF}}_{j,\mu}\right\rangle,
\]
and their total flow over one \(\lambda\)-cycle obeys
\[
\sum_{\mu}\left[\phi_{\mu}(2\pi)-\phi_{\mu}(0)\right]=C.
\]
The RSEG is the condition that keeps adjacent multiband Stark states nondegenerate during this evolution [2507.21501].

For the minimal \(n_b=2\), \(C=1\) case, the adiabatic permutation is
\[
\left|{\cal W}_{j,1}\right\rangle\rightarrow\left|{\cal W}_{j,2}\right\rangle,\qquad
\left|{\cal W}_{j,2}\right\rangle\rightarrow\left|{\cal W}_{j+1,1}\right\rangle.
\]
A many-body state that occupies only one Stark species therefore does not return to itself after one \(2\pi\) cycle in \(\lambda\); it returns only after two cycles. The paper interprets this as non-interacting fractionalization protected by the RSEG, with a many-body Chern number \(C'=C\) over the extended \(2\pi n_b\) cycle and a physical fractional topological number \(C/n_b\) [2507.21501].

In the \(1/2\) fractional topological Stark insulator, the two-cycle pumped charge is quantized to
\[
C=1,
\]
so the charge pumped per single \(2\pi\) cycle is
\[
\Delta N=\frac{1}{2}.
\]
The supplement gives the analogous \(1/3\) example, in which the pumped charge is \(1/3\), \(2/3\), and \(1\) after one, two, and three cycles, respectively [2507.21501].

This establishes the role of RSEG as a protection mechanism for a phase that is neither an ordinary integer band insulator nor an interaction-driven fractional topological phase. The paper is explicit that there are no interactions, no topological order, and no spontaneous symmetry breaking. The analogy to fractional quantum Hall physics lies in multi-cycle return and fractional pumping, while the microscopic mechanism is Stark-localized interband hybridization [2507.21501].

## 4. Prehistory and adjacent real-space gap constructions

Before the term was formalized, several works developed real-space gap ideas that are conceptually adjacent to RSEG but technically distinct.

| Work | System | Gap notion |
|---|---|---|
| [1307.2282] | \(2H\)-NbSe\(_2\) | CDW gap inferred from real-space phase reversal near \(-0.7\ \mathrm{eV}\) |
| [1708.05564] | Zigzag SWCNTs | Global DOS gap from real-space \(O(N)\) TBMD eDOS |
| [1612.07713] | Disordered 2D Si\(_x\)C\(_{1-x}\) | Configuration-averaged disorder gap from TB-LMTO-vLB-ASR |
| [2006.09753] | Solid hydrogen | Gap from long-\(\tau\) decay of full Green’s function |
| [2310.14204] | Disordered Rice–Mele chain | Real-space invariant \(Q\) protected by a finite spectral gap |

In pristine \(2H\)-NbSe\(_2\), STM and STS were used to identify a charge-density-wave spectroscopic anomaly not through a conventional symmetric tunneling gap at \(E_F\), but through a \(180^\circ\) real-space phase reversal of the CDW modulation as a function of bias [1307.2282]. The phase remains aligned with a high-energy reference from \(+1.4\ \mathrm{eV}\) down to about \(-0.7\ \mathrm{eV}\), then undergoes a \(\pi\) reversal. The authors interpret this as evidence that the principal CDW-related gap lies around \(-0.7\ \mathrm{eV}\) below \(E_F\), with weak changes at the Fermi energy. The gap is therefore momentum-selective and CDW-wavevector-specific, inferred from real-space contrast inversion rather than from a direct hard gap in local \(dI/dV\) [1307.2282].

In vacancy-defected zigzag SWCNTs, the phrase “real space” refers to the computational framework rather than to a local spectroscopic observable. The electronic structure is computed by an \(O(N)\) tight-binding molecular dynamics method, and the band gap is inferred from the electronic density of states near the Fermi level [1708.05564]. The reported quantity is explicitly a global DOS-derived gap, not a spatially resolved real-space gap. Concrete gap values include \(0.13\ \mathrm{eV}\) for a divacancy-defected \((12,0)\) tube, and monovacancy-induced gap reductions from \(0.44\ \mathrm{eV}\) to \(0.07\ \mathrm{eV}\) in \((13,0)\) and from \(0.55\ \mathrm{eV}\) to \(0.09\ \mathrm{eV}\) in \((14,0)\), followed by gap reopening at divacancy [1708.05564].

A similar distinction holds for disordered 2D Si\(_x\)C\(_{1-x}\), where the gap is obtained in a real-space TB-LMTO-vLB-ASR formalism without Bloch’s theorem [1612.07713]. Here the key object is a configuration-averaged spectral gap shaped by local chemical environments and disorder fluctuations. The gap reaches a maximum near \(x=0.50\), with Table II listing \(E_g=5.835\ \mathrm{eV}\) at \(x=0.50\), and lower values on either side such as \(2.094\ \mathrm{eV}\) at \(x=0.10\) and \(5.719\ \mathrm{eV}\) at \(x=0.55\) [1612.07713]. This is a real-space disorder formalism for gap extraction, but not a local gap field and not the Stark-type RSEG.

In the finite-temperature GW space-time method, \(G\) and \(W\) are stored in real space and imaginary time, and the gap is extracted from the asymptotic decay of the full interacting Green’s function,
\[
|G(\tau)|
\rightarrow
a^{e} e^{-\xi^{e}\tau}
+
a^{h} e^{-|\xi^{h}|(\beta-\tau)}.
\]
This avoids analytic continuation and yields, for example, Si gaps of \(3.33(4)\ \mathrm{eV}\) for \(\Gamma\to\Gamma\) and \(1.42(2)\ \mathrm{eV}\) for \(\Gamma\to X\) [2006.09753]. Again, the representation is real-space, but the estimator is based on imaginary-time asymptotics rather than on a local real-space gap observable.

Finally, in disordered Rice–Mele chains without chiral symmetry, the principal innovation is a real-space topological invariant
\[
Q=\frac{1}{\pi}\arg \sum_r z_r,
\]
not an RSEG definition [2310.14204]. The paper shows numerically that when a finite spectral gap persists, \(Q\) remains sharply quantized; when disorder closes the spectral gap and fills it with localized states, \(Q\) becomes statistically indefinite. This establishes a real-space topology protected by an ordinary spectral gap, not a separate real-space energy gap [2310.14204].

## 5. Distinction from band gaps, spectral gaps, and local spectroscopic anomalies

RSEG is easily conflated with several other gap notions. The literature summarized here supports a sharper taxonomy.

First, an RSEG is not merely a conventional band gap described in real-space coordinates. In the Stark-insulator formulation, the original topological flat-band structure survives as the background \(H_0(\lambda)\), but the relevant low-energy protection shifts from the internal band gap \(\Delta_s\) to the hybridization splitting \(\Delta^{\mathrm{RS}}\) between resonant Stark-localized states [2507.21501]. The paper is explicit that the transition is associated with closing of the RSEG, not necessarily with closure of the original momentum-space bulk gap.

Second, an RSEG is not identical to a DOS gap extracted from a real-space calculation. The SWCNT study determines the energy band gap from eDOS near \(E_F\) in an \(O(N)\) TBMD framework, and the disordered siliphene study reports a configuration-averaged spectral gap from ASR electronic structure [1708.05564][1612.07713]. Both are real-space computational routes to a gap, but neither defines a localized real-space hybridization gap of the Stark type.

Third, a real-space topological invariant is not itself a real-space energy gap. The Rice–Mele invariant \(Q\) is a global quantity built from all eigenstates and the position operator, while the energetic protection in that work remains the ordinary spectral gap diagnosed by the density of states [2310.14204].

Fourth, a real-space gap signature need not be centered at the Fermi level or appear as a symmetric hard gap in local spectroscopy. In \(2H\)-NbSe\(_2\), the central observation is a phase reversal of the CDW pattern near \(-0.7\ \mathrm{eV}\), with only weak changes at \(E_F\) and no obvious direct spectral-gap signature in point STS [1307.2282]. This is not a formal RSEG, but it is an important counterexample to the common assumption that “real-space gap” must mean a local symmetric gap at the chemical potential.

A plausible implication is that the term RSEG is most precise when reserved for the Stark-localization mechanism of interband real-space hybridization, while earlier usages are better classified as real-space gap inference, real-space gap extraction, or real-space topology protected by an ordinary spectral gap.

## 6. Broader real-space gap physics

The broader literature places RSEG within a larger movement away from exclusively Bloch-band descriptions of gap phenomena. In holographic CFTs on \(\mathbb{R}_t \times \Sigma\), the lowest scalar excitation gap obeys
\[
\omega_{\min}^2 \ge \frac{\Delta^2\,\bar R_{\min}}{(D-1)(D-2)},
\]
or equivalently
\[
\Omega \equiv \omega_{\min}\mathcal R \ge \Delta,\qquad
\mathcal R^2=\frac{(D-1)(D-2)}{\bar R_{\min}},
\]
with the bound saturated for a sphere [1507.00015]. Here the gap is controlled by real-space geometry through the minimum scalar curvature \(\bar R_{\min}\), not by disorder, Stark localization, or a local DOS observable.

In nonlinear SSH-type chains with Kerr nonlinearity, the relevant real-space spectral structure is the gap between the original linear energy bands under open boundary conditions, inside which localized nonlinear stationary states and in-gap solitons appear [2006.06308]. The paper argues that these in-gap solitons can be understood as self-consistent edge states bound to nonlinearity-induced effective edges, and can occur even when the associated linear system is topologically trivial [2006.06308]. This is again not formal RSEG, but it demonstrates another route by which real-space localization and gap protection can become inseparable.

Taken together, these works suggest that “real-space gap physics” now spans at least three non-equivalent regimes: hybridization gaps between localized Stark states, spectroscopic anomalies inferred from spatial phase structure, and ordinary spectral gaps accessed through real-space formulations of disorder, nonlinearity, geometry, or Green’s-function asymptotics. Within that broader family, the RSEG proper is the most specific and technically constrained notion: a real-space hybridization gap
\[
\Delta^{\mathrm{RS}} = 2\left|\langle w_{j_1}^{(1)}|F\hat{x}|w_{j_2}^{(2)}\rangle\right|
\]
that replaces the usual band gap as the operative protection mechanism of a fractional topological Stark insulator [2507.21501].

Source: https://www.emergentmind.com/topics/real-space-energy-gap-rseg