{"id":"d4d7f82f-d07e-49c6-b719-10365d2d1cb8","arxiv_id":"2607.20910","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"The spectrum of the (r-1)-up-Laplacian is majorized by the conjugate degree sequence of (r-1)-faces, and the Duval-Reiner conjecture fails for all r≥2.","lead":"An r-dimensional simplicial complex has its top up-Laplacian eigenvalue sums controlled by the degrees of its (r-1)-faces, a higher-dimensional version of a classical graph inequality. The paper also constructs explicit complexes that disprove a related 2002 conjecture in every dimension r≥2.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: the cited signature-similarity lemma is the main dependency, but the reduction and counterexample are sound.","rationale":"The reader correctly isolates Lemma 3.1 and the delegation to [8, Lemma 3.2] as the most load-bearing step, and I agree that this is where the proof leans on an external result. However, my own spot-checks of the sign structure indicate the reduction is sound: the local matrix is the Gram matrix of a signed incidence matrix of the link graph, and the sign indeterminacy is resolved by a switching similarity. The entrywise decomposition in Lemma 3.2 is correct, and the double-counting in Theorem 1.3 is valid. For the counterexample, the root existence for p(x) follows from elementary sign changes and the intermediate value theorem, and the stated intervals for alpha1 and alpha2 are narrow enough to certify by interval arithmetic. No step I examined invalidates the reader's ACCEPT verdict.","tokens_in":12216,"tokens_out":40672,"duration_ms":348409,"concrete_test":"Run an exhaustive script over all vertex orderings of small complexes, e.g., the r=3 complex with eta={a,b} and three 3-faces eta-uv, eta-vw, eta-uw, computing L^down_r(K;eta), applying a diagonal +/-1 similarity and permutation, and checking equality with L^down_1(lk eta) plus zeros. If equality fails for any ordering, Theorems 1.3 and 1.6 would require revisiting; if it passes, the central dependency is verified.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim is the majorization theorem Theorem 1.3, proved through Lemmas 3.1 and 3.2: the r-dimensional down Laplacian decomposes into local down Laplacians indexed by (r-2)-faces, and each local operator is, up to signed permutation, the edge Laplacian of the link graph. The general-ordering case of Lemma 3.1 is delegated to [8, Lemma 3.2]. I examined the sign behavior on small interleaved-order examples, including r=3 with eta={a,b} and link vertices ordered u<a<v<b<w. Individual columns can have two equal signs, but the resulting signed local matrix can be switched by a diagonal sign matrix to the standard graph edge Laplacian, and the spectra agree; the cited lemma is plausible. Lemma 3.2 itself checks out by direct entrywise bookkeeping. The Section 5 numerical claim is also not a real vulnerability: p(x)=x^5-19x^4+133x^3-413x^2+527x-175 has p(0)<0, p(1)>0, p(2)>0, p(3)<0, p(4)<0, p(5)>0, p(6)<0, p(7)>0, which forces five real roots in (0,1), (2,3), (4,5), (5,6), and (6,7); alpha1 and alpha2 are the top two, and the quoted interval arithmetic is routine. No significant objection identified.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves two main results for finite simplicial complexes. Theorem 1.3 states that for every r-dimensional complex K, the spectrum of the (r-1)-dimensional up-Laplacian is majorized by the conjugate degree sequence of its (r-1)-faces, recovering the Grone-Merris-Bai theorem for graphs when r=1. Theorem 1.6 establishes a Brouwer-type upper bound for partial sums of these Laplacian eigenvalues, recovering the graph theorem of Brouwer-Kothari-Tudose when r=1. The proofs are based on a local decomposition of the r-dimensional down Laplacian into local Laplacians indexed by (r-2)-faces, together with a reduction of each local Laplacian to the edge Laplacian of a link graph. In Section 5 the paper constructs, for every r>=2 and n>=r+5, an explicit pure r-dimensional complex K_{n,r} that violates the fifth partial-sum inequality of the Duval-Reiner conjecture, thereby disproving that conjecture in all dimensions r>=2.","tokens_in":12521,"tokens_out":10668,"duration_ms":88651,"significance":"If the results are correct, as the referee's independent check and the stress-test notes indicate, the paper makes a substantial contribution: it gives the first higher-dimensional majorization theorem of Grone-Merris-Bai type for face degrees, a credible higher-dimensional Brouwer-type bound, and an explicit counterexample to a named conjecture. The proofs are largely self-contained and display an elegant double-counting technique; the reduction to link graphs makes the arguments transparent and reuses strong known graph results. The counterexample is constructive and checkable, and the recovery of the r=1 graph theorems is a nice sanity check. The paper is clearly written and the main claims are well motivated.","major_comments":[],"minor_comments":[{"comment":"The general-ordering case of the key reduction is delegated to [8, Lemma 3.2] without stating that lemma or its proof. Since Lemma 3.1 is load-bearing for both Theorems 1.3 and 1.6, please state [8, Lemma 3.2] explicitly or include a proof in an appendix so the paper is self-contained.","section":"Section 3, Lemma 3.1"},{"comment":"The assertion that p(x) has five real roots with 6.16382<alpha1<6.16383 and 5.83636<alpha2<5.83637 is supported only by \"a direct computation.\" Please include a short certificate, for example a Sturm sequence or exact interval arithmetic, so that the counterexample is fully verifiable without rerunning a numerical computation.","section":"Section 5"},{"comment":"There are several typos: \"EIGENV ALUE\" in the abstract and title spacing, and \"Laplaican\" for \"Laplacian\" in the abstract and in Sections 2 and 3. These should be corrected.","section":"Abstract and Section 3"},{"comment":"The notation \"eigenvalues lambda in lambda_2(K)\\setminus{7}\" is slightly informal; it should be clarified that this refers to the multiset of eigenvalues with 7 removed according to its multiplicity.","section":"Section 5, Lemma 5.1"},{"comment":"The equality case in the simplex example is stated without proof; a one-line verification that the up-Laplacian has eigenvalues (r+1,0,\\ldots,0) would be helpful.","section":"Section 1, Remark 4.1"}],"recommendation":"minor_revision","confidential_remarks":"The paper acknowledges the concurrent independent work of Huang [15]; the construction here is explicit and appears sound. The only substantive dependency on an unpublished source is [8, Lemma 3.2], which the authors can readily incorporate. I do not see a scope or novelty concern."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The short version: this is a genuine advance. The paper proves a higher-dimensional analog of the Grone–Merris–Bai theorem—majorization of the (r−1)-up-Laplacian spectrum by the conjugate (r−1)-degree sequence—and complements it with a Brouwer-type bound. It also settles the Duval–Reiner conjecture negatively: for every r≥2 there are pure r-complexes violating the fifth partial sum. That is a real open problem in spectral combinatorics, and the counterexample is explicit and checkable.\n\nThe main proof is a clean reduction. Lemma 3.2 decomposes the r-down Laplacian into local pieces indexed by (r−2)-faces; Lemma 3.1 identifies each local piece with the edge Laplacian of the link graph. Then Grone–Merris–Bai and Brouwer–Kothari–Tudose apply link-by-link, and the partial-sum inequalities follow by double counting. I checked the bookkeeping in Lemma 3.2; it is correct. The general-ordering case of Lemma 3.1 is delegated to [8, Lemma 3.2], a signature-similarity lemma from the authors' earlier work. I did not find a flaw there; the stress-test examples work out. It would be nice to include a proof of that lemma, but as a citation to a preprint it is acceptable if the referee presses on it.\n\nThe counterexample section is also solid. The base 2-complex on 7 vertices is small enough to be verified by hand; the characteristic polynomial is displayed and the claimed root intervals are narrow. The join/shift argument lifting it to arbitrary dimension is elegant. There is a tiny typo in the degree list ('d0(vh)=1, h≥7' should presumably be h=1..q), but the meaning is clear.\n\nWhat is not here is a resolution of Lew's conjecture; the bound in Theorem 1.6 is weaker, and the paper says so explicitly. That is honest. The numerical root check is not machine-certified, but it is the kind of thing any referee can verify in ten minutes.\n\nThe citation pattern looks fair: the paper acknowledges Huang's concurrent independent counterexample, and the self-citation is to a technical lemma rather than to inflate importance.\n\nBottom line: this paper deserves a serious referee and, on current evidence, acceptance after minor revisions. It will be cited by anyone working on Laplacian spectra of complexes. I would bring it to our reading group if the group is in the mood for combinatorics; for a general audience it is a bit specialized.","headline":"Solid paper: proves the right higher-dimensional analog of Grone–Merris–Bai, gives a Brouwer-type bound, and kills the Duval–Reiner conjecture with an explicit construction; only soft spot is a delegated lemma and a routine numerical claim.","tokens_in":13052,"tokens_out":2634,"would_cite":true,"duration_ms":21030,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["05C50","05E45","15A18","55U10"],"pacs":[],"model":"deepseek-v4-flash","headline":"In every simplicial complex, the top up-Laplacian spectrum is majorized by the conjugate face-degree sequence.","keywords":["simplicial complex","Laplacian eigenvalues","majorization","conjugate degree sequence","Grone-Merris-Bai theorem","Brouwer-type inequality","Duval-Reiner conjecture","link graph"],"falsifier":"Falsify the reduction by computing, in exact arithmetic, the eigenvalues of $L^{\\mathrm{down}}_r(K;\\eta)$ for a complex with an arbitrary vertex ordering that does not put the vertices of $\\eta$ first: they must match the eigenvalues of $L^{\\mathrm{down}}_1(\\mathrm{lk}\\,\\eta)\\oplus O$. Falsify the counterexample by running a Sturm sequence or interval-arithmetic certificate on $p(x)=x^5-19x^4+133x^3-413x^2+527x-175$; if the five roots are not real, or if $\\alpha_1+\\alpha_2\\le12$, the constructed complex no longer violates the fifth partial sum.","tokens_in":12029,"feed_emoji":"🔺","tokens_out":13934,"duration_ms":99190,"temperature":0.7,"pith_summary":"This paper proves that the eigenvalues of the top up-Laplacian of any $r$-dimensional simplicial complex are controlled, in the majorization order, by the degrees of its top faces: the sums of the $\\ell$ largest eigenvalues never exceed the sums of the $\\ell$ largest conjugate face degrees, with total sums equal. This is a direct higher-dimensional analogue of the Grone-Merris-Bai theorem for graphs, recovered when $r=1$, and it replaces the Duval-Reiner conjecture, which proposed the same control by vertex degrees. The paper also establishes a Brouwer-type bound on sums of the $\\ell$ largest up-Laplacian eigenvalues in terms of the numbers of $r$-faces and $(r-2)$-faces, recovering the graph Brouwer-Kothari-Tudose theorem when $r=1$. On the negative side, it constructs pure $r$-dimensional complexes on any $n\\ge r+5$ vertices whose spectra violate the Duval-Reiner conjecture at the fifth partial sum, so that conjecture fails in every dimension $r\\ge2$. A sympathetic reading is that the right degree sequence for higher-dimensional Laplacian majorization is the face-degree sequence, not the vertex-degree sequence.","feed_headline":"Face degrees bound higher Laplacian eigenvalue sums","feed_subtitle":"A new theorem extends the graph Grone-Merris-Bai bound to all dimensions while the Duval-Reiner conjecture fails.","key_machinery":"The load-bearing identity is the decomposition $rL^{\\mathrm{down}}_r(K)=\\sum_{\\eta\\in S_{r-2}(K)}L^{\\mathrm{down}}_r(K;\\eta)$, where $L^{\\mathrm{down}}_r(K;\\eta)$ is the local down Laplacian acting on the $r$-faces that contain $\\eta$. Lemma 3.1 identifies each local piece, up to a signed permutation, with $L^{\\mathrm{down}}_1(\\mathrm{lk}\\,\\eta)\\oplus O$, i.e., the edge Laplacian of the link graph of $\\eta$ together with a zero block. This reduces the whole high-dimensional problem to graph Laplacians: the partial eigenvalue sums of $L^{\\mathrm{down}}_r(K)$ are at most the sums of the corresponding partial sums over the link graphs, and those are controlled by the classical graph theorems. The double-counting steps then convert the link-level degree information into face-degree information for $K$, producing both Theorem 1.3 and Theorem 1.6.","core_discovery":"The positive core claim is Theorem 1.3: for every $r$-dimensional simplicial complex $K$, the spectrum of the $(r-1)$-dimensional up-Laplacian satisfies $\\lambda_{r-1}(K)\\preccurlyeq d^\\top_{r-1}(K)$, meaning that for each $\\ell$ the $\\ell$ largest eigenvalues sum to at most the $\\ell$ largest conjugate degrees of the $(r-1)$-faces, and the total sums coincide. The proof works with the $r$-dimensional down Laplacian, which shares the nonzero spectrum of the $(r-1)$-up-Laplacian. The paper decomposes $rL^{\\mathrm{down}}_r(K)$ into local down Laplacians attached to each $(r-2)$-face $\\eta$, shows each local piece is, up to a signed permutation, the edge Laplacian of the link graph $\\mathrm{lk}\\,\\eta$ padded with zeros, and then applies the graph Grone-Merris-Bai theorem link by link and sums the results. The second positive claim, Theorem 1.6, is the Brouwer-type inequality $\\sum_{i=1}^\\ell \\lambda_{r-1,i}(K)\\le \\frac{r+1}{2}f_r(K)+\\frac{f_{r-2}(K)}{r}\\binom{\\ell+1}{2}$, obtained by applying the graph Brouwer bound to each link and double counting edges of links against $r$-faces. The negative claim is Proposition 5.3: for every $r\\ge2$ and $n\\ge r+5$, the constructed pure complex $K_{n,r}$ violates the fifth partial-sum inequality of the Duval-Reiner conjecture, whose conjectured upper bound uses conjugate vertex degrees $d^\\top_0(K)$.","pith_inferences":["A natural extension is to test whether Theorem 1.3 survives under weighted faces: the link-graph reduction is linear, so face weights should translate to edge weights on each link graph, and weighted versions of the graph theorems would transfer if they are available.","The counterexample family may violate not only the fifth partial sum but later ones; computing the gap $\\alpha_1+\\alpha_2-12$ as a function of $q$ would show how far the vertex-degree majorization is from being true.","A rigorous interval-arithmetic or Sturm-sequence certificate for the five real roots would remove the only numerical computation in the paper, making Proposition 5.3 fully symbolic.","The failure of the vertex-degree conjecture suggests testing weakened majorization statements, for instance with conjugate degrees of $(r-2)$-faces or with an absolute constant factor, as the next plausible conjecture in this direction."],"forward_implications":["For any $r$-dimensional complex, the $\\ell$-th partial sum of the top $(r-1)$-up-Laplacian eigenvalues is bounded by the $\\ell$-th partial sum of the conjugate $(r-1)$-face degree sequence, so the face-degree sequence is the correct higher-dimensional analogue of graph degrees for majorization.","The largest $(r-1)$-up-Laplacian eigenvalue is bounded by $\\frac{r+1}{2}f_r(K)+\\frac{f_{r-2}(K)}{r}$, a number computable directly from face counts; at $r=1$ this recovers the graph bound $f_1(G)+1$.","The Duval-Reiner conjecture fails in every dimension $r\\ge2$: there is no way to repair the conjecture as a statement about vertex degrees, and the obstruction is already visible at the fifth partial sum.","Both main theorems collapse to the known graph results at $r=1$, so the extension is exact rather than approximate.","The counterexample family supplies pure complexes on every sufficiently large vertex count, so the failure of the Duval-Reiner conjecture is not confined to sporadic small examples."],"supporting_citations":[{"why":"With [11], supplies the Grone-Merris theorem for graphs that Theorem 1.3 generalizes and that is invoked on each link graph.","marker":"[2]"},{"why":"The Grone-Merris graph majorization theorem used in the proof of Theorem 1.3 on the link graphs.","marker":"[11]"},{"why":"Posed the Duval-Reiner conjecture on vertex-degree majorization that Proposition 5.3 disproves.","marker":"[5]"},{"why":"Ky Fan's variational characterization of eigenvalue partial sums, used to move from the sum of local Laplacians to local eigenvalue sums.","marker":"[7]"},{"why":"Its Lemma 3.2 provides the signed-permutation similarity that makes the link-graph identification of Lemma 3.1 valid under arbitrary vertex orderings.","marker":"[8]"},{"why":"The graph Brouwer bound proved by Brouwer-Kothari-Tudose that is applied to each link graph in the proof of Theorem 1.6.","marker":"[17]"}],"fun_headline_variants":["Simplicial Laplacian spectrum: conjugate-degree majorization","Brouwer-type Laplacian bound proven for simplicial complexes","Duval-Reiner conjecture fails from dimension two upward","Laplacian majorization theorem extends to all dimensions","New eigenvalue inequalities for simplicial complex Laplacians"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole argument rests on the local identity that, up to a signed permutation, the part of the down Laplacian on $r$-faces around an $(r-2)$-face $\\eta$ is exactly the edge Laplacian of the link graph of $\\eta$ padded with zeros; if that identity ever failed for some vertex ordering, the reduction to graph theorems would collapse, and the counterexample separately depends on the computed claim that $p(x)$ has five real roots whose top two sum to more than $12$.","fun_headline_variants_meta":{"raw":{"variants":["Simplicial Laplacian spectrum: conjugate-degree majorization","Brouwer-type Laplacian bound proven for simplicial complexes","Duval-Reiner conjecture fails from dimension two upward","Laplacian majorization theorem extends to all dimensions","New eigenvalue inequalities for simplicial complex Laplacians"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001089,"raw_usage":{"total_tokens":4692,"prompt_tokens":1227,"completion_tokens":3465,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":843,"completion_tokens_details":{"reasoning_tokens":3382}},"tokens_in":843,"tokens_out":3465,"duration_ms":23652,"temperature":1.0,"reasoning_tokens":3382,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T15:31:27.574229+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Falsify the reduction by computing, in exact arithmetic, the eigenvalues of $L^{\\mathrm{down}}_r(K;\\eta)$ for a complex with an arbitrary vertex ordering that does not put the vertices of $\\eta$ first: they must match the eigenvalues of $L^{\\mathrm{down}}_1(\\mathrm{lk}\\,\\eta)\\oplus O$. Falsify the counterexample by running a Sturm sequence or interval-arithmetic certificate on $p(x)=x^5-19x^4+133x^3-413x^2+527x-175$; if the five roots are not real, or if $\\alpha_1+\\alpha_2\\le12$, the constructed complex no longer violates the fifth partial sum.","supporting_citations":[{"cited_title":"Bai, The Grone–Merris conjecture,Trans","cited_arxiv_id":null,"evidence_quote":"With [11], supplies the Grone-Merris theorem for graphs that Theorem 1.3 generalizes and that is invoked on each link graph."},{"cited_title":"Grone and R","cited_arxiv_id":null,"evidence_quote":"The Grone-Merris graph majorization theorem used in the proof of Theorem 1.3 on the link graphs."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Posed the Duval-Reiner conjecture on vertex-degree majorization that Proposition 5.3 disproves."},{"cited_title":"Fan, On a theorem of Weyl concerning eigenvalues of linear transformations I,Proc","cited_arxiv_id":null,"evidence_quote":"Ky Fan's variational characterization of eigenvalue partial sums, used to move from the sum of local Laplacians to local eigenvalue sums."}],"review_version":1}