Pith. sign in

REVIEW 2 major objections 4 minor 32 references

Logarithmic regularity of spectral measures on infinite graphs

T0 review · 2 major / 4 minor · reviewed 2026-08-02 · deepseek-v4-flash

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

Pith's one-line read For a broad class of infinite graphs, the expected spectral measure satisfies µ(I) ≤ C / ln(1/|I|) on every interval I.

desk verdict Serious, ambitious extension of Craig–Simon to unimodular graphs, but the proof of the central Lemma 1 has a real gap that propagates to Theorems 4–5. read the letter →

arxiv 2606.03006 v2 pith:5NHNQ7OR submitted 2026-06-02 math-ph math.GRmath.MPmath.PR

Charles Bordenave This is my paper · ORCID
classification math-phmath.GRmath.MPmath.PR MSC 47B8046L1005C80
keywords spectralmeasurelogarithmicHölderregularitymonotonelabellingunimodularrandomgraphgroupalgebraindicableAndersonmodelCraig–Simontheorem
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

This paper proves that the expected spectral measure of many self-adjoint operators on infinite graphs is logarithmically Hölder regular: the mass of any short interval I is at most a constant over ln(1/|I|), so in particular the measure has no atoms. The operators covered include group-algebra elements on Cayley graphs, Anderson-type tight-binding models, anisotropic percolation operators, and operators on quasi-transitive graphs. The engine is a strengthened monotone labelling method: an invariant labelling splits vertices into 'prodigy', 'level', and 'bad' classes, and an invertible transfer operator propagates eigenfunction estimates from level to level. The main hypothesis, the (a,k)-indicable condition, is that a homomorphism to Z sends one generator strictly above all others; when it holds, the paper improves earlier no-atoms theorems to a quantitative bound. This extends the classical Craig–Simon theorem from Z^d to many non-amenable groups and random settings.

What carries the argument

Monotone labelling with transfer operator. Given an invariant labelling η:V→{0,...,n-1}, a vertex is a prodigy if it has a neighbour x̂ with η(x̂)<η(x) and all other neighbours of x̂ have label <η(x); level vertices have all neighbours of label ≤η(x); the rest are bad. The proof defines T_j from prodigy vertices of label j to their parents x̂; the key requirement is that T_j be invertible with inverse norm at most 1/p*. The recursive eigenfunction estimate (Lemma 1) propagates the size of the spectral projection from level to level, yielding the interval bound. In the infinite setting, the finite-dimensional spectral count is replaced by the von Neumann dimension of the associated tracial al

What would settle it

Numerically diagonalize large finite quotients of a surface-group Cayley graph with the Anderson model (or a group-algebra element) and measure the integrated density of states on intervals of length ε; if for some ε the mass of a length-ε interval decays slower than 1/ln(1/ε) as ε→0, the claimed logarithmic regularity is false. More directly, any explicit (a,k)-indicable group with a self-adjoint p having non-vanishing p_a and a non-zero atom in µ_p would falsify Theorem 1.

Watch

Extended reading notes

Core claim

The central result is a quantitative monotone-labelling estimate: in any unimodular weighted rooted graph with an invariant labelling and deterministic bounds ∥A_G∥≤K and inf|p(x,x̂)|≥p*>0, the expected spectral measure at the root obeys E[µ_G^{δ_o}(I)] ≤ P(o∈B) + Σ_j E[µ_{G_j}^{δ_o}(I)1_{o∈L_j}] for intervals of length at most κ α^{-n}. Specializing to Cayley graphs of (a,k)-indicable groups, where the invariant labelling comes from a homomorphism φ:Γ→Z and ω mod n, this gives the explicit regularity bound µ_p(I) ≤ 2k ln(α)/(ln(β/|I|))_+ for every interval I, with α=3∥p∥_op/|p_a|, β=2√2|p_a|, and consequently no atoms. The same inequality is proved for right-invariant random arrays (Theorem

Load-bearing premise

The argument requires the graph's group to admit a homomorphism to Z under which some generator has a strictly larger image than all others (the (a,k)-indicable condition), and requires the coefficient of that generator in the operator to be bounded away from zero; if either fails, the main bound can be vacuous or false.

Editorial extensions

If this is right

  • For every self-adjoint element of the group algebra of an (a,1)-indicable group (free groups, even Artin groups, surface groups), the spectral measure has no atoms and satisfies the explicit log bound.
  • The density of states of Anderson-type models on Cayley graphs of indicable groups has logarithmic Hölder regularity for arbitrary compactly supported potential distributions, without independence or density assumptions.
  • For anisotropic percolation models in which the distinguished generator's edge is always present, the expected spectral measure is uniformly log-regular regardless of the other bond probabilities.
  • The block-labelling extension applies to quasi-transitive operators, yielding iterative bounds and, in examples, finitely many atoms contained in the spectrum of a finite matrix.

Reading between the lines

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

  • The bound depends on only the coefficient p_a and the global norm, suggesting that the spectral regularity mechanism is essentially one-dimensional: the monotone labelling along a single homomorphism is the only structure doing the work. One could try to exploit several independent homomorphisms to obtain stronger regularity (e.g., power-law) for groups with higher rank structure.
  • The phenomenon seen in the lamplighter group, where breaking the one-sidedness of the generator produces purely atomic spectra, suggests that the (a,k)-indicable condition might be close to a necessary condition for any universal logarithmic bound on Cayley graphs of indicable groups. Testing this necessity on other groups with multiple generators of equal φ-value would sharpen the picture.
  • The method may extend to operators on periodic manifolds with a co-compact group action, but the finite-trace von Neumann algebra is absent; a limiting argument using finite approximants (along the lines of Benjamini–Schramm convergence) might still yield a logarithmic regularity statement for the integrated density of states.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 4 minor

Summary. The paper develops a monotone labelling method to prove logarithmic Hölder regularity of expected spectral measures for self-adjoint local operators on unimodular random weighted graphs. The main technical results are Lemma 1 (finite graphs), Theorem 4 (unimodular graphs, vertex labelling), and Theorem 5 (block labelling). These are applied to elements of group algebras of indicable groups (Theorem 1), invariant random operators and Anderson-type models (Theorem 2), quasi-transitive operators (Theorem 3), and a construction for general indicable groups (Theorem 6). The claimed results would substantially extend Craig–Simon log-Hölder continuity and atomlessness of the density of states beyond Euclidean lattices, with explicit constants.

Significance. If the proofs were correct, the paper would be a significant advance in the spectral theory of operators on infinite graphs. The framework is elegant and unifies several models: group-algebra elements, invariant random operators, and Benjamini–Schramm limits. The constants α and β are explicit, and no parameter is fitted to data, which is a strength. The applications to Anderson-type potentials with arbitrary compactly supported distributions and to quasi-transitive graphs are natural and would be valuable. However, the central finite-dimensional lemma contains a false linear-algebra step, and this step is inherited by the infinite-dimensional theorems. Since the main results all rest on this lemma, the current manuscript does not establish its claims as written.

major comments (2)
  1. [Section 3, proof of Lemma 1] The assertion 'Since F is contained in the sum of F′, B and the L_j’s' is not justified and is false in general. Let W = B + Σ L_j. Then F′ = F ∩ W^⊥, and a general f ∈ F need not have its orthogonal projection onto W^⊥ again in F. Concretely, on the 3-vertex path 1–2–3 with labels η = 0,1,2 and unit edge weights, the eigenvalue √2 has eigenvector f = (1,√2,1). Then B = span(e_1), L_j = 0, P = span(e_2,e_3), and F′ = F ∩ B^⊥ ∩ ∩_j L_j^⊥ = 0, yet F is not contained in B. Thus the linear-algebra decomposition on which the entire proof rests is invalid. The subsequent induction only proves triviality of F∩P, i.e. of vectors in F supported entirely on prodigy vertices; it does not control the prodigy components of a general eigenvector. Since Theorem 4 and Theorem 5 repeat this step, the interval estimates in Theorems 1–3 and 6 are not established as written.
  2. [Theorem 4, proof] In the infinite-dimensional proof, F is defined as 'the vector space spanned by eigenvectors of A_G with eigenvalues in I'. For a self-adjoint operator with continuous spectrum, the spectral subspace E(I)H is not spanned by eigenvectors. The equality dim(F) = E[µ_G^{δ_o}(I)], used immediately after the definition, does not follow for the continuous part of the spectral measure. The proof should take F to be the direct-integral spectral subspace E(I)H_ρ (or prove a limiting argument), not a space of eigenvectors. As written, Theorem 4 is not proved for the continuous part of µ_G^{δ_o}, which is essential to the claimed log-Hölder regularity.
minor comments (4)
  1. [Abstract] Typo: 'spectal measure' should be 'spectral measure'.
  2. [Section 1 / Theorem 1] The convention for a singleton interval I={λ} is stated, but the formula contains ln(β/|I|). For |I|=0 the expression is undefined; the intended limiting convention (right-hand side equals 0) should be stated explicitly.
  3. [Section 3, Lemma 1] The remark that the constant 3 in α and κ is 'somewhat arbitrary' and that 'the proof gives a constant 2.09±0.01' is informal and not used; either remove it or make the purported better constant part of a theorem.
  4. [Section 4.3 / Remark 2] The name 'Grigorchuk and ˙Zuk' contains a typographical artifact; use 'Grigorchuk and Żuk'.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the main estimate is derived directly from the stated geometric and non-degeneracy hypotheses; self-citations are contextual rather than load-bearing.

full rationale

The paper's derivation chain is not circular. The interval regularity estimates in Theorems 1-3 are consequences of Lemma 1 and Theorem 4, which are proved in Section 3 using Definition 2, the transfer operators T_j, the lower bound |p(x,hat x)| >= p_*, and the von Neumann dimension. The constants alpha = 3K/p_* and beta = 2*sqrt(2) p_* are explicit functions of the assumed operator norm and coefficient lower bound; no parameter is fitted to a data subset and then renamed a prediction. The geometric condition '(a,k)-indicable' is an input hypothesis, not the conclusion, and the proof actually supplies the monotone labelling construction rather than taking it as an unproved black box. The self-citations to [8] identify the source of the monotone labelling method, but the present paper states Definition 2 and carries out the finite-dimensional and von Neumann-dimension proofs, so the central claim does not reduce to the self-citation. Theorem 6 constructs coefficients satisfying an explicit invertibility condition via Theorem 5; it does not import an unproved uniqueness or ansatz. The only substantive concern is a proof gap in Lemma 1: the sentence 'Since F is contained in the sum of F', B and the L_j's' is not a valid linear-algebra consequence in general, because a vector in F can have nonzero prodigy components even when F ∩ P = 0. That is a correctness risk in the written proof, not a circularity: even if the step fails, the claimed theorem is not equivalent to its own assumptions by definition. Therefore no circular step is identified and the circularity score is 0.

Assumptions & free parameters 1 free parameters · 6 assumptions · 0 invented entities

The ledger contains no fitted parameters and no invented physical entities: the argument is a theorem chain from stated hypotheses. The input costs are: (i) the unimodularity assumption that guarantees the trace (excluding non-unimodular random graphs and manifold settings, acknowledged by the paper); (ii) the (a,k)-indicable geometric condition plus the deterministic bounds K and p*; and (iii) standard von Neumann-algebra and functional-analysis background. The only hand-tuned quantity is the universal constant 3 in α, which the paper itself flags as arbitrary. Compared to the prior program [8, 13], the new cost is the interval-counting (not atom-only) version of the transfer argument and the matrix-valued block-labelling induction.

free parameters (1)
  • Universal constant 3 in α = 3K/p* = 3 (any value ≳ 2.09 works)
    Hand-chosen universal constant in the definition of α (Lemma 1 and Theorems 1–5). It is not fitted to data and does not change the qualitative logarithmic-regularity claim; the paper openly states it is 'somewhat arbitrary'. Listed for completeness.
assumptions (6)
  • domain assumption Unimodularity of the rooted marked graph (G, η, o) and the faithful normal tracial state τ(p) = E_ρ[⟨δ_o, p_G δ_o⟩] (Section 2, eq. (10))
    The method computes expected spectral measures as von Neumann traces; this restricts the scope to unimodular graphs / right-invariant random operators / Benjamini–Schramm limits. The paper itself notes in 'Some perspectives' that the manifold analogue fails because the corresponding von Neumann algebra is not of finite type.
  • domain assumption Existence of a surjective φ ∈ Hom(Γ, Z) with k = φ(a) > φ(b) for all b ∈ S\{a}, i.e. (Γ, S) is (a,k)-indicable (Definition 1)
    This is the 'natural geometric condition' of the abstract. It makes the random labelling η(x) = φ(x) + ω mod n produce enough prodigy vertices (label ≥ 2k) for the transfer argument. Section 4.3 shows that when it fails only the weaker existence theorem (Theorem 6) is available.
  • domain assumption Deterministic bounds ∥P∥_op ≤ K and |p_a(x)| ≥ p* a.s. (Theorems 2 and 4), or invertibility of the block maps with inverse norm ≤ p*⁻¹ (Theorem 5)
    Needed so the transfer operator T_j has controlled inverse; both α = 3K/p* and β = 2√2 p* are built from these constants, so dropping p* > 0 makes the bound vacuous.
  • standard math von Neumann dimension satisfies the dimension formula (12): dim(H1+H2) = dim H1 + dim H2 − dim(H1∩H2)
    Cited to [17, exercice 8.7.31] and [24, Theorem 1.12(2)]; this is what transfers the finite-graph counting lemma to the direct-integral Hilbert space H_ρ.
  • standard math Spectral-restriction lemma (Lemma 2): H_A(I)∩V ⊂ H_{PAP}(I), proved via the step-function approximation A_n → A and [29, Theorem VIII.23] together with Portemanteau
    Bridges the global spectral subspace F to the restricted operators on level vertices G_j. The continuous-spectrum case depends on weak convergence of the spectral measures µ_{P A_n P} → µ_{P A P}.
  • standard math Faithfulness of the left-regular representation λ_Γ (identification of p with λ_Γ(p), Section 1, eq. (1))
    Standard group-algebra fact used implicitly when Theorem 1 works directly with group-algebra elements.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Logarithmic regularity of spectral measures on infinite graphs." pith.science (2026). https://pith.science/paper/5NHNQ7OR

@misc{pith2026260603006,
  author       = {Pith},
  title        = {Pith review of: Logarithmic regularity of spectral measures on infinite graphs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5NHNQ7OR}},
  note         = {Machine review of arXiv:2606.03006}
}
read the original abstract

We study the regularity of spectral measures of self-adjoint operators on infinite weighted graphs in the unimodular setting. This framework encompasses operators in the group algebra of a finitely generated group, random operators whose distribution is quasi-invariant under a group action, and Benjamini--Schramm limits of operators on finite graphs. Under a natural geometric condition on the underlying graph, we prove that the expected spectral measure satisfies a logarithmic H\"older regularity estimate. The proof relies on a strengthened version of the monotone labelling method previously introduced with Sen and Vir\'ag to control the pure point part of the spectal measure. Applications include operators in group algebras of indicable groups, Anderson-type models with arbitrary compactly supported potentials on Cayley graphs, anisotropic percolation operators, and operators on quasi-transitive graphs. In particular, our results extend the classical Craig--Simon theorem beyond the Euclidean lattice.

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

32 extracted references

  1. [8]

    Bordenave, A

    C. Bordenave, A. Sen, and B. Vir´ ag. Mean quantum percolation.J. Eur. Math. Soc. (JEMS), 19(12):3679–3707, 2017

    2017

  2. [1]

    Aizenman and S

    M. Aizenman and S. Warzel.Random operators. Disorder effects on quantum spectra and dynamics, volume 168 ofGrad. Stud. Math.Providence, RI: American Mathematical Society (AMS), 2015

    2015

  3. [2]

    Aldous and R

    D. Aldous and R. Lyons. Processes on unimodular random networks.Electron. J. Probab., 12:no. 54, 1454–1508, 2007

    2007

  4. [3]

    Processes on unimodular random networks

    D. Aldous and R. Lyons. Errata to “Processes on unimodular random networks”.Electronic Journal of Probability, 22(none):1 – 4, 2017

    2017

  5. [4]

    Processes on unimodular random networks

    D. Aldous and R. Lyons. Second errata to “Processes on unimodular random networks”. Electron. J. Probab., 24:Paper No. 25, 2, 2019

    2019

  6. [5]

    Bellissard

    J. Bellissard. K-theory of C*-algebras in solid state physics. In T. C. Dorlas, N. M. Hugenholtz, and M. Winnink, editors,Statistical Mechanics and Field Theory: Mathematical Aspects, pages 99–156, Berlin, Heidelberg, 1986. Springer Berlin Heidelberg

    1986

  7. [6]

    Bordenave

    C. Bordenave. Spectrum of random graphs. InAdvanced topics in random matrices, pages 92–150. Paris: Soci´ et´ e Math´ ematique de France (SMF), 2017

    2017

  8. [7]

    Bordenave

    C. Bordenave. Sparse graphs and their Benjamini-Schramm limits: a spectral tour. arXiv:2510.10299, 2025

    arXiv 2025

Show all 32 references
  1. [9]

    Cherix and A

    P.-A. Cherix and A. Valette. On spectra of simple random walks on one-relator groups. With an appendix by Paul Jolissaint.Pac. J. Math., 175(2):417–438, 1996

    1996

  2. [10]

    Craig and B

    W. Craig and B. Simon. Log H¨ older continuity of the integrated density of states for stochastic Jacobi matrices.Commun. Math. Phys., 90:207–218, 1983

    1983

  3. [11]

    de la Harpe, A

    P. de la Harpe, A. G. Robertson, and A. Valette. On the spectrum of the sum of generators for a finitely generated group.Isr. J. Math., 81(1-2):65–96, 1993

    1993

  4. [12]

    Garza-Vargas and A

    J. Garza-Vargas and A. Kulkarni. Spectra of infinite graphs via freeness with amalgamation. Can. J. Math., 75(5):1633–1684, 2023. 22

    2023

  5. [13]

    Grigorchuk and C

    R. Grigorchuk and C. Pittet. Laplace and Schr¨ odinger operators without eigenvalues on ho- mogeneous amenable graphs.Asian J. Math., 28(2):287–316, 2024

    2024

  6. [14]

    R. I. Grigorchuk and A. ˙Zuk. On the asymptotic spectrum of random walks on infinite families of graphs. InRandom walks and discrete potential theory. Cortona 1997, pages 188–204. Cam- bridge: Cambridge University Press; Roma: Istituto Nazionale di Alta Matematica Francesco Se...

    1997

  7. [15]

    Higuchi and T

    Y. Higuchi and T. Shirai. The spectrum of magnetic Schr¨ odinger operators on a graph with periodic structure.Journal of Functional Analysis, 169(2):456–480, 1999

    1999

  8. [16]

    Jaikin-Zapirain and D

    A. Jaikin-Zapirain and D. L´ opez-´Alvarez. The strong Atiyah and L¨ uck approximation conjec- tures for one-relator groups.Math. Ann., 376(3-4):1741–1793, 2020

    2020

  9. [17]

    R. V. Kadison and J. R. Ringrose.Fundamentals of the theory of operator algebras: special topics. Vol. 4: Advanced theory - an exercise approach. Boston: Birkh¨ auser, 1992

    1992

  10. [18]

    A. J. Koll´ ar, M. Fitzpatrick, P. Sarnak, and A. A. Houck. Line-graph lattices: Euclidean and non-Euclidean flat bands, and implementations in circuit quantum electrodynamics.Commun. Math. Phys., 376(3):1909–1956, 2020

    1909

  11. [19]

    Korotyaev and N

    E. Korotyaev and N. Saburova. Magnetic Schr¨ odinger operators on periodic discrete graphs. Journal of Functional Analysis, 272(4):1625–1660, 2017

    2017

  12. [20]

    Kotowski and B

    M. Kotowski and B. Vir´ ag. Dyson’s spike for random Schroedinger operators and Novikov- Shubin invariants of groups.Commun. Math. Phys., 352(3):905–933, 2017

    2017

  13. [21]

    Linnell, B

    P. Linnell, B. Okun, and T. Schick. The strong Atiyah conjecture for right-angled Artin and Coxeter groups.Geom. Dedicata, 158:261–266, 2012

    2012

  14. [22]

    Linnell and T

    P. Linnell and T. Schick. Finite group extensions and the Atiyah conjecture.J. Am. Math. Soc., 20(4):1003–1051, 2007

    2007

  15. [23]

    P. A. Linnell. Division rings and group von Neumann algebras.Forum Mathematicum, 5(6):561–576, 1993

    1993

  16. [24]

    L¨ uck.L 2-invariants: Theory and applications to geometry andK-theory, volume 44 of Ergeb

    W. L¨ uck.L 2-invariants: Theory and applications to geometry andK-theory, volume 44 of Ergeb. Math. Grenzgeb., 3. Folge. Berlin: Springer, 2002

    2002

  17. [25]

    R. Lyons. Identities and inequalities for tree entropy.Comb. Probab. Comput., 19(2):303–313, 2010

    2010

  18. [26]

    J. C. McLaughlin.Random Walks and Convolution Operators on Free Products. PhD thesis, New York University, 1987. Order No. 8712766, ProQuest Dissertations & Theses Global. 23

    1987

  19. [27]

    Mohar and W

    B. Mohar and W. Woess. A survey on spectra of infinite graphs.Bulletin of the London Mathematical Society, 21(3):209–234, 1989

    1989

  20. [28]

    O’Donnell and X

    R. O’Donnell and X. Wu. Explicit near-fully X-Ramanujan graphs . In2020 IEEE 61st Annual Symposium on Foundations of Computer Science (FOCS), pages 1045–1056, Los Alamitos, CA, USA, Nov. 2020. IEEE Computer Society

    2020

  21. [29]

    Reed and B

    M. Reed and B. Simon. Methods of modern mathematical physics. 1: Functional analysis. New York-London: Academic Press, Inc. xvii, 325 p.$12.50 (1972)., 1972

    1972

  22. [30]

    S´ anchez-Peralta

    P. S´ anchez-Peralta. Universal localizations, Atiyah conjectures and graphs of groups.Geom. Funct. Anal., 35(3):842–876, 2025

    2025

  23. [31]

    T. Sunada. GroupC ∗-algebras and the spectrum of a periodic Schr¨ odinger operator on a manifold.Can. J. Math., 44(1):180–193, 1992

    1992

  24. [32]

    W. Woess. Context-free pairs of groups. II: Cuts, tree sets, and random walks.Discrete Math., 312(1):157–173, 2012. 24

    2012

Pith tools

Reviewed August 2, 2026 · model on record in the stance chip above.