Eigenvalue growth of the discrete Hodge Laplacian across dimensions

Matthias Keller; Philipp Bartmann

REVIEW 1 major objections 2 minor 23 references

Eigenvalue growth of the discrete Hodge Laplacian across dimensions

T0 review · 1 major / 2 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 The largest eigenvalue of the discrete Hodge Laplacian never increases with dimension: $\lambda^H_k \le \lambda^H_{k-1}$ for every finite simplicial complex.

desk verdict A likely true main theorem, but the proof for the λ^+_k case misses the zero-eigenvalue case; the gap is real and patchable. read the letter →

arxiv 2608.11170 v1 pith:DWLCGUBP submitted 2026-08-11 math.CO math.GT

Philipp Bartmann , Matthias Keller This is my paper · ORCID

classification math.COmath.GT MSC 05E4555U1015A18

keywords discreteHodgeLaplaciansimplicialcomplexeigenvaluemonotonicitycohomologyvanishinglocalizationmethodAlexanderdualFormancurvaturespectralgap

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, for any finite simplicial complex with $n$ vertices, the largest eigenvalues of the up-, down-, and Hodge Laplacians in dimension $k$ are bounded above by those in dimension $k-1$. In particular $\lambda^H_k \le \lambda^H_{k-1}$, confirming O's conjecture and settling the graph Helmholtzian question $\lambda^H_1 = \lambda(G)$. The same comparison produces a lower bound on the smallest eigenvalue and, via the discrete Hodge theorem, a vanishing criterion for cohomology groups $H^k(\Sigma,\mathbb{R})$. The estimates generalize earlier results that were limited to flag complexes, and they do not require large missing faces. The proof works by viewing $\Sigma$ as a subcomplex of the complete complex and controlling the error term with constants that count missing faces.

What carries the argument

The paper uses the localization method for cochains (from [1]) adapted to the error operator $R_k = \tilde\delta_k\iota_k - \iota_{k+1}\delta_k$, which records the difference between the coboundary on the complete complex and on the subcomplex $\Sigma$. Localization decomposes a $k$-cochain $\varphi$ into vertex-localized $(k-1)$-cochains $\varphi_v$ with $(k+1)\|\varphi\|^2 = \sum_v \|\varphi_v\|^2$, and Lemma 4.3 shows $|R_k\varphi(\sigma)| = |R_{k-1}\varphi_v(\tau')|$ for $\sigma = v\tau'$. Lemma 4.4 then bounds the localized error sum $\sum_v \|R_{k-1}\varphi_v\|^2$ between $\alpha_k\|R_k\varphi\|^2$ and $\beta_k\|R_k\varphi\|^2$. These identities convert the scalar complete-complex Laplacian into an eigenvalue inequality in one dimension lower.

What would settle it

Search all graphs on five or six vertices and compute the largest eigenvalue of the edge Helmholtzian $\Delta^H_1$ and of the graph Laplacian. The paper's resolution of Problem 1.1 predicts $\lambda^H_1 = \lambda(G)$ in every case; a single graph where these two numbers differ would refute the central claim. Equivalently, a brute-force check of all simplicial complexes on five vertices for any $k$ with $\lambda^H_k > \lambda^H_{k-1}$ would settle the theorem.

Watch

Extended reading notes

Core claim

The central result, Theorem 4.1, states that for every finite simplicial complex $\Sigma$ with $n$ vertices and every $k \ge 1$, $$(k+1-\alpha_k)\,\$\lambda$^\circ_k + \alpha_k n \le (k+1)\,\$\lambda$^\circ_{k-1}$$ for $\circ \in \{+, -, H\}$, where $\alpha_k$ is one less than the minimum number of $k$-simplices of $\Sigma$ contained in a missing $(k+1)$-simplex, with $\alpha_k=0$ if no such missing simplex exists. An analogous inequality with $\beta_k$ holds for the smallest eigenvalue $\mu^H_k$. Because the Hodge Laplacian of the complete complex is $n$ times the identity, Lemma 2.1 gives $n\|\varphi\|^2 = \|\delta_k\varphi\|^2 + \|\partial_{k-1}\varphi\|^2 + \|R_k\varphi\|^2$, and from this the authors derive $\lambda^\circ_k \le \lambda^\circ_{k-1}$, together with the identity $\lambda^H_k = \lambda^-_k$. Corollary 4.5 turns the smallest-eigenvalue bound into the vanishing criterion $H^k(\Sigma,\mathbb{R}) = \{0\}$ whenever $\mu^H_0 > n\bigl(1 - \frac{1}{(k+1)!}\prod_{j=1}^k (j+1-\beta_j)\bigr)$.

Load-bearing premise

The argument depends on the fact that, in the complete complex on $n$ vertices, every Hodge Laplacian eigenvalue is exactly $n$, together with a consistent choice of signs for the coboundary maps; if that fact or the sign system failed, the proof's comparison between dimensions would break.

Editorial extensions

If this is right

O's conjecture and Problem 1.1 are settled: $\lambda^+_k \le \lambda^+_{k-1}$ for every simplicial complex, and for every graph $G$ the largest Helmholtzian eigenvalue equals the largest graph Laplacian eigenvalue $\lambda(G)$.
A new vanishing criterion follows: if the smallest eigenvalue in dimension 0 exceeds the threshold $n\bigl(1 - \frac{1}{(k+1)!}\prod_{j=1}^k (j+1-\beta_j)\bigr)$, then $H^k(\Sigma,\mathbb{R}) = \{0\}$.
The flag-complex bounds of [1] carry over to arbitrary simplicial complexes, with $\beta_k$ replacing the flag-complex bound $\beta_k \le 1$.
Alexander duality translates the main bounds into spectral estimates and a top-dimensional vanishing criterion for $H^k(\Sigma,\mathbb{R})$.
Lower bounds on $\mu^H_k$ from free faces and Forman curvature are recorded, including a higher-dimensional analogue of Fiedler's algebraic connectivity bound.

Reading between the lines

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

Extension: the eigenvalue monotonicity suggests a 'no spectral gap opening upward' principle: if a complex has a spectral gap in dimension $k$, the gap in dimension $k-1$ is at least as large; this could constrain high-dimensional expander constructions where gaps are desired in all dimensions.
Extension: because $\alpha_k$ and $\beta_k$ count only missing faces, the bounds could be tested statistically on random complexes, asking whether the inequality is typically strict and how fast $\lambda^H_k$ decays with $k$.
Extension: the same proof strategy might adapt to weighted or non-uniform simplicial complexes whenever a reference complex has scalar Hodge Laplacian; if no such reference exists, the comparison would need a more general error bound.

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.

Desk Editor's Note

read the letter

Zoe,

The paper proves the full dimensional monotonicity conjecture of O, and the main technique—localization with the missing-face parameters α_k and β_k—is a genuinely useful addition to the toolkit. The Hodge case of Theorem 4.1 is clean, and the corollaries (vanishing criteria, the λ^H_k = λ^-_k identity, the Alexander dual estimates) follow honestly from the stated machinery. This is a serious piece of work, and if the main theorem stands, it resolves a real open problem.

But the stress-test note is right: the proof of the λ^+_k case has a gap. The argument assumes that an eigenfunction of λ^+_k satisfies ∂_{k-1}φ = 0. That holds for positive eigenvalues because those eigenspaces lie in im ∂_k, where ∂_{k-1}∂_k = 0. It fails identically when λ^+_k = 0: then the 0-eigenspace is ker δ_k, and an indicator of a single k-simplex can be an eigenfunction with nonzero ∂_{k-1}φ. The inequality in that case, α_k n ≤ (k+1) λ^+_{k-1}, is not vacuous—the boundary of the tetrahedron (k=2) gives equality 12≤12—and the paper does not treat it separately. The already-proved λ^H_k version does not rescue it, because its right-hand side involves λ^H_{k-1} = λ^+_{k-2}, not λ^+_{k-1}.

This is a load-bearing hole in the written proof, but it looks patchable. A separate argument for the zero case—using that Σ_{k+1} is empty and bounding λ^+_{k-1} below via the missing faces—should work, and the example suggests the inequality is sharp. I do not see why the theorem itself would be false.

The minor issues the reader flagged are real but just as minor as stated: the missing minus sign in the Lemma 4.3 display does not affect the absolute-value lemma, and the product in Corollary 4.5 has a degenerate factor that makes the hypothesis impossible, not a mathematical error. The reliance on the folklore complete-complex identity in Lemma 2.1 is a bit heavy, but the identity is standard and cited to two sources.

Who should read this: anyone working on discrete Hodge theory, spectral gaps for simplicial complexes, or cohomology vanishing. It deserves a serious referee, but the referee should demand a fix for the zero-eigenvalue case before acceptance. I would bring it to reading group once the gap is addressed; the method is worth knowing either way.

Referee Report

1 major / 2 minor

Summary. The paper studies the spectra of the up/down/Hodge Laplacians of a finite simplicial complex Σ. The main result, Theorem 4.1, asserts that for every k ≥ 1 and o ∈ {+, −, H}, (k+1−α_k) λ^o_k + α_k n ≤ (k+1) λ^o_{k−1}, together with a lower bound (k+1−β_k) μ^H_k + β_k n ≥ (k+1) μ^H_{k−1}, where α_k, β_k are explicit combinatorial constants counting missing faces. Corollary 4.2 derives λ^o_k ≤ λ^o_{k−1}, confirming O's conjecture, and λ^H_k = λ^−_k. The proof embeds Σ into the complete complex, controls the error term R_k via localization identities, and derives vanishing criteria for cohomology groups.

Significance. If the main theorem is fully established, the paper resolves O's conjecture in full generality and gives quantitative spectral decay across dimensions with explicit, checkable constants. The paper is largely self-contained: it reproves the localization identities from [1], makes the dependence on the complete-complex identity explicit in Lemma 2.1, and formulates concrete inequalities that can be tested on examples. The main mechanism is transparent and the constants are natural in the flag-complex case. However, the proof of the λ^+_k case has a genuine gap in the zero-eigenvalue case, so the claimed theorem is not yet fully established; the result is likely true but requires a separate argument.

major comments (1)

[Section 4, proof of Theorem 4.1, λ^+_k case] The stated example in the stress-test note, the boundary of the tetrahedron, does not illustrate the gap because for that complex A_2 is empty and α_2 = 0. The star K_{1,3} is a cleaner witness to the failure of the proof's assertion in the zero-eigenvalue case.

minor comments (2)

[Lemma 4.3] The abstract contains a typo: `\mathbb{R)}` should be `\mathbb{R})`.
[Section 4, proof of Theorem 4.1] The sentence 'The proof for λ^−_k follows because λ^−_k = λ^H_k' is terse; the identity is proved in Corollary 4.2 using the λ^+_k statement. Once the zero-eigenvalue case is repaired, the authors should make explicit that this use is not circular.

Circularity Check

0 steps flagged · score 2.0 of 10

No circular derivation found: Theorem 4.1 is built from external folklore and internally proved lemmas; the main issue is a zero-eigenvalue correctness gap, not circularity.

full rationale

The derivation chain for Theorem 4.1 is not circular. Lemma 2.1 is the main external input; it is attributed to the folklore complete-complex identity nI for the Hodge Laplacian, with external citations [9] and [24], and it is used as a norm identity rather than as the target inequality. Lemmas 3.1, 4.3, and 4.4 are proved in the paper from the coboundary definitions and the relation delta_{k+1}delta_k=0. The H-case proof is an algebraic comparison using Lemma 2.1 and Lemma 4.4; the lambda_plus case repeats the same comparison using the additional positive-eigenvalue fact that partial_{k-1}phi=0; and the lambda_minus case is deduced from the already-proved H-case through Corollary 4.2, whose proof uses the independently proved lambda_plus monotonicity plus standard spectral identities, so there is no second-order cycle back to the lambda_minus statement. The self-citation [2] is used in Lemma 2.1 only as an alternative verification and in Section 5 for the Schroedinger-operator/curvature representation; Section 5 is explicitly independent of the main result, so this self-citation is not load-bearing for the central claim. A genuine correctness problem is present but is not circularity: in the lambda_plus proof, the assertion that an eigenfunction of lambda_plus_k satisfies ||partial_{k-1}phi||^2=0 fails when lambda_plus_k=0, since then phi may lie in ker delta_k rather than im partial_k; the lambda_minus_1 case also needs lambda_minus_0=lambda^H_0, which Corollary 4.2 does not state. These are gaps in the proof, not reductions of the theorem to its assumptions.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The central claim rests on the complete-complex identity, the sign-system equivalence, and two standard cited theorems. No free parameters are fitted to data and no new entities are postulated. The new alpha_k and beta_k parameters are definitions that quantify missing faces, not fitted or postulated objects.

assumptions (4)

standard math The Hodge Laplacian of the complete complex K on n vertices is n times the identity on every k-cochain space.
Invoked in Lemma 2.1 to obtain the identity n times the squared norm of phi equals the squared norm of delta_k phi plus the squared norm of partial_{k-1} phi plus the squared norm of R_k phi. Cited to [9, Lemma 8] and [24, Lemma 3.2].
domain assumption There exists a choice of coboundary signs theta(tau,sigma) in {plus or minus 1, 0} on the complete complex satisfying delta_{k+1} delta_k = 0 and unitarily equivalent to the usual oriented simplicial coboundary.
Section 2, Remark. The definitions of delta_k, partial_k, and R_k all depend on this equivalence, and Lemma 2.1 relies on the complete-complex Hodge Laplacian being n times the identity under this sign system.
standard math Alexander dual spectral identity: sigma(delta^H_k(Sigma)) without n equals sigma(delta^H_{n-k-3}(Sigma*)) without n.
Used in Section 6, Theorem 6.1 and Proposition 6.3, cited to Duval and Reiner [4, Corollary 4.7].
standard math Discrete Hodge theorem: H^k(Sigma,R) is isomorphic to the kernel of delta^H_k.
Used in Corollaries 4.5 and 6.2 to turn positivity of the smallest eigenvalue into vanishing of cohomology. Cited to Eckmann [5].

how reviews work

0 comments

Cite this review

Pith. "Pith review of Eigenvalue growth of the discrete Hodge Laplacian across dimensions." pith.science (2026). https://pith.science/paper/DWLCGUBP

@misc{pith2026260811170,
  author       = {Pith},
  title        = {Pith review of: Eigenvalue growth of the discrete Hodge Laplacian across dimensions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DWLCGUBP}},
  note         = {Machine review of arXiv:2608.11170}
}

abstract

We prove several bounds on the largest and smallest eigenvalues of the combinatorial Hodge Laplacian $\Delta^H_k$ of a finite simplicial complex $\Sigma.$ As a consequence, we obtain new vanishing criteria for cohomology groups $H^k(\Sigma,\mathbb{R)}$ and confirm a conjecture of O on the dimensional monotonicity of the largest eigenvalue.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

23 extracted references · 13 canonical work pages

[1]

Aharoni, E

R. Aharoni, E. Berger, and R. Meshulam. Eigenvalues and homology of flag complexes and vector representations of graphs.Geom. Funct. Anal., 15(3):555–566, 2005

2005
[2]

Bartmann and M

P. Bartmann and M. Keller. On Hodge Laplacians on General Simplicial Complexes.arXiv preprint arXiv:2508.07761, 2025

arXiv 2025
[4]

A. M. Duval and V. Reiner. Shifted simplicial complexes are Laplacian integral.Trans. Amer. Math. Soc., 354(11):4313–4344, 2002

work page 2002
[5]

B. Eckmann. Harmonische Funktionen und Randwertaufgaben in einem Komplex.Comment. Math. Helv., 17:240–255, 1944

work page 1944
[6]

M. Fiedler. Algebraic connectivity of graphs.Czechoslovak Math. J., 23(98):298–305, 1973

work page 1973
[7]

R. Forman. Bochner’s method for cell complexes and combinatorial Ricci curvature.Discrete Comput. Geom., 29(3):323–374, 2003

work page 2003
[8]

Garland.p-adic curvature and the cohomology of discrete subgroups ofp-adic groups

H. Garland.p-adic curvature and the cohomology of discrete subgroups ofp-adic groups. Ann. of Math. (2), 97:375–423, 1973

1973
[9]

Math., 216(2):545– 582, 2016

A.Gundert andU.Wagner.Oneigenvaluesofrandomcomplexes.Israel J. Math., 216(2):545– 582, 2016

2016

Show all 23 references

[10]

Horak and J

D. Horak and J. Jost. Spectra of combinatorial Laplace operators on simplicial complexes. Adv. Math., 244:303–336, 2013

2013
[11]

Jost and F

J. Jost and F. Münch. Characterizations of Forman curvature.arXiv preprint arXiv:2110.04554, 2021

2021 arXiv
[12]

W. Leal, G. Restrepo, P. F. Stadler, and J. Jost. Forman-Ricci curvature for hypergraphs. Adv. Complex Syst., 24(1):Paper No. 2150003, 24, 2021

2021
[13]

A. Lew. Spectral gaps, missing faces and minimal degrees.J. Combin. Theory Ser. A, 169:105127, 14, 2020

2020
[14]

A. Lew. The spectral gaps of generalized flag complexes and a geometric Hall-type theorem. Int. Math. Res. Not. IMRN, (11):3364–3395, 2020

2020
[15]

A. Lew. An eigenvalue interlacing approach to Garland’s method.arXiv preprint arXiv:2508.17279, 2025

2025 arXiv
[16]

L. Lu, Y. Shi, Z. Stanić, J. Wang, and Y. Wang. Helmholzian spectra of graphs: basic properties.arXiv preprint arXiv:2605.03478, 2026

2026 arXiv
[17]

S. O. An upper bound on the largest eigenvalue of the Helmholtzian of a graph.arXiv preprint arXiv:2606.19742v1, 2026

2026 arXiv
[18]

Oppenheim

I. Oppenheim. Local spectral expansion approach to high dimensional expanders Part I: Descent of spectral gaps.Discrete Comput. Geom., 59(2):293–330, 2018. 12

2018
[19]

Parzanchevski and R

O. Parzanchevski and R. Rosenthal. Simplicial complexes: spectrum, homology and random walks.Random Structures Algorithms, 50(2):225–261, 2017

2017
[20]

Rosenthal and L

R. Rosenthal and L. Tenenbaum. Simplicial spanning trees in random Steiner complexes. Combinatorica, 43(3):613–650, 2023

2023
[21]

Saucan, A

E. Saucan, A. Samal, and J. Jost. A simple differential geometry for complex networks. Network Science, 9(S1):S106–S133, 2021

2021
[22]

Shukla and D

S. Shukla and D. Yogeshwaran. Spectral gap bounds for the simplicial Laplacian and an application to random complexes.J. Combin. Theory Ser. A, 169:105134, 20, 2020

2020
[23]

R. P. Sreejith, K. Mohanraj, J. Jost, E. Saucan, and A. Samal. Forman curvature for complex networks.J. Stat. Mech. Theory Exp., (6):063206, 26, 2016

2016
[24]

Zhang and Y.-Z

H.-Z. Zhang and Y.-Z. Fan. Upper Bounds for the Largest Laplacian Eigenvalue of Simplicial Complexes.arXiv preprint arXiv:2606.21233, 2026. Philipp Bartmann: Institut für Mathematik, Universität Potsdam 14476 Potsdam, Germany Email address:philipp.bartmann@uni-potsdam.de Matth...

2026 arXiv

Pith tools

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

[1] [1]

Aharoni, E

R. Aharoni, E. Berger, and R. Meshulam. Eigenvalues and homology of flag complexes and vector representations of graphs.Geom. Funct. Anal., 15(3):555–566, 2005

2005

[2] [2]

Bartmann and M

P. Bartmann and M. Keller. On Hodge Laplacians on General Simplicial Complexes.arXiv preprint arXiv:2508.07761, 2025

arXiv 2025

[3] [4]

A. M. Duval and V. Reiner. Shifted simplicial complexes are Laplacian integral.Trans. Amer. Math. Soc., 354(11):4313–4344, 2002

work page 2002

[4] [5]

B. Eckmann. Harmonische Funktionen und Randwertaufgaben in einem Komplex.Comment. Math. Helv., 17:240–255, 1944

work page 1944

[5] [6]

M. Fiedler. Algebraic connectivity of graphs.Czechoslovak Math. J., 23(98):298–305, 1973

work page 1973

[6] [7]

R. Forman. Bochner’s method for cell complexes and combinatorial Ricci curvature.Discrete Comput. Geom., 29(3):323–374, 2003

work page 2003

[7] [8]

Garland.p-adic curvature and the cohomology of discrete subgroups ofp-adic groups

H. Garland.p-adic curvature and the cohomology of discrete subgroups ofp-adic groups. Ann. of Math. (2), 97:375–423, 1973

1973

[8] [9]

Math., 216(2):545– 582, 2016

A.Gundert andU.Wagner.Oneigenvaluesofrandomcomplexes.Israel J. Math., 216(2):545– 582, 2016

2016