{"id":"e58589c2-f35f-4ee5-ae74-79358af0c61c","arxiv_id":"1908.07909","paper_version":2,"verdict":"REJECT","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper asserts jellyfish graphs are determined by Laplacian and signless Laplacian spectra, but the Laplacian proof uses a false spectral radius bound and the signless proof has an unproved step.","lead":"Jellyfish graphs are built by placing a star on every vertex of a cycle. This paper claims they are uniquely identifiable from their Laplacian spectra, but a central bound used in the proof fails for even cycles.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 3.1's Laplacian spectral radius bound is false for even q (e.g. JFG(2,4)); Lemma 3.2's degree-sequence argument depends on it, so the even-q DLS proof collapses.","rationale":"The reader's weakest assumption is correct and is the cleanest single gap. JFG(2,4) is a valid jellyfish graph with even q, and its Laplacian spectral radius exceeds the bound asserted in Lemma 3.1. This is not a minor numerical slip: Lemma 3.2 has no other mechanism to bound the maximum degree of an L-cospectral graph, so the degree-sequence conclusion that drives Theorem 3.1 is unsupported. The DQS half is also damaged by the p=1-only spectrum formula in Lemma 3.3 and the unsupported subgraph assertion in Lemma 3.6, independently reinforcing rejection. The theorems may be true, but the manuscript as written does not prove them; the reader's REJECT verdict should stand.","tokens_in":8699,"tokens_out":14733,"duration_ms":131102,"concrete_test":"Compute the Laplacian spectrum of JFG(2,4) using the Fourier ansatz on the 4-cycle. The alternating eigenvector with y_i=-x_i/(λ-1) gives characteristic equation λ^2-7λ+4=0, whose largest root is (7+√33)/2≈6.372. Since 6.372 exceeds p+3+2/(p+2)=5.5, Lemma 3.1's bound is false; rerunning Lemma 3.2 without that bound shows that the deduction d1(H)≤p+2 no longer follows.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Lemma 3.1 asserts μ1(JFG(p,q)) ≤ p+3+2/(p+2). For JFG(2,4), take the alternating cycle mode x_i=(-1)^i and leaf values y_i=-x_i/(λ-1). The Laplacian eigenvalue λ satisfies (λ-p-4)(λ-1)=p, i.e. λ^2-7λ+4=0, so μ1=(7+√33)/2≈6.372. The claimed bound is 5+2/4=5.5, so Lemma 3.1 is false exactly in the even-q case that Theorem 3.1 targets. Lemma 3.2 uses Lemma 3.1 plus Theorem 2.3 to infer d1(H)≤p+2; once the bound is false, that inference is unsupported, and Equations (4)-(5) no longer force the degree sequence. Thus Theorem 3.1 is not established. The DQS half has independent defects: Lemma 3.3's determinant uses 1/(x-1) although the displayed B,C blocks give p/(x-1), so the stated Q-spectrum formula is only valid for p=1, and Lemma 3.6 contains an unsupported assertion that a component contains a jellyfish subgraph.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript claims that jellyfish graphs JFG(p,q), obtained by attaching p leaves to each vertex of a q-cycle, are determined by their signless Laplacian spectrum (DQS) and, when q is even, by their Laplacian spectrum (DLS). The proof strategy for the DLS half is to show via a spectral-radius bound that any L-cospectral graph has the same degree sequence and is unicyclic, hence isomorphic; the DQS half proceeds by computing the Q-spectrum explicitly, then using spectral moments and a connectivity argument to force the degree sequence and the graph structure. The paper also poses a conjecture for odd q in the Laplacian case.","tokens_in":8902,"tokens_out":6343,"duration_ms":59022,"significance":"The topic is appropriate for a spectral graph theory journal: extending determinability results from sun graphs to a natural family is of interest, and the paper includes explicit formulas (the Q-spectrum expression in Lemma 3.3 and the spectral-moment equations in Lemma 3.5) that are potentially useful. However, the central technical lemmas contain errors that invalidate both main theorems. Lemma 3.1's spectral radius bound is demonstrably false for even q, and Lemma 3.3's Q-spectrum computation is algebraically inconsistent with its own block matrix. These are not superficial typos: the subsequent degree-sequence arguments and connectivity proof depend directly on these claims. As a result, the manuscript does not establish either DQS or DLS for jellyfish graphs.","major_comments":[{"comment":"The claimed upper bound μ1(JFG(p,q)) ≤ p+3+2/(p+2) is false for even q. For JFG(2,4), the alternating cycle mode x_i = (-1)^i with leaf values y_i = -x_i/(λ-1) gives Laplacian eigenvalue λ satisfying (λ-p-4)(λ-1)=p, i.e., λ² - 7λ + 4 = 0, so μ1 = (7+√33)/2 ≈ 6.372, exceeding the claimed bound 5.5. Since Lemma 3.2 uses this bound together with Theorem 2.3 to conclude d1(H) ≤ p+2, Eq. (4) does not follow and the degree-sequence argument in Lemma 3.2 collapses. This invalidates Theorem 3.1, which is the paper's DLS claim for even q.","section":"Lemma 3.1"},{"comment":"The displayed block matrix has B and C blocks consisting of p copies of I_q stacked vertically, so the Schur complement of D = I_{n-q} in the characteristic determinant is p/(x-1), not 1/(x-1). Consequently the expression for the Q-spectrum is valid only for p=1. Corollary 3.2 and the subsequent use of the spectral radius in Lemma 3.6 rely on this formula, so the DQS proof is not established for p ≥ 2.","section":"Lemma 3.3"},{"comment":"In case (2a), the proof asserts that 'by Lemma 3.5, there exists a subgraph G1 of H_j such that G1 ≅ JFG(p,q')' for some unspecified q'. Lemma 3.5 establishes only that H and G have the same degree sequence; it does not imply the existence of a jellyfish subgraph with the same p and some cycle length. This unsupported assertion is load-bearing for the connectivity argument, and the same gap reappears in case (2b).","section":"Lemma 3.6"}],"minor_comments":[{"comment":"The proof cites Lemmas 2.3 and 2.4 for the two spectral-radius inequalities, but the statements of those lemmas do not directly supply the bounds; the citations appear to be mismatched.","section":"Lemma 3.1"},{"comment":"The notation n'_{p+2} for the count of degree-(p+2) vertices in G is introduced with a prime that carries no meaning and could be confused with a complementary quantity; please remove the prime.","section":"Lemma 3.2"},{"comment":"The text 'det(Q(H) = 4' is missing a closing parenthesis before '= 4'.","section":"Lemma 3.4"},{"comment":"The step 'H is an unicyclic graph and so H = G' needs justification: the degree sequence alone does not immediately force H to be unicyclic, and the citation to Lemma 3.6 should be made explicit here.","section":"Theorem 3.2"}],"recommendation":"reject","confidential_remarks":"The false bound in Lemma 3.1 is easily verified by direct computation and is decisive for the DLS claim. The algebraic error in Lemma 3.3 similarly undermines the DQS claim. These are not local typographical issues; the main theorems collapse without a substantially different proof approach."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a natural extension of the sun-graph results to jellyfish graphs, but the proof of the Laplacian half is invalid and the Q-spectrum computation is wrong, so as written the paper does not establish its theorems.\n\nWhat's genuinely new: for p>1, nothing in the cited literature says JFG(p,q) is DQS or DLS. The paper sets up the family cleanly and uses standard tools—spectral moments, line graphs, degree-sequence forcing. If the results were true, they'd be a modest addition to the list of graphs determined by their spectra.\n\nThe problems are not minor. Lemma 3.1 asserts μ1(JFG(p,q)) ≤ p+3+2/(p+2). That bound is false for even q. Take JFG(1,4): the alternating cycle mode gives Laplacian eigenvalue 3+√5 ≈ 5.236, while the claimed upper bound is 4+2/3 ≈ 4.667. Lemma 3.2 uses that bound to conclude d1(H) ≤ p+2, which then forces the degree sequence. With the bound false, the degree-sequence argument collapses, and Theorem 3.1 (the even-q DLS claim) is not established.\n\nThe Q-side has its own defect. In Lemma 3.3 the matrix blocks are stated with B = [I_q ... I_q] (p copies), so the Schur complement should contain p/(x-1). The displayed determinant uses 1/(x-1), which is only right for p=1. The resulting Q-spectrum formula and Corollary 3.2 are therefore wrong for p>1. Lemma 3.5 and Lemma 3.6 inherit the problem; Lemma 3.6 also jumps from degree sequences to the existence of jellyfish subgraphs without justification.\n\nOn the plus side, the paper is readable, the definitions are clear, the literature is appropriately cited, and the conjecture about odd q is sensible. But the abstract also overstates the DLS result by omitting the even-q condition that the body carries.\n\nMy take: this should not go to peer review as is. The counterexample to Lemma 3.1 and the p/(x-1) error are easy to verify, and they are load-bearing. If the authors repair those and re-check Lemma 3.6, the paper may be salvageable, but the current version is not a proof.","headline":"Nice generalization idea, but the proofs have load-bearing errors: Lemma 3.1's bound is false for even q, and the Q-spectrum formula doesn't match the paper's own matrix.","tokens_in":9465,"tokens_out":8257,"would_cite":false,"duration_ms":71990,"reading_group":"no","serious_thinker":"no","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["05C50"],"pacs":[],"model":"deepseek-v4-flash","headline":"Jellyfish graphs are claimed to be determined by their signless Laplacian spectrum, and by their Laplacian spectrum when the cycle length is even.","keywords":["jellyfish graphs","sun graphs","signless Laplacian spectrum","Laplacian spectrum","spectral determination","DQS","DLS","unicyclic graphs"],"falsifier":"Compute the Laplacian spectrum of the 8-vertex graph $JFG(1,4)$, a 4-cycle with one pendant leaf at each vertex: its largest Laplacian eigenvalue is $3+\\sqrt5 \\approx 5.236$, which exceeds the claimed upper bound $14/3 \\approx 4.667$ in Lemma 3.1. This falsifies the spectral-radius bound behind the even-$q$ degree-sequence argument; a search for other graphs with the same degree sequence and spanning-tree count as $JFG(p,q)$ for small even $q$ would settle whether the DLS conclusion itself survives.","tokens_in":8452,"feed_emoji":"🪼","tokens_out":10957,"duration_ms":98275,"temperature":0.7,"pith_summary":"This paper sets out to prove that jellyfish graphs—graphs formed by attaching $p$ pendant leaves to each vertex of a cycle of length $q$—are determined by their spectra in two senses: no other graph can share the signless Laplacian spectrum of a jellyfish graph, and, when $q$ is even, no other graph can share its Laplacian spectrum either. These properties are called DQS and DLS, respectively. The results extend known spectral-determination theorems for sun graphs, which are the special case $p=1$. The proof strategy is to show that any cospectral candidate must have the same degree sequence as the jellyfish graph, and then to use the fact that a unicyclic graph with that degree sequence is forced to be the jellyfish graph itself.","feed_headline":"Jellyfish graphs claimed determined by their spectra","feed_subtitle":"For even cycles the paper claims the Laplacian spectrum also determines the graph.","key_machinery":"The load-bearing object is the degree sequence $(1^{pq}, (p+2)^q)$ of $JFG(p,q)$—the $q$ cycle vertices each carry degree $p+2$, and the $pq$ leaves have degree 1. The proofs force any cospectral candidate to reproduce this sequence by combining spectral moments of the Laplacian or signless Laplacian with counts of closed walks and triangles in the line graph, using the spanning-tree count to fix the cycle length. On the Laplacian side, the cap $\\mu_1(JFG(p,q)) \\leq p+3+\\frac{2}{p+2}$ is what limits the maximum degree of a cospectral graph to $p+2$; on the Q-side, an explicit block-matrix formula gives the eigenvalues $\\lambda_i + p+3 \\pm \\frac{1}{2}\\sqrt{\\lambda_i^2+(2p+2)\\lambda_i+p^2+2p+5}$ with $\\lambda_i=\\cos(2\\pi i/q)$, plus a multiplicity of 1.","core_discovery":"The paper's central claim is that the jellyfish graph $JFG(p,q)$ is DQS for all $p,q$, and DLS when $q$ is even. For the signless Laplacian side, the paper derives the full spectrum of any Q-cospectral graph from a block decomposition of the signless Laplacian, then uses spectral-moment identities to recover the degree sequence and connectedness to force isomorphism. For the Laplacian side, it argues that any L-cospectral graph has the same order, size, number of spanning trees, and first Zagreb index as the jellyfish graph; under a bound on the largest Laplacian eigenvalue, these force the same degree sequence, and the spanning-tree count fixes the cycle length, so the candidate must be the jellyfish graph. A corollary transfers the Laplacian determination to complements of such graphs.","pith_inferences":["If the even-$q$ Laplacian bound fails generally, the DLS conclusion may still hold, but the proof would need a different argument; the natural next target is the correct Laplacian spectral radius bound for $JFG(p,q)$.","The conjecture that odd-$q$ jellyfish graphs are also DLS could be tested with the same degree-sequence equations, which do not depend on the disputed bound on the Q-side.","Because the mechanism is essentially degree-sequence forcing, the technique may extend to other unicyclic graphs built by coalescing stars with cycles, a direction the paper does not explore."],"forward_implications":["Every jellyfish graph $JFG(p,q)$ is claimed to be uniquely determined by its signless Laplacian spectrum, so any graph with the same $Q$-spectrum must be isomorphic to it.","For even $q$, any graph with the same Laplacian spectrum as a jellyfish graph is claimed to be isomorphic to it, and so is the complement of that graph.","A graph Q-cospectral with $JFG(p,q)$ must have degree sequence $(1^{pq}, (p+2)^q)$, according to the proof.","The explicit $Q$-spectrum formula gives the spectral radius $\\frac{p+5+\\sqrt{p^2+6p+13}}{2}$ for any jellyfish graph and for anything cospectral with it."],"supporting_citations":[{"why":"Proves the sun graphs are DQS, the result that this paper generalizes to jellyfish graphs.","marker":"[10]"},{"why":"Proves the sun graphs are DLS, the Laplacian-side precedent this paper extends.","marker":"[2]"},{"why":"Supplies the closed-walk counts and the bound on $\\mu_1(G)$ used to cap the Laplacian spectral radius in Lemma 3.1.","marker":"[15]"},{"why":"Relates Laplacian eigenvalues of a connected unicyclic bipartite graph to adjacency eigenvalues of its line graph, used to link closed walks with degree counts.","marker":"[13]"},{"why":"States that Q-cospectral graphs have A-cospectral line graphs, used to transfer line-graph triangle counts to degree equations.","marker":"[16]"},{"why":"Gives the general bound $\\mu_1(G) \\geq d_1(G)+1$ used to convert the spectral-radius cap into a degree cap.","marker":"[6]"},{"why":"Gives the first four coefficients of the characteristic polynomial, from which the degree-sum identities are obtained.","marker":"[11]"},{"why":"Provides the signless Laplacian spectral moments and the determinant criteria used in Lemmas 2.1–2.3.","marker":"[4]"},{"why":"Relates Laplacian spectra of a graph and its complement, used to derive the complement corollary.","marker":"[9]"}],"fun_headline_variants":["Jellyfish graphs: spectra tell them apart","Spectra uniquely determine jellyfish graphs","Jellyfish graphs proven DLS and DQS","No two jellyfish graphs share L or Q spectra","Spectra force jellyfish graph isomorphism"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The Laplacian-side proof rests on the bound $\\mu_1(JFG(p,q)) \\le p+3+\\frac{2}{p+2}$, used to conclude that any L-cospectral graph has maximum degree at most $p+2$; for even $q$ the bound is asserted but not generally true, since $JFG(1,4)$ has Laplacian spectral radius $3+\\sqrt5 \\approx 5.236$, larger than $14/3 \\approx 4.667$.","fun_headline_variants_meta":{"raw":{"variants":["Jellyfish graphs: spectra tell them apart","Spectra uniquely determine jellyfish graphs","Jellyfish graphs proven DLS and DQS","No two jellyfish graphs share L or Q spectra","Spectra force jellyfish graph isomorphism"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000205,"raw_usage":{"total_tokens":1275,"prompt_tokens":712,"completion_tokens":563,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":328,"completion_tokens_details":{"reasoning_tokens":494}},"tokens_in":328,"tokens_out":563,"duration_ms":5517,"temperature":1.0,"reasoning_tokens":494,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:32:44.674478+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the Laplacian spectrum of the 8-vertex graph $JFG(1,4)$, a 4-cycle with one pendant leaf at each vertex: its largest Laplacian eigenvalue is $3+\\sqrt5 \\approx 5.236$, which exceeds the claimed upper bound $14/3 \\approx 4.667$ in Lemma 3.1. This falsifies the spectral-radius bound behind the even-$q$ degree-sequence argument; a search for other graphs with the same degree sequence and spanning-tree count as $JFG(p,q)$ for small even $q$ would settle whether the DLS conclusion itself survives.","supporting_citations":[{"cited_title":"Mirzakhah and D","cited_arxiv_id":null,"evidence_quote":"Proves the sun graphs are DQS, the result that this paper generalizes to jellyfish graphs."},{"cited_title":"Boulet, Spectral characterizations of sun graphs and brok en sun graphs, Discrete Math","cited_arxiv_id":null,"evidence_quote":"Proves the sun graphs are DLS, the Laplacian-side precedent this paper extends."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the closed-walk counts and the bound on $\\mu_1(G)$ used to cap the Laplacian spectral radius in Lemma 3.1."},{"cited_title":"Shen, and H","cited_arxiv_id":null,"evidence_quote":"Relates Laplacian eigenvalues of a connected unicyclic bipartite graph to adjacency eigenvalues of its line graph, used to link closed walks with degree counts."},{"cited_title":"Zhou and C","cited_arxiv_id":null,"evidence_quote":"States that Q-cospectral graphs have A-cospectral line graphs, used to transfer line-graph triangle counts to degree equations."},{"cited_title":"Grone and R","cited_arxiv_id":null,"evidence_quote":"Gives the general bound $\\mu_1(G) \\geq d_1(G)+1$ used to convert the spectral-radius cap into a degree cap."},{"cited_title":"Oliveira N","cited_arxiv_id":null,"evidence_quote":"Gives the first four coefficients of the characteristic polynomial, from which the degree-sum identities are obtained."},{"cited_title":"Cvetkovi´ c, P","cited_arxiv_id":null,"evidence_quote":"Provides the signless Laplacian spectral moments and the determinant criteria used in Lemmas 2.1–2.3."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Relates Laplacian spectra of a graph and its complement, used to derive the complement corollary."}],"review_version":1}