pith. sign in

arxiv: 1907.10359 · v1 · pith:UR7KHESEnew · submitted 2019-07-21 · 🧮 math.GT

Checkerboard graph links and simply laced Dynkin diagrams

Lucas Fernandez Vilanova This is my paper

Pith reviewed 2026-05-24 18:27 UTC · model grok-4.3

classification 🧮 math.GT
keywords signed graphsDynkin diagramseigenvaluescheckerboard graph linksquasipositive linkslink signatureADE classificationfibred links
0
0 comments X

The pith

Signed graphs with all eigenvalues larger than -2 are equivalent under integer congruence of adjacency matrices to one of the ADE Dynkin diagrams.

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

The paper introduces an equivalence on signed graphs under which their adjacency matrices become congruent over the integers. It proves that any signed graph satisfying the strict eigenvalue lower bound greater than -2 must fall into this equivalence class of one of the simply laced Dynkin diagrams A_n, D_n, E_6, E_7 or E_8. The same spectral classification is then applied to checkerboard graph links, defined as a family of fibred strongly quasipositive links that contains all positive braid links. The result establishes that any such link attaining maximal signature must be isotopic to a link realized by one of the ADE diagrams. A reader would care because the eigenvalue condition supplies a concrete numerical test that forces the combinatorial type of both the graphs and the links they produce.

Core claim

We define an equivalence relation on graphs with signed edges such that the associated adjacency matrices of two equivalent graphs are congruent over Z. We show that signed graphs whose eigenvalues are larger than -2 are equivalent to one of the simply laced Dynkin diagrams A_n, D_n, E_6, E_7 and E_8. Checkerboard graph links are a class of fibred strongly quasipositive links which include positive braid links. We use the previous result to prove that a checkerboard graph link with maximal signature is isotopic to one of the links realized by the simply laced Dynkin diagrams.

What carries the argument

The equivalence relation on signed graphs making their adjacency matrices congruent over the integers, which together with the eigenvalue bound greater than -2 classifies the graphs as ADE Dynkin diagrams.

If this is right

  • Every signed graph meeting the eigenvalue condition is equivalent to an ADE Dynkin diagram.
  • Checkerboard graph links attaining maximal signature are isotopic only to the links coming from ADE diagrams.
  • Positive braid links that arise as checkerboard graph links and achieve maximal signature fall into the same isotopy classes.
  • The spectral bound supplies a numerical criterion that restricts the possible isotopy types within the larger family of fibred strongly quasipositive links.

Where Pith is reading between the lines

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

  • The same congruence relation might preserve additional link invariants such as the Seifert form or the Alexander polynomial beyond signature alone.
  • One could check whether the classification extends to signed graphs with eigenvalues at least -2 rather than strictly greater, or to other families of links.
  • The result supplies a route to compute the possible signatures of checkerboard graph links by enumerating only the ADE cases.

Load-bearing premise

Defining equivalence via Z-congruence of adjacency matrices and imposing the eigenvalue lower bound greater than -2 is enough to force every such signed graph into an ADE Dynkin diagram.

What would settle it

Exhibit a signed graph whose adjacency matrix has every eigenvalue strictly larger than -2 yet is not congruent over Z to the adjacency matrix of any ADE Dynkin diagram.

Figures

Figures reproduced from arXiv: 1907.10359 by Lucas Fernandez Vilanova.

Figure 1
Figure 1. Figure 1: An example of the Seifert surface of a positive braid link and the corresponding linking graph. The [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Forbidden minors, reading from left to right: [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 4
Figure 4. Figure 4: An example of a t-move on the vertex and edge marked with a circle where the dashed lines indicate negative edges. Let G be a signed graph and ∣G∣ the graph we obtain from G by ignoring the signs of its edges, this is usually called the underlying graph of G, a name that we adopt in this paper. Observe that the underlying graph of a signed graph that results from a t-move on G does not depend on the signs … view at source ↗
Figure 5
Figure 5. Figure 5: Graph B. Similarly, one can check the following relations [PITH_FULL_IMAGE:figures/full_fig_p005_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: The graphs (A) and (C) are t-equivalent to Dn. The graphs B, A and C are not only an instructive example, but they will also be useful in the proof of Proposition 3.1. Lemma 2.3. Let G1 be a signed graph the cycles of which are positive and let G2 be a signed graph. If G1 ∼ G2, then 2I + A(G1) ≅ 2I + A(G2). 5 [PITH_FULL_IMAGE:figures/full_fig_p005_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Dashed lines represents graphs connected to the line’s endpoints. [PITH_FULL_IMAGE:figures/full_fig_p007_7.png] view at source ↗
Figure 9
Figure 9. Figure 9: An example of ∣Gn∣ for m = 3. Notice that, if m > 6, then we can easily find an induced subgraph of type X (see [PITH_FULL_IMAGE:figures/full_fig_p008_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: The non-signed graph without the starred vertex is [PITH_FULL_IMAGE:figures/full_fig_p008_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: The non-signed graph is t ′ -equivalent to a graph that contains a T minor. If x > 5, then Gn clearly contains a minor of type D˜. Case 3: For m = 3, there are two cycles in Gn. If both of them have length 3, we can perform a t ′ -move as in [PITH_FULL_IMAGE:figures/full_fig_p009_11.png] view at source ↗
Figure 13
Figure 13. Figure 13: An example of ∣G∣ for m = 2. 9 [PITH_FULL_IMAGE:figures/full_fig_p009_13.png] view at source ↗
Figure 15
Figure 15. Figure 15: The non-signed graph is t ′ -equivalent to a graph as in the proof of Proposition 3.1, case 2. • If there is only one cycle with length 3, we need to consider either the case: or 10 [PITH_FULL_IMAGE:figures/full_fig_p010_15.png] view at source ↗
Figure 17
Figure 17. Figure 17: Examples of ∣G∣ for m = 2. Left: Gn−1 ∼t ′ E6. Right: Gn−1 ∼t ′ E7. We can construct the E8 case by simply connecting a new vertex, say v8, to v7. Case 2.1: First, consider that Gn−1 ∼t ′ E6. If m = 1 and v is connected to the vertex v1 or v6, then G ∼t ′ E7. If v is connected to the vertex v3, then X ⊂ G. If v is connected to a vertex different from those mentioned above, then G contains an induced subgr… view at source ↗
Figure 19
Figure 19. Figure 19: The graphs encircled with green contain an induced subgraph of type [PITH_FULL_IMAGE:figures/full_fig_p012_19.png] view at source ↗
Figure 21
Figure 21. Figure 21: Two graphs t ′ -equivalent to E6. Now, consider the graphs in [PITH_FULL_IMAGE:figures/full_fig_p013_21.png] view at source ↗
Figure 23
Figure 23. Figure 23: ΓA and ΓB represents checkerboard graphs connected to the dashed line’s endpoints. Proof: Recall that we can associate an abstract open book to a checkerboard graph. The goal is to show that the open books associated to the graphs Γ1 and Γ′ 1 are equivalent, i.e. let (Σ1, φ1) and (Σ ′ 1 , φ′ 1 ) be the open books associated to the graphs, then there is a diffeomorphism, h, between the surfaces Σ1 and Σ′ 1… view at source ↗
Figure 24
Figure 24. Figure 24: This drawing is a modification of a drawing in [2]. Right: four positive Hopf bands plumbed according [PITH_FULL_IMAGE:figures/full_fig_p016_24.png] view at source ↗
Figure 25
Figure 25. Figure 25: Right: four positive Hopf bands plumbed according to the graph in Figure 23 with three 2- [PITH_FULL_IMAGE:figures/full_fig_p016_25.png] view at source ↗
read the original abstract

We define an equivalence relation on graphs with signed edges, such that the associated adjacency matrices of two equivalent graphs are congruent over $\mathbb{Z}$. We show that signed graphs whose eigenvalues are larger than $-2$ are equivalent to one of the simply laced Dynkin diagrams: $A_{n}$, $D_{n}$, $E_{6}$, $E_{7}$ and $E_{8}$. Checkerboard graph links are a class of fibred strongly quasipositive links which include positive braid links. We use the previous result to prove that a checkerboard graph link with maximal signature is isotopic to one of the links realized by the simply laced Dynkin diagrams.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper defines an equivalence relation on signed graphs such that adjacency matrices of equivalent graphs are congruent over Z. It proves that any signed graph whose adjacency matrix has all eigenvalues strictly larger than -2 is equivalent to one of the simply laced Dynkin diagrams A_n, D_n, E_6, E_7 or E_8. This classification is then applied to checkerboard graph links (a class of fibred strongly quasipositive links) to show that those attaining maximal signature are isotopic to the links realized by the ADE diagrams.

Significance. If the classification holds, the result supplies a concrete bridge between the congruence classification of signed graphs under a spectral constraint and the isotopy types of a family of fibred links. The explicit identification with ADE diagrams is a strength, as it imports a well-studied combinatorial object into link theory and yields a finite list of maximal-signature examples.

major comments (2)
  1. [Main classification theorem (proof in §3)] The equivalence is defined by B = P^T A P for P in GL(n,Z) (likely §2). The eigenvalue bound > -2 is not preserved under this relation, since the transformed quadratic form is not a standard congruence of A + 2I. The proof of the main classification theorem must therefore explicitly rule out the possibility that a non-ADE congruence class nevertheless contains at least one representative with spectrum > -2. No invariant that characterizes precisely those classes admitting such a representative is identified in the abstract or the statement of the theorem.
  2. [Link application (Theorem on checkerboard graph links)] Application to checkerboard graph links (later sections): the reduction from maximal signature of the link to the eigenvalue condition on the associated signed graph is load-bearing for the isotopy claim. The manuscript must verify that the signature-maximality hypothesis indeed produces a signed graph satisfying the strict eigenvalue bound before invoking the ADE classification.
minor comments (2)
  1. [Abstract] The abstract states the two main theorems but supplies no indication of the proof strategy or the key reduction steps; a one-sentence outline of the argument would improve readability.
  2. [§2 (definitions)] Notation for signed edges and the precise definition of the adjacency matrix (including the treatment of negative edges) should be fixed early and used consistently.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading and constructive comments on the classification of signed graphs and its application to checkerboard graph links. We address each major comment below and will revise the manuscript accordingly to strengthen the exposition.

read point-by-point responses

  1. Referee: [Main classification theorem (proof in §3)] The equivalence is defined by B = P^T A P for P in GL(n,Z) (likely §2). The eigenvalue bound > -2 is not preserved under this relation, since the transformed quadratic form is not a standard congruence of A + 2I. The proof of the main classification theorem must therefore explicitly rule out the possibility that a non-ADE congruence class nevertheless contains at least one representative with spectrum > -2. No invariant that characterizes precisely those classes admitting such a representative is identified in the abstract or the statement of the theorem.

    Authors: We agree that the congruence B = P^T A P does not preserve the eigenvalues of A + 2I directly, as the referee correctly notes. The proof in Section 3 establishes the result by direct analysis: starting from a signed graph whose adjacency matrix has all eigenvalues > -2, it derives that the graph must be equivalent to an ADE diagram via reductions on the possible connected components and forbidden subgraphs. To address the concern explicitly, we will revise the proof to include a short argument showing that if a congruence class admits any representative with spectrum > -2, then it must be the ADE class (using the fact that the determinant of A + 2I is preserved up to sign under the equivalence and equals 1 only for ADE diagrams). We will also update the theorem statement to clarify that the equivalence classes in question are precisely those containing at least one representative satisfying the strict eigenvalue bound. revision: yes

  2. Referee: [Link application (Theorem on checkerboard graph links)] Application to checkerboard graph links (later sections): the reduction from maximal signature of the link to the eigenvalue condition on the associated signed graph is load-bearing for the isotopy claim. The manuscript must verify that the signature-maximality hypothesis indeed produces a signed graph satisfying the strict eigenvalue bound before invoking the ADE classification.

    Authors: We agree that the link between maximal signature and the eigenvalue condition > -2 requires explicit verification in the text. In the sections applying the classification to checkerboard graph links, the argument relies on the signature of the link being maximal if and only if the associated signed graph satisfies the spectral hypothesis. We will add a short lemma (or expand the existing discussion) that derives the strict eigenvalue bound directly from the maximality of the signature, using the relation between the signature of the Seifert form and the inertia of the adjacency matrix of the checkerboard graph. This will be inserted before invoking the main classification theorem. revision: yes

Circularity Check

0 steps flagged

No significant circularity; direct classification via congruence classes

full rationale

The paper defines an equivalence on signed graphs by Z-congruence of adjacency matrices and proves that any such graph whose adjacency matrix has all eigenvalues > -2 belongs to the congruence class of an ADE Dynkin diagram. This is a structural classification result resting on properties of quadratic forms and root systems rather than any fitted parameter, self-definitional loop, or load-bearing self-citation. The eigenvalue bound is applied directly to representatives within each class; no step reduces the claimed equivalence to a renaming or to an input that is already the output by construction. The derivation is therefore self-contained against external benchmarks in linear algebra and Lie theory.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Only the abstract is available; no explicit free parameters, axioms, or invented entities are described.

pith-pipeline@v0.9.0 · 5627 in / 1022 out tokens · 20909 ms · 2026-05-24T18:27:33.265233+00:00 · methodology

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

12 extracted references · 12 canonical work pages · 1 internal anchor

  1. [1]

    S. Baader. Positive braids of maximal signature . Enseign. Math. 59 (2013), no. 3-4, 351-358

    work page 2013

  2. [2]

    Baader, L

    S. Baader, L. Lewark. Positive two strand torus links . Preprint

    work page

  3. [3]

    Baader, L

    S. Baader, L. Lewark, L. Liechti. Checkerboard graph monodromies . L’Enseignement Math´ ematique 64 (2018), no. 2, pp. 65-88

    work page 2018

  4. [4]

    Branched covers of quasipositive links and L-spaces

    M. Boileau, S. Boyer, C. McA. Gordon. Branched covers of quasipositive links and L-spaces . arXiv:1710.07658v2 (2019)

    work page internal anchor Pith review Pith/arXiv arXiv 2019

  5. [5]

    Boileau, S

    M. Boileau, S. Boyer, C. McA. Gordon. On definite strongly quasipositive links and L-space branched covers. arXiv:1811.08862v1 (2018)

    work page arXiv 2018

  6. [6]

    Cameron, J.M

    P.J. Cameron, J.M. Goethals, J.J. Seidel, E.E. Shult. Line graphs, root systems, and elliptic geometry. Journal of algebra 43 (1976), 305-327

    work page 1976

  7. [7]

    Dehornoy

    P. Dehornoy. On the zeros of the Alexander polynomial of a Lorenz knot . Annals de l’Institut Fourrier 65 (2015), 509-548

    work page 2015

  8. [8]

    L. Liechti. Positive braid knots of maximal topological 4-genus . Math. Proc. Cambridge Phi- los. Soc. 161 (2016), no. 3, 559-568

    work page 2016

  9. [9]

    P. M. Melvin, H. R. Morton. Fibred knots of genus 2 formed by plumbing Hopf bands . J. London Math. Soc.(2) 34 (1986), 159-168

    work page 1986

  10. [10]

    Carl D. Meyer. Matrix analysis and applied linear algebra . Siam (2000), 472

    work page 2000

  11. [11]

    Susanna S. Epp. Discrete mathematics with applications . Thomson, third edition

    work page

  12. [12]

    H. F. Trotter. Homology of group systems with applications to knot theory . Annals of Math- ematics, Vol. 76, No. 3 (1962), 464-498. Mathematics Institut, University of Bern, Sidlerstrasse 5, 3012 Bern, Switzerland E-mail address: lucas.fernandez@math.unibe.ch 17

    work page 1962