Pith. sign in

REVIEW 4 major objections 4 minor 24 references

This paper proves that in one-dimensional non-Hermitian photonic crystals, the winding number of a dispersion curve equals a transfer-matrix eigenvalue count, and this invariant determines whether edge modes localize on the left or right bo

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-02 17:35 UTC pith:TUXE7BIG

load-bearing objection Transfer-matrix index for continuous non-Hermitian photonic crystals is a genuine contribution, but Example 3 does not verify the path assumption that the key theorems rely on. the 4 major comments →

arxiv 2603.22515 v2 pith:TUXE7BIG submitted 2026-03-23 math-ph math.MP

Spectral topology and edge modes for one-dimensional non-Hermitian photonic crystals

Junshan Lin , Hai Zhang This is my paper
classification math-ph math.MP MSC 35P0578A4078A45
keywords non-Hermitian photonic crystalsphotonic skin effectspectral winding numbertransfer matrixedge modesbulk-edge correspondencespectral reciprocity
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes a rigorous bulk-edge correspondence for the photonic skin effect in continuous one-dimensional media, where the Toeplitz-matrix theory used for discrete lattices does not apply. Its central claim is that a transfer-matrix eigenvalue count — half the difference between the number of eigenvalues with modulus below one and the number above one — is constant on each component of the complement of the dispersion curves and equals the winding number of the enclosing dispersion loop. When that count is +1, every non-branch-point frequency inside the loop is an edge mode of the semi-infinite crystal on the right, decaying exponentially; when it is -1, the same holds on the left. This yields a criterion, read directly off the transfer matrix, for where and whether photonic skin modes appear in continuous periodic media.

Core claim

The central discovery is the equality Ind(D_i_n;1)=W(γ_n;ω_B): the new index, defined from the four transfer-matrix eigenvalues, takes the same value as the winding number of the nth dispersion curve around any point in that component. The paper proves that the index is locally constant on gapped components, that the unbounded principal component has index 0 by homotopy from a Hermitian reference crystal, and that crossing a positively oriented loop changes the index by +1 (negatively oriented, -1). It then constructs the edge mode: for index +1, three transfer-matrix eigenvectors lie inside the unit circle, and a linear combination chosen to match the outgoing boundary condition at z=0 deca

What carries the argument

The transfer matrix M(ω) for one period: it evolves the four-component field across a unit cell, and its eigenvalues λ_j(ω) are the Bloch phase factors e^{ik}, with product 1. The new index Ind(ω)=1/2(#{|λ_j|<1} - #{|λ_j|>1}) is constant on every connected component of the complement of the dispersion curves; Theorem 9 pins the outer component to 0 via a Hermitian limit, Theorem 12 shows the jump across a loop equals its winding number, and Theorem 14 uses the sign of Ind to select the decaying subspace of M(ω) whose linear combination matches the outgoing boundary condition, producing an exponentially decaying edge eigenfunction. The A/F layered stack — anisotropic A layers plus magnetized

Load-bearing premise

The argument needs a continuous path g(t) that avoids the spectrum while deforming back to a Hermitian reference crystal — assumed, not verified, for the Example 3 stack — plus the separate non-intersection of dispersion curves (partially relaxed by Proposition 13); if either fails, the index value, and with it the predicted edge side, can change.

What would settle it

For the three-layer A/F stack of Example 3 (δ=6, φ1=0, φ2=0.8, α=β=0.5, ε0=13+5i), pick a non-branch-point frequency inside one of the closed dispersion loops and compute the four transfer-matrix eigenvalues λ_j(ω). If 1/2(#{|λ_j|<1} - #{|λ_j|>1}) does not equal the winding number of that loop around ω, the central equivalence Theorem 12 fails. Alternatively, solve the semi-infinite problem on I+ at such a frequency; if no exponentially decaying eigenfunction exists despite the index being +1, the bulk-edge statement fails.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • For a frequency inside a positively wound dispersion loop that is not a branch point, the semi-infinite crystal on the right supports an exponentially localized edge mode; inside a negatively wound loop, the left crystal does (Theorem 14).
  • The winding number of a dispersion curve can be computed from the transfer-matrix eigenvalue count without first mapping the full Bloch band, since Ind(D)=W(γ;ω_B) (Theorem 12).
  • Under the rank condition in Proposition 15, the number of edge modes equals the index magnitude: one for |W|=1, two for |W|=2, so the correspondence gives a counting statement, not just existence.
  • When different dispersion curves intersect, the conclusion still holds away from the other bands' enclosed regions (Proposition 13 and Remark 5), so the correspondence is stable under band crossings.
  • The edge-mode construction carries over to perfect-electric or perfect-magnetic conducting terminations (Remark 4), so the prediction is robust to the choice of boundary condition.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • A cheap numerical check is to fix a point inside one of the plotted loops of Example 3 and evaluate Ind from the four roots of det(M(ω)-λI); the paper does not report such a direct verification, and this would confirm or refute Theorem 12 for that stack.
  • The proof of Theorem 9 needs a spectral-avoiding path g(t) from a Hermitian reference crystal to the target component; for the A/F stack this path is not exhibited, so finding it explicitly or showing that band touchings block it would settle whether the outer index is always 0.
  • Because the invariant only uses the relative positions of the four transfer-matrix eigenvalues to the unit circle, it should be computable directly from the coefficients of the quartic characteristic equation, suggesting an algebraic criterion for point-gap membership.
  • The same transfer-matrix construction should apply to other continuous 1D non-Hermitian mechanisms (gain/loss, imaginary gauge potentials), since only periodicity and det M=1 are used; testing one such system would show whether the bulk-edge correspondence is a general feature of continuous wave models rather than specific to photonic stacks.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. The paper develops a transfer-matrix framework for the non-Hermitian skin effect in one-dimensional continuous photonic crystals. For the Bloch eigenvalue problem with quasi-periodic boundary condition, the transfer matrix M(ω) has four eigenvalue branches λ_j(ω). The authors define Ind(ω)=1/2(#{|λ_j|<1}−#{|λ_j|>1}), prove it is locally constant on each connected component of the complement Γ^c of the dispersion curves, and study its jump across a dispersion curve. Under a homotopy from a reference Hermitian crystal and the existence of a spectral-avoiding path (Theorem 9), the exterior region is assigned index 0; crossing a positively oriented loop changes the index by +1 and a negatively oriented loop by −1, so that Ind(D_i^n)=W(γ_n;ω_B). Theorem 14 then asserts that every non-branch-point frequency in a bounded component with index +1 is an eigenvalue of the semi-infinite problem on I+ with an exponentially decaying eigenfunction, and similarly for index −1 on I−. The paper includes a concrete A/F layered example and explicit transfer-matrix formulas.

Significance. If correct, this is a valuable rigorous bulk-edge correspondence for continuous 1D non-Hermitian photonic crystals, where Toeplitz spectral theory does not apply. The transfer-matrix index is a natural invariant that directly controls the decay of the physical solution and is explicitly computable from the 4×4 monodromy matrix. The paper is well organized, and the local jump argument in Theorem 12 is plausible and internally consistent. The inclusion of explicit transfer matrices for the A and F layers and a concrete three-layer example is a strength. However, the main topological input—the existence of the spectral-avoiding path and the principal-gap assignment for the exterior region—is not verified for the paper's flagship example, and the non-Hermitian band-labeling assumption is asserted rather than proved. These issues make the central claim conditional and require substantial revision.

major comments (4)
  1. [§3.2, Theorem 9; Example 3] The conclusion Ind(D;1)=0 for the exterior component is not established for the example used throughout the paper. Theorem 9 requires a continuous curve g:[0,1]→C with g(t)∈Γ^c_t and g(1)∈D; this spectral-avoiding path is never constructed or checked for the A/F stack of Example 3. Figure 1 (right) shows only the endpoint loops, and the text gives no information about how the point-gap loops are born during the deformation. If they nucleate through a band touching on the real axis, such a path may be blocked, and Ind(D) need not be 0. Since Theorems 12 and 14 use Ind(D)=0 as the base value, a shift in this value would change every Ind(D_i^n) by a constant and can reverse or remove the predicted side of localization. Please verify the existence of the path for the example, or state the existence of the spectral-avoiding path as an explicit assumption in the main theorems; Assumption 7(ii)
  2. [§2.3, band-labeling assumption] The paper assumes that the complex eigenfrequencies can be ordered by increasing real part with each ω_n(k) continuous in k and 'no ambiguity'. For non-Hermitian operators, eigenvalue curves can cross, and the real-part ordering can change discontinuously; without a separation condition or a precise continuation argument, the objects γ_n and their winding numbers are not well-defined. This is load-bearing for the definitions of point gaps and for Theorems 11 and 12. The authors should either prove that the stated ordering is well-defined for the class of media considered, or add it as an explicit hypothesis and verify it for Example 3.
  3. [§3.3, Theorem 12 and Proposition 13] Theorem 12 assumes γ_n∩γ_m=∅ and the existence of a smooth point where ω_n is analytic with a well-defined outward normal. None of these conditions is verified for the dispersion curves in Example 3. The plot in Figure 1 is not enough to exclude intersections or self-tangencies; the jump argument requires a neighborhood where the only relevant curve is γ_n and where ω_n(ξ) is locally injective. Proposition 13 relaxes the non-intersection condition only partially: it still requires the smooth-point and analyticity assumptions and only establishes the index on the punctured domain D_i^n\∪_{m≠n}D_m. Please check these hypotheses for the numerical example or provide a version of the theorem that can be applied directly to it.
  4. [§3.4, Theorem 14 and sign of the radiation condition] In equation (25), the outgoing condition is c2=Q c1 for z>0 and c2=−Q c1 for z<0. The matrix B defined in (30) is H+QE, which vanishes for the z<0 condition but not for the z>0 condition. The proof of part (i) (I+) therefore appears to use the wrong sign unless the ambient medium is placed on the opposite side than the named interval; for part (ii) the sign should be changed accordingly. This is easily fixed, but as written it is confusing and affects the boundary-value problem that Theorem 14 claims to solve. The authors should also state explicitly why the nonzero α in (28) gives a nontrivial solution Φ; at non-branch points with distinct λ_j this follows from linear independence of the eigenvectors, but the argument should be written out.
minor comments (4)
  1. [§2.3 and Example 3] The text refers to 'Figure 2 (Right)' for the non-Hermitian band structure of Example 3; this should be Figure 1 (Right). Similarly, Example 3 refers to 'Figure 2 (Left)' for the Hermitian bands; this should be Figure 1 (Left).
  2. [Throughout] There are several typos: 'eigenfrquency' in §2.1, 'phtononic' in Assumption 7, and the use of 'attain' where 'has' is meant. A careful proofreading pass is needed.
  3. [§3.3, Theorem 11] Theorem 11 is stated for a general closed curve γ_n, but the proof assumes that each bounded component D_i^n has a simple closed boundary γ_i^n and that W(γ_j^n;ω_B)=0 for j≠i. If γ_n self-intersects or has multiple loops, these points require additional justification; Remark 2 gestures at this but the theorem statement should carry the same qualification.
  4. [Appendix A] The displayed formula for the discriminant of the quartic is hard to parse and appears to contain a typo: Δ = − Δ1−4Δ2_0/27. Please check and correct this expression.

Circularity Check

0 steps flagged

No load-bearing circularity: Ind=W is proved, not assumed; the unverified path in Example 3 is a correctness gap, not a circular reduction.

full rationale

The central derivation is self-contained. The relation Ind(D_i^n;1)=W(γ_n;ω_B) in Theorem 12 is not posited as an input: Ind is a pointwise count of transfer-matrix eigenvalues inside/outside the unit circle at a fixed ω, while W is an integral winding number of the dispersion image ω_n(T). The proof derives the equality by local conformal inversion of ω_n near a smooth boundary point and by tracking which eigenvalue branch crosses |λ|=1; Theorem 11 independently relates orientation to W. Theorems 12 and 14 then use this equality as a theorem hypothesis, not as a restatement of a fitted or assumed quantity. There are no fitted constants, no subset of data being 'predicted', and no ansatz imported by citation. The only self-citation is [16] in Remark 1 for the scalar relation tr(M(ω))=2cos(k); that fact is standard and is not load-bearing for the main theorem. A genuine gap is that Example 3 does not exhibit a spectral-avoiding path g(t) required by Theorem 9, nor verify γ_n∩γ_m=∅; but this is a hypothesis-checking/correctness issue, not circularity, because the theorem's conclusion is not assumed in constructing the model, the transfer matrix, or the invariant.

Axiom & Free-Parameter Ledger

0 free parameters · 7 axioms · 0 invented entities

The central claim rests on standard Floquet-Bloch and complex-analysis tools, plus several domain assumptions about band labeling and spectral separation. The most fragile premise is the existence of the spectral-avoiding homotopy path in Theorem 9, which is not established for the concrete A/F stack. No free parameters are fitted to data, and no new physical entities are introduced.

axioms (7)
  • standard math Floquet-Bloch spectral theory: for Hermitian media the quasi-periodic Maxwell operator has countable real eigenvalues arranged in continuous bands.
    Invoked in §2.2.1 to define the band structure and spectral gaps of the reference Hermitian medium.
  • domain assumption For non-Hermitian media, the eigenfrequencies can be ordered by real part and the dispersion curves ω_n(k) are well-defined continuous closed curves without band-label ambiguity.
    Stated in §2.3 as an assumption ('no ambiguity to define the dispersion curves'); not proven for general complex spectra or near exceptional points.
  • domain assumption Assumption 7: the reference Hermitian photonic crystal has a band gap and a point ω0 on the spectrum with exactly two distinct unit-modulus transfer-matrix eigenvalues.
    Used in Lemma 8 to prove the principal component has Ind=0; not derived from the layer parameters.
  • ad hoc to paper Existence of a continuous curve g(t)∈Γ^c_t connecting the Hermitian reference gap to the target component D (Theorem 9(ii)).
    This spectral-avoiding path is a strong topological premise required to carry Ind=0 from the Hermitian reference to the non-Hermitian crystal; it is not verified for Example 3.
  • domain assumption The dispersion curves γ_n and γ_m do not intersect for m≠n, and ω_n is analytic at the chosen boundary point (Theorem 12).
    Assumed in the jump theorem; the authors only partially extend to intersecting curves via Proposition 13 and Remark 5.
  • domain assumption In Theorem 14, ω is not a branch point of the transfer-matrix eigenvalues.
    Exceptional points are excluded from the edge-mode construction; behavior at branch points is not treated.
  • domain assumption The ambient medium is vacuum and the boundary condition is the outgoing radiation condition (24b)-(25).
    Defines the semi-infinite eigenvalue problem; other boundary conditions are mentioned in Remark 4 but not developed.

pith-pipeline@v1.3.0-alltime-deepseek · 16665 in / 18604 out tokens · 184793 ms · 2026-08-02T17:35:38.388895+00:00 · methodology

0 comments
read the original abstract

This work investigates edge modes in non-Hermitian photonic crystals with broken spectral reciprocity. In such systems, the spectra of the underlying operators generally form closed loops over the complex plane with nontrivial spectral topology, which gives rise to the so-called skin effect characterized by edge modes localized at interfaces. For discrete lattice models, the skin effect can be understood through the spectral theory of Toeplitz matrices. However, this mathematical framework no longer applies to continuous wave models, where finite-dimensional approximations break down. In this work, we employ a transfer matrix approach to describe wave propagation in one-dimensional periodic media and introduce a new spectral topological invariant based on the eigenvalues of the transfer matrix. The new topological invariant is equivalent to the winding number of the non-Hermitian spectrum and it enables the characterization of edge modes in one-dimensional non-Hermitian photonic crystals. The mathematical theory provides the theoretical foundation for the skin effect in continuous wave models.

Figures

Figures reproduced from arXiv: 2603.22515 by Hai Zhang, Junshan Lin.

Figure 1
Figure 1. Figure 1: Band structures of the three-layer periodic media which attain spectral asymmetry. The permittivity and permeability are given in the form of (17), with δ = 6, φ1 = 0, φ2 = 0.8, α = β = 0.5. Left: Hermitian photonic crystal with ε0 = 13, ε˜0 = 1; Right: Non-Hermitian photonic crystal with ε0 = 13 + 5i, ε˜0 = 1. . 2.3 Band structure of non-Hermitian photonic crystals When ε(z) ̸= ε ∗ (z) or µ(z) ̸= µ ∗ (z),… view at source ↗
Figure 2
Figure 2. Figure 2: Point gap (left) and line gap (right) for the spectrum of non-Hermitian operators. . Similar to Hermitian photonic crystals, we distinguish two types of non-Hermitian photonic crystals: (i) The spectral reciprocity holds such that ωn(k) = ωn(−k) for all n. (ii) The spectral reciprocity is broken with ωn(k) ̸= ωn(−k) for some n. Again the spectral reciprocity holds when the symmetry group of the period medi… view at source ↗
Figure 3
Figure 3. Figure 3: Spectrum of non-Hermitian operators over the complex plane: a trivial gap when ωn(k) = ωn(−k) (left) and a non-trivial gap when the spectral symmetry is broken (right). . 3 Spectral topology and edge modes in non-Hermitian photonic crystals In this section, we investigate the edge modes in semi-infinite photonic crystals for which the dispersion curves for the quasi-periodic problem (7) attain point gaps. … view at source ↗
Figure 4
Figure 4. Figure 4: A schematic plot of dispersion curves γ1, γ2, γ3, · · · . . For the simple closed curve γ i n , we say that it is oriented counterclockwise (or positively oriented) if the mapping ωn preserves the counterclockwise orientation of the unit circle T for ωn(ξ) ∈ γ i n . More precisely, let ξ0 = (cos(k0),sin(k0)) ∈ T be such that ωn(ξ0) is a smooth point of γ i n . Let τ = (− sin(k0), cos(k0)) be the unit tange… view at source ↗
Figure 5
Figure 5. Figure 5: The domains enclosed by γn are not Jordan domains. . 3.4 Edge modes in semi-infinite photonic crystals with non-trivial spectral topol￾ogy We consider a semi-infinite photonic crystal sitting in the interval I− := (−∞, 0) or I+ := (0, ∞). For z ∈ I±, the medium parameter is given by (ε(z), µ(z)), and the ambient medium outside the photonic crystal is assumed to be vacuum. The corresponding spectral problem… view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

24 extracted references · 1 canonical work pages

  1. [1]

    Mathematical foundations of the non-Hermitian skin effect

    Habib Ammari et al. “Mathematical foundations of the non-Hermitian skin effect” . In: Archive for Rational Mechanics and Analysis 248.3 (2024), p. 33

    2024

  2. [2]

    Spectra and pseudo-spectra of tridiagonal k-Toeplitz matrices and the topological origin of the non-Hermitian skin effect

    Habib Ammari et al. “Spectra and pseudo-spectra of tridiagonal k-Toeplitz matrices and the topological origin of the non-Hermitian skin effect” . In: Journal of Physics A: Mathematical and Theoretical 58.20 (2025), p. 205201

    2025

  3. [3]

    Non-hermitian physics

    Yuto Ashida, Zongping Gong, and Masahito Ueda. “Non-hermitian physics” . In: Advances in Physics 69.3 (2020), pp. 249–435

    2020

  4. [4]

    Exceptional topology of non- Hermitian systems

    Emil J Bergholtz, Jan Carl Budich, and Flore K Kunst. “Exceptional topology of non- Hermitian systems” . In: Reviews of Modern Physics 93.1 (2021), p. 015005

    2021

  5. [5]

    Non-Hermitian boundary modes and topology

    Dan S Borgnia, Alex Jura Kruchkov, and Robert-Jan Slager. “Non-Hermitian boundary modes and topology” . In: Physical review letters 124.5 (2020), p. 056802

    2020

  6. [6]

    Spectral properties of banded Toeplitz matrices

    Albrecht Böttcher and Sergei M Grudsky. Spectral properties of banded Toeplitz matrices . SIAM, 2005

    2005

  7. [7]

    Ordinary differential equations with applications

    Carmen Chicone. Ordinary differential equations with applications . Springer, 2006

    2006

  8. [8]

    Non-Hermitian topology and exceptional-point geometries

    Kun Ding, Chen Fang, and Guancong Ma. “Non-Hermitian topology and exceptional-point geometries” . In: Nature Reviews Physics 4.12 (2022), pp. 745–760

    2022

  9. [9]

    An optimal bound for the spectral variation of two matrices

    Ludwig Elsner. “An optimal bound for the spectral variation of two matrices” . In: Linear Algebra and its Applications 71 (1985), pp. 77–80. doi: 10.1016/0024-3795(85)90236-1

    work page doi:10.1016/0024-3795(85)90236-1 1985

  10. [10]

    Geometry-dependent skin effects in reciprocal photonic crystals

    Zhening Fang et al. “Geometry-dependent skin effects in reciprocal photonic crystals” . In: Nanophotonics 11.15 (2022), pp. 3447–3456

    2022

  11. [11]

    Nonreciprocal magnetic photonic crystals

    Alex Figotin and I Vitebsky. “Nonreciprocal magnetic photonic crystals” . In: Physical Review E 63.6 (2001), p. 066609

    2001

  12. [12]

    Symmetry and topology in non-Hermitian physics

    Kohei Kawabata et al. “Symmetry and topology in non-Hermitian physics” . In: Physical Review X 9.4 (2019), p. 041015

    2019

  13. [13]

    Biorthogonal bulk-boundary correspondence in non-Hermitian systems

    Flore K Kunst et al. “Biorthogonal bulk-boundary correspondence in non-Hermitian systems” . In: arXiv preprint arXiv:1805.06492 (2018)

    Pith/arXiv arXiv 2018

  14. [14]

    Edge modes, degeneracies, and topological numbers in non-Hermitian systems

    Daniel Leykam et al. “Edge modes, degeneracies, and topological numbers in non-Hermitian systems” . In: Physical review letters 118.4 (2017), p. 040401

    2017

  15. [15]

    Topological phases in the non-Hermitian Su-Schrieffer-Heeger model

    Simon Lieu. “Topological phases in the non-Hermitian Su-Schrieffer-Heeger model” . In: Phys- ical Review B 97.4 (2018), p. 045106

    2018

  16. [16]

    Mathematical theory for topological photonic materials in one dimension

    Junshan Lin and Hai Zhang. “Mathematical theory for topological photonic materials in one dimension” . In:Journal of Physics A: Mathematical and Theoretical 55.49 (2022), p. 495203

    2022

  17. [17]

    Non-Hermitian topological phenomena: A review

    Nobuyuki Okuma and Masatoshi Sato. “Non-Hermitian topological phenomena: A review” . In: Annual Review of Condensed Matter Physics 14.1 (2023), pp. 83–107

    2023

  18. [18]

    Topological origin of non-Hermitian skin effects

    Nobuyuki Okuma et al. “Topological origin of non-Hermitian skin effects” . In: Physical review letters 124.8 (2020), p. 086801

    2020

  19. [19]

    Topological band theory for non-Hermitian Hamilto- nians

    Huitao Shen, Bo Zhen, and Liang Fu. “Topological band theory for non-Hermitian Hamilto- nians” . In: Physical review letters 120.14 (2018), p. 146402. 20

    2018

  20. [20]

    Non-hermitian chern bands

    Shunyu Yao, Fei Song, and Zhong Wang. “Non-hermitian chern bands” . In: Physical review letters 121.13 (2018), p. 136802

    2018

  21. [21]

    Edge states and topological invariants of non-Hermitian systems

    Shunyu Yao and Zhong Wang. “Edge states and topological invariants of non-Hermitian systems” . In: Physical review letters 121.8 (2018), p. 086803

    2018

  22. [22]

    Correspondence between winding numbers and skin modes in non-Hermitian systems

    Kai Zhang, Zhesen Yang, and Chen Fang. “Correspondence between winding numbers and skin modes in non-Hermitian systems” . In: arXiv preprint arXiv:1910.01131 (2019)

    Pith/arXiv arXiv 1910

  23. [23]

    A review on non-Hermitian skin effect

    Xiujuan Zhang et al. “A review on non-Hermitian skin effect” . In: Advances in Physics: X 7.1 (2022), p. 2109431

    2022

  24. [24]

    Nontrivial point-gap topology and non-Hermitian skin effect in photonic crystals

    Janet Zhong et al. “Nontrivial point-gap topology and non-Hermitian skin effect in photonic crystals” . In: Physical Review B 104.12 (2021), p. 125416. 21

    2021