Pith. sign in

REVIEW 3 major objections 5 minor 4 references

Convergence of supercell and superspace methods for computing spectra of quasiperiodic operators

T0 review · 3 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read This paper proves that continued-fraction supercell approximants converge to the true quasiperiodic spectrum with explicit error bounds, and that the superspace lift has the same spectrum.

desk verdict The supercell convergence theorem is a solid, citable result with explicit rates, but the superspace spectral equality is not proven as printed due to a Fourier-mode mismatch that looks like a typo. read the letter →

arxiv 2411.15906 v1 pith:SNHUYG3S submitted 2024-11-24 math.SP math-phmath.APmath.MP

Bryn Davies , Clemens Thalhammer This is my paper · ORCID
classification math.SPmath-phmath.APmath.MP MSC 35P0581Q10
keywords quasiperiodicoperatorssupercellmethodsuperspaceFloquet-BlochtheorycontinuedfractionsFibonaccitilingspectralgapsinterfacemodes
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

Quasicrystalline materials have exotic wave properties, but the differential operators that describe them have spectra that are notoriously hard to compute reliably. This paper targets the two most common computational shortcuts: the supercell method, which repeats a finite patch of the material periodically, and the superspace method, which treats the aperiodic material as a slice of a higher-dimensional periodic structure. For the supercell method it proves that, when the approximating periods come from continued fractions of the irrational parameter, the Floquet-Bloch spectra converge to the true quasiperiodic spectrum in Hausdorff distance at an explicit rate, with errors of order $(1+|\lambda|)q_l^{-1}$. A practical consequence is that band gaps that persist across supercell approximations, known as super band gaps, are guaranteed to be genuine gaps of the limiting operator. For the superspace method the paper proves that the lifted periodic operator has exactly the same spectrum, and shows that the spurious eigenvalues seen in earlier implementations depend on the choice of numerical discretisation.

What carries the argument

The engine of the supercell result is the continued-fraction approximant $\theta_l = p_l/q_l$ combined with a localised approximate-eigenfunction construction. A Bloch eigenfunction of the periodic approximant is multiplied by a smooth cut-off supported on a window of length comparable to $q_l$; the commutator with the operator is then $O(q_l^{-1})$, and the coefficient mismatch between approximants is controlled by the continued-fraction bound $|\theta_m-\theta_l| \le (q_l q_{l+1})^{-1}$. For the superspace claim the load-bearing object is the lifted operator $B_\theta = \sum_k a_k(x,y) D_\theta^k$ with $D_\theta = \partial_x + \theta\,\partial_y$, whose Fourier modes are matched to the generalised Fourier modes of the one-dimensional operator. A secondary mechanism is the trace-map recursion for Fibonacci tilings, which provides the explicit super-band-gap criterion used to certify persistent gaps from finitely many iterates.

What would settle it

Carry out the displayed mode matching directly: applying $D_\theta = \partial_x + \theta\,\partial_y$ to $e^{2\pi i(mx+\theta n y)}$ gives frequency $m + \theta^2 n$, whereas the one-dimensional operator acts on $e^{2\pi i(m+\theta n)x}$ with frequency $m + \theta n$; the two match only if $\theta\in\{0,1\}$, so this computation would refute the superspace theorem as printed and force a corrected lift.

Watch

Extended reading notes

Core claim

The paper's central claim is that two widely used computational heuristics for quasiperiodic spectra are convergent in a strong, quantitative sense. Theorem 3.4 states that if $A_\theta$ is a self-adjoint elliptic operator with quasiperiodic coefficients and $\theta_l = p_l/q_l$ is the $l$-th continued fraction approximant of $\theta$, then every point of the supercell spectrum $\sigma(A_{\theta_l})$ lies within distance $C(1+|\lambda|) q_l^{-1}$ of $\sigma(A_\theta)$, and conversely every point of $\sigma(A_\theta)$ lies within an explicitly bounded distance of $\sigma(A_{\theta_l})$ expressed as a convergent series. Hence the supercell spectra converge to the quasiperiodic spectrum in Hausdorff distance, and persistent spectral gaps in a supercell band diagram must be gaps of the limiting operator. Theorem 4.1 asserts the same spectrum for the superspace lift $B_\theta = \sum_k a_k(x,y) D_\theta^k$ on the two-dimensional torus, with $D_\theta = \partial_x + \theta\partial_y$, thereby validating the superspace method and attributing its observed spectral pollution to the numerical discretisation rather than to the lifting idea.

Load-bearing premise

The superspace part of the paper hinges on a frequency-matching identity between the original operator's modes and the plane-wave modes of its two-dimensional lift; if that identity is wrong, the claimed equality of spectra does not follow.

Editorial extensions

If this is right

  • Supercell band diagrams built from continued-fraction approximants converge to the true quasiperiodic spectrum at a quantified rate, so persistent spectral gaps seen in finite supercell plots are provably gaps of the infinite system.
  • The explicit error bound makes supercell computations a controlled numerical method: a target accuracy fixes the continued-fraction depth needed, with no guesswork about supercell size.
  • For golden-mean Fibonacci tilings the analogous convergence holds with errors controlled by Fibonacci numbers, and the trace-map criterion identifies super band gaps from finitely many iterates.
  • The superspace spectrum equals the original quasiperiodic spectrum when the lift is discretised consistently, and the spectral pollution previously reported is tied to the discretisation (notably plane-wave truncation) rather than the method itself.
  • Localised interface modes in reflection-symmetric quasicrystals persist from supercell approximants, with estimates for their eigenfrequencies and exponential decay rates.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the cut-off construction carries over to several variables, the same strategy would give supercell convergence for higher-dimensional quasiperiodic operators, with the continued-fraction rate replaced by a multi-dimensional Diophantine approximation error.
  • The mode-matching inconsistency in the superspace proof suggests a repaired formulation: taking the lifted plane-wave frequency to be $m + n$ rather than $m + \theta n$ would make the identity consistent, and the equality of spectra would then plausibly hold as stated.
  • The error bound can be read as a design rule for quasicrystal metamaterials: one can work backwards from a desired gap resolution to the required continued-fraction index, a practical step the paper leaves implicit.
  • The same localised-mode argument likely applies to other symmetry-induced interfaces, such as phase-shift or anti-reflection defects, provided the defect eigenvalue is sufficiently well separated from the approximant spectrum.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. This paper develops convergence theory for two numerical strategies for spectra of one-dimensional quasiperiodic elliptic operators. For the supercell method, Theorem 3.4 asserts an explicit Hausdorff-type rate of order q_l^{-1} when the quasiperiodic slope theta is replaced by its continued-fraction approximants p_l/q_l, and the authors give a largely self-contained argument for the key localization lemma. Theorem 4.1 claims equality between the spectrum of the quasiperiodic operator A_theta and that of its superspace lift B_theta, and Section 5 applies the supercell estimates to prove existence and decay of localised interface modes. Numerical examples for Schrödinger and Helmholtz-type operators illustrate both methods.

Significance. If fully established, Theorem 3.4 would provide an explicit convergence rate for supercell approximations of quasiperiodic spectra and would rigorously justify the common practice of reading persistent gaps from supercell band diagrams; this is a useful and timely result for the metamaterials literature. The superspace equality of Theorem 4.1 would likewise put a widely used heuristic on a firm footing. The paper’s supercell arguments are mostly elementary and self-contained, and the continued-fraction construction is a clear strength. However, the superspace proof is invalid as printed, and the interface-mode theorem in Section 5 is not supported by the arguments given; these points must be repaired before the advertised claims can be accepted.

major comments (3)
  1. [4.1, Eq. (4.6)] Equation (4.6) does not define a lift that matches the slice used in (4.5). The mode e^{2\pi i(mx+\theta ny)} is mapped by D_\theta = \partial_x + \theta \partial_y to 2\pi i(m+\theta^2 n)e^{2\pi i(mx+\theta ny)}, whereas the generalized Fourier coefficient in (4.5) has frequency m+\theta n. Thus the claim that B_\theta acts on the Fourier coefficients of ~F exactly as A_\theta acts on the generalized Fourier coefficients of ~f is false as printed. Replacing the y-frequency \theta n by n repairs the algebraic identity, but the converse inclusion also needs a spectral theory for the non-elliptic operator B_\theta, which is not supplied by Theorem 2.7. As it stands, Theorem 4.1 is not proven.
  2. [5, Theorem 5.1 and Eq. (5.2)] The interface-mode theorem is not established. The condition (5.2) uses an undefined index n and has a quantifier over lambda in sigma(A_theta) that conflicts with the fixed lambda whose separation is being assumed. More substantively, “interface eigenvalue for A_{theta_{l0}}” is never defined, and the periodic operator A_{theta_{l0}} has no eigenvalues on the whole line; some truncated or defect problem must be specified. The proof then uses Theorem 3.4 to infer the existence of a limiting eigenvalue, but Theorem 3.4 only bounds distances to the spectra sigma(A_{theta_l}) and sigma(A_theta); it does not by itself produce an isolated eigenvalue of A_theta. Finally, the appeal to [1,17,47] for periodic band-gap interface modes does not automatically transfer to the quasiperiodic limit. The authors should either give a precise statement with a complete proof or clearly label this as a conjecture.
  3. [3.1, proof of Theorem 3.4] The proof of the upward bound is not rigorous as written. After estimating a point lambda in sigma(A_{theta_{l+2}}), the text introduces a recursion d_{n-1} \le C(1+|\lambda|+d_n) q_n^{-1} + d_n with d_n called the distance from lambda to sigma(A_{theta_n}) and d_0 = 0; if d_n is the distance for a fixed lambda, then the base case d_{l+2}=0 holds only for lambda in sigma(A_{theta_{l+2}}), and the passage to arbitrary lambda in sigma(A_theta) requires the spectral-inclusiveness argument from Proposition 3.1 (cited as “Theorem 3.1”), which is not spelled out. Likewise, the first inequality “follows from Lemma 3.5 by letting m \to \infty” needs a no-spectral-pollution statement for the sequence A_{theta_m}. These gaps are likely repairable, but they are load-bearing for the central theorem.
minor comments (5)
  1. [Title] The title contains spacing typos (“supersp ace”, “opera tors”); the whole manuscript should be proofread for similar artifacts.
  2. [3.1, end of proof of Theorem 3.4] The sentence “The claim now follows from Theorem 3.1” should presumably refer to Proposition 3.1 rather than Theorem 3.1.
  3. [3.1, Eq. (3.22)] The displayed inequality compares a_k(x,\theta x) with a_k(x,\theta_l), but the surrounding argument compares A_{\theta_m} and A_{\theta_l}; the first argument should be \theta_m x.
  4. [3.3, Figure 3.1 caption] The caption says “as the length of the periodic unit cell decreases”, but q_l increases along the sequence; the intended statement is that the unit cell becomes longer.
  5. [4.2] No convergence theorem is stated for the finite-difference or plane-wave discretisations used in the numerical examples; the figures are therefore heuristic illustrations rather than rigorous consequences of Theorems 3.4 and 4.1.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: supercell and superspace convergence proofs are first-principles; self-citations are auxiliary tools, and the main caveat is a non-circular Fourier-mode error in Theorem 4.1.

full rationale

The central supercell theorem (Theorem 3.4) is derived from the spectral-theoretic approximate-eigenvector criterion (Lemma 3.2), explicit bump-function localisation, commutator estimates, and the classical continued-fraction approximation rate (Theorem 2.11). No parameter is fitted to the target spectrum and the conclusion is not assumed in the hypotheses. The tiling application uses the transfer-matrix recurrence of Kohmoto–Kadanoff–Tang [31] and the super band gap criterion of [22]; the latter is a self-citation by the first author, but it is used only to certify that certain λ lie in super band gaps, while the implication 'persistent supercell gap implies limiting gap' is exactly Theorem 3.4, so the citation is not load-bearing for the main result. The superspace section attempts a direct proof of σ(Aθ)=σ(Bθ). As printed, the proof is not correct: equation (4.6) defines ~F with y-frequency θn, while Dθ = ∂x + θ∂y sends e^{2πi(mx+θny)} to the multiplier m+θ²n, not the m+θn appearing in the reduced operator's Fourier series; the displayed 'acts like' identity therefore fails. This is a mathematical error in an omitted or invalid proof step, not a circular reduction—the claim does not follow from its own statement or from a fitted parameter. The authors do flag the related non-ellipticity of Bθ and the absence of a general Floquet theory for it, which further supports reading this as a correctness gap rather than circularity. Theorem 5.1 cites [1] together with independent references [17,47]; even if [1] is a self-citation, the isolation argument is not reduced solely to it. Overall, the derivation chain is self-contained against external benchmarks; the only issues are a minor non-load-bearing self-citation and a non-circular proof defect in the superspace theorem.

Assumptions & free parameters 0 free parameters · 9 assumptions · 0 invented entities

The paper introduces no fitted free parameters: the convergence constants C are generic and the numerical discretisations use standard mesh and plane-wave cutoffs. The main assumptions are standard results from Shubin [44], continued-fraction theory, elliptic regularity, and the specific hypotheses of the interface-mode theorem. The most paper-specific assumptions are the Davies-Morini super band gap criterion and the well-separated condition (5.2).

assumptions (9)
  • standard math Spectral theorem for self-adjoint operators (Weyl sequences characterise spectrum)
    Lemma 3.2 uses Weyl sequences to characterise the spectrum; this is standard and not proved.
  • standard math Shubin's Theorem 2.7: spectrum on CAP(R) equals spectrum on C_c^infty(R) for almost periodic elliptic operators
    Cited from [44]; used in the proof of Theorem 4.1 to identify the relevant spectrum.
  • standard math Shubin's Theorem 2.9: all operators in the hull H(A) share a common spectrum
    Cited from [44]; used to justify Sigma(A_theta) = sigma(A_theta) for irrational theta.
  • standard math Almost periodic approximation theorem (Theorem 2.2 of [44])
    Used to justify the supercell construction; cited, not re-derived.
  • domain assumption Elliptic regularity estimate ||f||_{H^k} <= C(1+|lambda|)||f||
    Used in Lemmas 3.5 and Theorem 3.3; requires smooth elliptic operators of the assumed form.
  • standard math Continued fraction error bound |theta - p_l/q_l| < 1/(q_l q_{l+1})
    Theorem 2.11 from [30]; controls the supercell convergence rates.
  • domain assumption Super band gap criterion of Theorem 3.9 from [22] (Davies-Morini)
    Stated and used in Section 3.2; co-authored by one of the present authors and not re-proven here.
  • domain assumption Periodic band-gap interface problems support only finitely many interface eigenvalues, and this persists in the quasiperiodic limit
    Invoked via [1,17,47] in the proof of Theorem 5.1 without derivation; the extension to the non-periodic limit is not demonstrated.
  • ad hoc to paper Well-separated interface eigenvalue condition (5.2)
    Explicit hypothesis in Theorem 5.1; the paper gives no evidence for how often such well-separated eigenvalues occur.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Convergence of supercell and superspace methods for computing spectra of quasiperiodic operators." pith.science (2026). https://pith.science/paper/SNHUYG3S

@misc{pith2026241115906,
  author       = {Pith},
  title        = {Pith review of: Convergence of supercell and superspace methods for computing spectra of quasiperiodic operators},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SNHUYG3S}},
  note         = {Machine review of arXiv:2411.15906}
}
read the original abstract

We study the convergence of two of the most widely used and intuitive approaches for computing the spectra of differential operators with quasiperiodic coefficients: the supercell method and the superspace method. In both cases, Floquet-Bloch theory for periodic operators can be used to compute approximations to the spectrum. We illustrate our results with examples of Schr\"odinger and Helmholtz operators.

Figures

Figures reproduced from arXiv: 2411.15906 by the authors.

Figure 2.1
Figure 2.1. Schematic of the slice along (ζ1, ζ2) ⊤ filling the unit cell densely Many fundamental results about almost periodic functions rely on the fact they can be approximated by periodic functions, as stated in Theorem 2.2, and it is the foundation of the supercell method considered here. To define the equivalent of L p spaces for almost periodic functions, we will first need to introduce the mean value functional. Defini… view at source ↗
Figure 3.1
Figure 3.1. The Floquet-Bloch spectra of a sequence of periodic approx￾imants of the Schrödinger equation (3.32). Proof. Let n be such that |xn+1|, |xn+2| > 2 + ε. Since |x1(ω)| = 2|cos (ω)| ≤ 2, we may assume without loss of generality that |xn| ≤ 2, or we reduce n by one. We then have |xn+3| ≥ |xn+2||xn+1| − |xn| ≥ |xn+2||xn+1| − |xn+2| ≥ |xn+2|(1 + ε) (3.30) and by induction it follows that |xN | > 2 for N > n. The converse … view at source ↗
Figure 3.2
Figure 3.2. The Floquet-Bloch spectra of a sequence of periodic approx￾imants of the generalised eigenvalue problem (3.34). 4.1. Theory In our example of the one-dimensional differential operator Aθ(x) = X k ak(x, θx) d k dxk , (4.1) we lift into the two-dimensional space R 2 and the new operator is Bθ = X k ak(x, y)Dk θ , (4.2) where Dθ denotes the directional derivative along the vector (1, θ) ⊤ in R 2 . Since the coeffi￾cien… view at source ↗
Figures from the paper (6 more)
Figure 2
Figure 2. Figure 2: ), the eigenvalues of the above equation should not depend on [PITH_FULL_IMAGE:figures/full_fig_p013_2.png]
Figure 4.1
Figure 4.1. Figure 4.1: Band diagrams computed using the superspace approach with the finite difference method using a rectangular mesh of char￾acteristic size h = 0.02. Hence there must exist some k ∈ R such that there exists an ε-almost eigenvector fe(x) = e ikxX m,n fmne 2πi(m+θn)x . (4.…
Figure 4.2
Figure 4.2. Figure 4.2: Two eigenmodes of the Schrodinger eigenvalue problem (3.32) computed using the superspace approach with a finite differ￾ence method and a rectangular mesh of size h = 0.02. (a) Dispersion relation of (3.32) calculated using the superspace method with N = 50 plane wav…
Figure 4.3
Figure 4.3. Figure 4.3: Spectra computed using the superspace method with the plane wave expansion method used to discretise in lifted space. (3.34) can be seen in [PITH_FULL_IMAGE:figures/full_fig_p015_4_3.png]
Figure 5.1
Figure 5.1. Figure 5.1: The localised interface modes for (5.3) decay exponentially away from the reflection-based interface. Alongside the modes the decay rate derived from the periodic approximation is shown as a dotted line. (a) ω 2 1 = 3.2403 (b) ω 2 2 = 8.1634 [PITH_FULL_IMAGE:figures…
Figure 5.2
Figure 5.2. Figure 5.2: The localised interface modes for the generalised eigenvalue problem (5.5). The decay rate derived from the periodic ap￾proximation is shown as a dotted line. where λk,min is the smaller of the two eigenvalues associated to the transfer matrix of the k th continued f…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

4 extracted references · 4 canonical work pages

  1. [48]

    Verbin Mor, Zilberberg Oded, Kraus Yaacov E, Lahini Yoav and Silberberg Yaron, ‘Observation of topological phase transitions in photonic quasicrystals’, in: Physical Review Letters110.7 (2013), p. 076403

    work page 2013

  2. [49]

    1101–1106

    WellanderNiklas, GuenneauSebastién andCherkaevElena, ‘Two-scale cut-and- projection convergence; homogenization of quasiperiodic structures’, in:Mathematical Methods in the Applied Sciences41.3 (2018), pp. 1101–1106

    work page 2018

  3. [50]

    Xia Yiwei, Erturk Alper and Ruzzene Massimo, ‘Topological edge states in qua- siperiodic locally resonant metastructures’, in:Physical Review Applied13.1 (2020), p. 014023

    work page 2020

  4. [51]

    Zolla F, Felbacq D and Guizal B, ‘A remarkable diffractive property of photonic quasi-crystals’, in:Optics Communications148.1-3 (1998), pp. 6–10. Bryn Davies Mathematics Institute, University of W arwick, Coventry CV4 7AL, UK Email address: bryn.davies@warwick.ac.uk Clemens Thalhammer ETH Zürich, Department of Mathematics, Rämistrasse 101, 8092 Zürich, S...

    work page 1998

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.