---
title: Non-Bloch Higher-Order Construction
url: https://www.emergentmind.com/topics/non-bloch-higher-order-construction
type: topic
---

# Non-Bloch Higher-Order Construction

Searching arXiv for recent and foundational papers on non-Bloch higher-order constructions, higher-dimensional non-Bloch band theory, and related Hermitian HOTI baselines.
Search query: "non-Bloch higher-order topology generalized Brillouin zone higher dimensions non-Hermitian skin effect"
Non-Bloch higher-order construction denotes the effort to formulate higher-order boundary phenomena—especially corner-, hinge-, and other codimension-\(>1\) effects—in the non-Hermitian regime where ordinary Bloch band theory fails under open boundary conditions. In this setting, the relevant objects are not ordinary real-momentum Bloch bands but complexified momenta, generalized Brillouin zones (GBZs), biorthogonal eigenstates, open-boundary projectors, and geometry-dependent spectral data. Across the present literature, the subject is best understood as a layered program rather than a single closed formalism: Hermitian higher-order topology provides the construction logic; higher-dimensional non-Bloch band theory provides the open-boundary spectral framework; and recent work on geometry dependence, strip GBZs, and non-Bloch quantum geometry supplies the technical ingredients needed for genuinely higher-order non-Hermitian formulations [1808.08965] [2210.04412] [2212.11743] [2407.01296] [2506.22743] [2605.19272].

| Paper | Direct contribution | Relation to non-Bloch higher-order construction |
|---|---|---|
| [1808.08965] | Anti-commuting Wilsonian mass hierarchy | Hermitian HOT baseline |
| [2011.11027] | Separable product construction from 1D edge states | Bottom-up HOT baseline |
| [2210.04412] | 2D non-Bloch band theory in two restricted classes | First 2D GBZ reductions |
| [2212.11743] | Amoeba/Ronkin formulation in arbitrary dimensions | Higher-dimensional GBZ backbone |
| [2008.07237] | Higher-order NHSE and modified higher-dimensional non-Bloch logic | Codimension-\(>1\) skin physics |
| [2407.01296] | Geometry-adaptive spectral-potential theory in arbitrary dimensions | Geometry-sensitive NHSE classification |
| [2506.22743] | Strip generalized Brillouin zone (SGBZ) for 2D GDSE | Sequential non-Bloch construction |
| [2605.19272] | Open-boundary non-Bloch quantum geometry and Wannier functions | Wannier-sector ingredients |
| [2007.00549] | Non-Abelian Bloch oscillations in Hermitian BBH model | Important contrast, not true non-Bloch |

## 1. Conceptual domain and terminology

The phrase combines two ideas that must be sharply separated. “Higher-order” refers to codimension-\(n\) boundary structure: in a \(D\)-dimensional \(n\)th-order topological phase, zero modes occur on a boundary of codimension \(d_c=n\), equivalently on a manifold of dimension \(D-n\) [1808.08965]. “Non-Bloch” refers to the non-Hermitian regime in which open-boundary spectra are not described by ordinary Bloch momenta, so one replaces \(e^{ik}\) by complex variables such as \(\beta\), works on a generalized Brillouin zone, and uses left-right biorthogonal states [2210.04412] [2212.11743].

This immediately excludes a common conflation. Non-Abelian multi-band Bloch dynamics is not, by itself, non-Bloch band theory. The BBH-based study of non-Abelian Bloch oscillations is entirely Hermitian, concerns a degenerate occupied-band \(SU(2)\) Berry structure, and explicitly does not formulate a non-Bloch generalized Brillouin zone, complex momentum, skin effect, or non-Hermitian bulk-boundary correspondence [2007.00549]. Its relevance is only that it goes beyond the simplest Abelian single-band Bloch picture.

A second conceptual distinction concerns scope. Some papers study higher-order skin accumulation, i.e. macroscopic codimension-\(>1\) NHSE, whereas others provide the open-boundary geometry and Wannier machinery from which a non-Bloch higher-order topological theory could later be built. The existing literature therefore contains both direct higher-order non-Hermitian boundary phenomena and foundational constructions that stop short of full higher-order topological invariants [2008.07237] [2605.19272].

## 2. Hermitian higher-order baselines

The most explicit Hermitian baseline is the general construction principle based on anti-commuting Wilsonian masses. If a \(D\)-dimensional first-order or regular topological phase involves \(m\) Hermitian matrices and there exist \(p-1\) additional mutually anti-commuting Hermitian matrices that anti-commute with the parent set, then one may realize \(n=1,\cdots,p\) order topology with suitable symmetry-breaking Wilsonian masses [1808.08965]. The organizing relation is the codimension rule \(d_c=n\). This produces the familiar hierarchy surface \(\to\) hinge \(\to\) corner in \(D=3\), and in the nodal-loop case even a “fourth-order” phase without localized boundary zero modes because \(d_c=4>D=3\) [1808.08965].

The significance of this Hermitian construction is that it is algebraic rather than specifically Bloch-theoretic. The central mechanism is a nested domain-wall reduction generated by mutually anti-commuting masses. This suggests that, in a future non-Bloch setting, one should preserve the mass-hierarchy and boundary-reduction logic even when ordinary Bloch momentum is replaced by complexified open-boundary data. That implication is structural rather than explicit in the Hermitian paper.

A second Hermitian baseline is the separable Kronecker-sum construction. In the off-diagonal Aubry–André–Harper setting, the 2D Hamiltonian takes the form
\[
H=H_x\otimes I_y+I_x\otimes H_y,
\]
so every 2D eigenstate factorizes into a product of 1D eigenstates, and a corner state is exactly
\[
|\Psi_{\rm corner}\rangle=|\psi_{\rm edge}^x\rangle\otimes |\psi_{\rm edge}^y\rangle
\]
with energy \(E_m^x+E_n^y\) [2011.11027]. This is not non-Bloch theory, but it provides a clean directional blueprint: higher-order localization can be assembled from lower-dimensional boundary-localized building blocks. A plausible implication is that a non-Bloch higher-order construction will often need an analogous direction-by-direction assembly, with generalized non-Bloch edge solutions replacing Hermitian edge states.

The Hermitian BBH quadrupole model enters the subject in a different way. In the non-Abelian Bloch-oscillation analysis, the model is a 2D Hermitian higher-order topological insulator with non-commuting mirrors, quantized quadrupole moment, vanishing bulk polarization, and corner charges, but the dynamical focus is on degenerate-band Wilson loops, non-Abelian Berry curvature, and Wannier-center dynamics rather than on non-Hermitian non-Bloch physics [2007.00549]. This paper therefore serves as a cautionary reference: “beyond simple Bloch” does not automatically mean “non-Bloch.”

## 3. Emergence of higher-dimensional non-Bloch band theory

The first substantial extension of non-Bloch band theory beyond one dimension was achieved only in restricted settings. In two classes of 2D non-Hermitian systems, the problem can be reduced to effective 1D non-Hermitian problems. In one class, a symmetry suppresses the skin effect in one direction, so one direction remains real-momentum and the other is complexified fiberwise. In the other, the characteristic equation separates into \(x\)- and \(y\)-dependent parts, so one builds the 2D GBZ direction by direction from two 1D equal-modulus conditions [2210.04412]. The corresponding non-Bloch Chern number is defined on the generalized Brillouin zone \(T_\beta\) using biorthogonal left/right eigenvectors,
\[
C_n=\frac{1}{2\pi i}\int_{T_\beta} d{\bm k}\, {\cal B}_n({\bm k}),
\]
and restores bulk-edge correspondence for the non-Hermitian Chern insulator on a rectangle [2210.04412].

A more general arbitrary-dimensional framework was then built from the amoeba of the characteristic polynomial. For
\[
f(\boldsymbol{\beta},E)=\det[E-h(\boldsymbol{\beta})],
\]
the amoeba is the image of the zero set under the logarithmic map \(\log|\boldsymbol{\beta}|\), and the physically selected decay exponents are determined by the minimizer \(\boldsymbol{\mu}_{\min}(E)\) of the Ronkin function [2212.11743]. The higher-dimensional GBZ is then defined by
\[
\det[E-h(\boldsymbol{\beta})]=0,\qquad \log|\boldsymbol{\beta}|=\boldsymbol{\mu}_{\min}(E).
\]
In this formulation, the 1D equal-modulus rule is reinterpreted as central-hole closure of the amoeba, and the open-boundary bulk spectrum is confined to energies for which the amoeba has no central hole [2212.11743].

An alternative arbitrary-dimensional route uses spectral potential rather than direct root matching. In this framework the thermodynamic spectral potential is obtained recursively from lower-dimensional slices, and in \(d\) dimensions the result is
\[
\Phi(E)=\min\{\tilde{\Phi}_1(E),\tilde{\Phi}_2(E),\ldots,\tilde{\Phi}_d(E)\}.
\]
This geometry-adaptive theory emphasizes that higher-dimensional non-Bloch spectra are geometry-dependent, that the relevant generalized Brillouin zone is determined by minimizing complex deformations, and that net winding numbers classify NHSE into critical and non-reciprocal types [2407.01296]. Its practical importance for higher-order work is that codimension-\(>1\) localization is encoded by simultaneous nonzero inverse localization lengths \(\mu_j\) along multiple open directions.

These constructions jointly establish that higher-order non-Bloch theory cannot be a naive tensor product of 1D GBZ rules. In higher dimensions, geometry, boundary orientation, and the order in which open limits are taken become intrinsic parts of the spectral problem [2210.04412] [2212.11743] [2407.01296].

## 4. Sequential non-Bloch construction and the strip generalized Brillouin zone

The most explicit 2D geometry-sensitive construction is the strip generalized Brillouin zone (SGBZ) theory for geometry-dependent skin effect. Its central move is to recover a controlled 2D non-Bloch framework by taking two sequential 1D thermodynamic limits: first along a chosen major axis and then along a minor axis [2506.22743]. Starting from a general non-Bloch Hamiltonian \(h(\beta_1,\beta_2)\), one first forms a strip Hamiltonian \(\mathcal H(\beta_1)\), open in the minor direction and parametrized by the complex major-axis factor \(\beta_1\). For finite strip width, standard 1D non-Bloch theory yields a quasi-1D major-axis GBZ (QMGBZ). In the thermodynamic strip limit, this is replaced by a parametric minor-axis GBZ (PMGBZ) and a strip winding number \(W_{\text{strip}}(E,r)\), whose sign change selects the SGBZ [2506.22743].

The construction is explicitly directional. The paper proves that the SGBZ depends only on the major axis, not on the minor-axis choice. Different major axes can therefore generate inequivalent SGBZs, and the competition between incompatible SGBZs is identified as the origin of geometry-dependent skin effect in fully open 2D samples [2506.22743]. This is the strongest existing formal statement that a higher-dimensional non-Bloch theory must be geometry-adapted.

The same paper provides a sufficient condition for geometry-dependent skin effect in terms of non-Bloch dynamical degeneracy splitting. Along an equal-frequency contour on the SGBZ, if \(\mathrm{Im}(E)\), \(|\beta_1|\), or \(|\beta_2|\) is nonuniform, then the continuum degeneracy of non-Bloch states breaks down into a discrete set, and GDSE follows [2506.22743]. This result is especially suggestive for higher-order constructions because corner or hinge matching depends not only on equal complex energy but also on compatible decay moduli in multiple directions.

The paper does not itself construct a corner SGBZ, a nested Wilson loop, or a corner invariant. Nevertheless, its sequential logic is already “nested” in an operational sense: open one direction, construct a strip non-Bloch object, then open the second direction and study compatibility. This suggests that a non-Bloch higher-order theory in 2D may have to be formulated as a matching problem between multiple edge-adapted SGBZs rather than from a single universal 2D momentum manifold. That is an inference from the formalism rather than a theorem stated in the paper.

## 5. Higher-order skin effects and the need to modify non-Bloch theory

Higher-order non-Hermitian skin effects show directly that codimension-\(>1\) localization can be a macroscopic non-Bloch phenomenon. In two dimensions, the second-order skin effect yields \(O(L)\) corner skin modes rather than the \(O(1)\) corner zero modes of Hermitian second-order topological insulators; in three dimensions, the third-order skin effect yields \(O(L)\) corner skin modes out of \(O(L^3)\) total modes [2008.07237]. This establishes that higher-order boundary accumulation in non-Hermitian systems is not merely a perturbation of Hermitian HOTI phenomenology.

The same work demonstrates why standard 1D non-Bloch logic fails in higher dimensions. In the 2D model, ordinary first-order skin effect is suppressed along \(x\) and \(y\) separately by transposition-associated mirror symmetries, so directional 1D winding numbers vanish, yet corner skin accumulation still occurs when both directions are open [2008.07237]. To solve the open-boundary problem, the paper uses a genuinely two-dimensional ansatz with two coupled complex variables \(\beta_x,\beta_y\) and four reflected sectors. The resulting corner-skin branch is fixed not by independent 1D GBZ rules but by simultaneous conditions such as
\[
\beta_y=-\frac{\gamma}{\lambda},\qquad
E=-i(\gamma+\lambda\beta_x),\qquad
|\beta_x|=|\beta_y|<1,
\]
which produce simultaneous decay in both directions [2008.07237].

Topologically, the higher-order skin effect is tied to intrinsic non-Hermitian point-gap topology rather than to occupied-band line-gap invariants. In the 2D case, the relevant bulk invariant is a \(\mathbb Z_2\)-quantized Wess-Zumino term,
\[
{\rm WZ}[H]\in\left\{0,\frac12\right\},
\]
protected by a rotation-type symmetry, and \({\rm WZ}[H]=1/2\) for \(|\gamma/\lambda|<1\) [2008.07237]. The importance of this result is not that it already furnishes a universal higher-order non-Bloch band theory, but that it proves such a theory must incorporate multidirectional complex momenta and simultaneous boundary matching.

A later arbitrary-dimensional spectral-potential theory clarifies the status of these modes. It explicitly captures the dominant \(O(L^d)\) skin sector and classifies NHSE into critical and non-reciprocal types using net winding numbers, but it also states that subleading modes—“hybrid skin-topological modes, topological boundary states, and higher-order skin modes, which are of order \(O(L^j)\) (\(j<d\))”—are not explicitly included in the spectral potential [2407.01296]. This marks an important boundary of the present theory: higher-order non-Bloch skin physics is partly accessible, but a complete non-Bloch treatment of subleading higher-order topological modes remains open.

## 6. Non-Bloch quantum geometry, Wannier objects, and current limitations

A fully open-boundary, non-Bloch quantum-geometric framework has recently supplied a different set of ingredients. For a 1D non-Hermitian system under open boundary conditions, the real-space integrated quantum metric
\[
\mathcal{Q}^{\rm rs}_{m,xx} = -\frac{1}{N}{\rm Tr} \left\{ \hat{P}_m \left[\hat{x},\hat{P}_m\right] \left[\hat{x},\hat{P}_m\right] \right\}
\]
is shown to equal exactly the GBZ integral of the left-right non-Bloch quantum metric,
\[
\mathcal{Q}^{\rm rs}_{m,xx}\equiv \mathcal{Q}^{LR}_{m,xx}
\]
[2605.19272]. Here \(\hat P_m\) is an open-boundary biorthogonal projector, either built directly from left/right OBC eigenstates or represented in the non-Bloch basis on the GBZ.

The same work defines non-Bloch Wannier functions adapted to the skin-effect regime,
\[
|w^R_{m,R_i}\rangle = \sqrt{N} \int \frac{d\beta}{2\pi i\beta} \beta^{-R_i} |\psi^R_{m,\beta}\rangle,
\]
and proves that the non-Bloch integrated quantum metric is exactly the gauge-invariant part of their spread functional [2605.19272]. It also introduces projected-position Wannier functions by diagonalizing
\[
\hat P_m\hat x\hat P_m,
\]
which is the precise object ordinarily used to construct Wannier sectors.

This paper does not explicitly construct higher-order topology, corner states, multipole moments, or nested Wilson loops. Its value lies instead in providing the OBC-compatible projector, GBZ measure, non-Bloch Berry phase, and non-Bloch Wannier basis from which such a theory could plausibly be built [2605.19272]. A plausible implication is that a non-Bloch higher-order theory should proceed by replacing ordinary Bloch projectors and Wannier-sector constructions with biorthogonal OBC/non-Bloch analogues, then iterating projected-position constructions in multiple directions.

The main limitations of the present field follow directly from the cited works. There is still no universally accepted non-Bloch nested Wilson-loop formalism for higher-order topology, no general corner or hinge invariant on a geometry-dependent higher-dimensional GBZ, and no complete treatment of the interplay among corner topology, skin accumulation, and scale-free modes [2210.04412] [2407.01296] [2506.22743]. Just as importantly, several influential papers near this subject are not themselves non-Bloch higher-order constructions: the Hermitian mass-hierarchy and separable HOT baselines provide construction logic [1808.08965] [2011.11027]; the BBH non-Abelian Bloch-oscillation work provides a Hermitian multi-band dynamical diagnostic but explicitly not a non-Bloch generalized Brillouin-zone theory [2007.00549].

Taken together, the literature supports a precise but limited synthesis. Non-Bloch higher-order construction presently consists of three compatible layers: a Hermitian higher-order codimension-reduction logic, a higher-dimensional non-Bloch open-boundary spectral formalism, and an emerging OBC-compatible quantum geometry/Wannier language. What remains incomplete is their full integration into a universal theory of non-Bloch higher-order topological invariants and codimension-\(>1\) bulk-boundary correspondence [1808.08965] [2212.11743] [2407.01296] [2506.22743] [2605.19272].

Source: https://www.emergentmind.com/topics/non-bloch-higher-order-construction