{"id":"3a90871e-d37e-43a1-92f2-a5ce01adff2e","arxiv_id":"2504.16721","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The spectrum of the cone of a non-reduced projective hypersurface with isolated singularities is determined by local spectral data of the reduced hypersurface together with global degree and multiplicity data.","lead":"This paper proves formulas for the spectrum of the cone over a projective hypersurface whose reduced underlying hypersurface has only isolated singularities. The formulas express spectral numbers in terms of local data at singular points plus global degree and multiplicity information, generalizing known results for line arrangements.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The bridge from the weighted-homogeneous case to the semi-weighted-homogeneous case in Theorem 2 depends on an unproved constancy of the cokernel E_{c,i}; all local counting is performed only at c=0.","rationale":"The central claim is Theorem 2: for a non-reduced plane curve whose reduced curve has only semi-weighted-homogeneous singularities, the spectrum of the cone is given by the explicit local-to-global formula (4). For this to hold, the local contribution of each singular point must depend only on the weighted-homogeneous limit data, and the only point where this reduction is proved is Section 4. The proof computes the wanted multiplicity at the weighted-homogeneous limit c=0 via (4.3), (4.7), and (4.11)-(4.12), then transfers it to c=1 by the µ-constant deformation. The transfer is the least secure part: the text moves from the snake lemma to 'independent of c∈S, using the local C*-action as in (3.3)' without demonstrating local freeness of the quotient C or the compatibility of the C*-action with the Deligne-extension twists that carry the residues. Nakayama's lemma in (4.8) is sufficient to propagate the vanishing of higher direct images once coherence is granted; the actual gap is the constancy of the cokernel dimension, which is what connects (4.11) to the original singularity. This is in the same region as the reader's weakest assumption, but I would locate it in the unproved E_{c,i}-constancy rather than in the Nakayama step itself. A direct Gröbner-basis computation of E_{1,i} for a non-weighted-homogeneous semi-weighted-homogeneous germ such as x^4+y^2+x^5 would settle the matter. The numerical examples in Sections 9-12 are consistent with the claimed formulas, but they only exercise the ordinary (semi-homogeneous) case w=w'=1, where the deformation issue is trivial; they do not test the fragile step. In good faith, the concern is about proof completeness rather than a known counterexample, so the conditional verdict remains appropriate.","tokens_in":20653,"tokens_out":25690,"duration_ms":251888,"concrete_test":"For a concrete semi-weighted-homogeneous but not weighted-homogeneous germ, e.g. h = x^4 + y^2 + x^5 with weights (w,w')=(1,2), d_j=4 and two branches of weighted degree 2 each, fix a residue datum (a_{j,l}, d, i) and compute the cokernel of ε_{1,i} directly: enumerate monomials u^{m_1-1}v^{m_2-1} and test the analogue of (4.12) with h in place of h' (or compute the image of Ω^2_{eY}(log eC_1)⊗eL_{1,i} in Ω^2_Y(log C_1)⊗L_{1,i} by Gröbner basis). Compare the resulting dimension with N_j(⌈γ_{j,i}⌉−1). If they differ, the claimed independence of E_{c,i} fails and the first equality in Theorem 2 is not established.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The first equality in Theorem 2 is proved in Section 4 by reducing the semi-weighted-homogeneous singularity to its weighted-homogeneous limit c=0, computing the cokernel dimension via (4.11)-(4.12), and then asserting that the cokernel E_{c,i} of the injection ε_{c,i} in (4.10) is independent of c∈S, using the local C*-action as in (3.3). This is the only step that transfers the formula from the model case c=0 to the original semi-weighted-homogeneous case c=1. The text does not prove that the snake-lemma quotient C is locally free over S, nor that the C*-action identifies E_{1,i} with E_{0,i} while preserving the Deligne-extension twists eL_{c,i} that carry the residue data β_{j,l,i}. If dim E_{1,i} differs from dim E_{0,i}, the first formula in (4) overcounts n_{f,i/d} by exactly that difference. The Nakayama part of the propagation is less problematic: from (4.8), vanishing of the c=0 fiber forces the relative direct image sheaf to vanish by Nakayama, so the higher direct images vanish for all nearby c. The genuinely fragile assertion is the constancy of the cokernel dimension, not the vanishing of R^nπ_*.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the spectrum (in the sense of Saito/Steenbrink) of the cone over a projective hypersurface Z whose underlying reduced variety Z' has only isolated singularities. Theorem 1 gives formulas for the spectral numbers of the cone of a reduced hypersurface in terms of local spectral numbers at the singular points of Z' and the global degree, and extends these formulas to non-reduced hypersurfaces under a constant-multiplicity condition. Theorem 2 treats non-reduced plane curves whose reduced support has only semi-weighted-homogeneous singularities: it expresses the three families n_{f,i/d}, n_{f,i/d+1}, n_{f,i/d+2} by the explicit closed formulas (4) in terms of local weights, local weighted degrees and multiplicities, and global degrees and multiplicities. Corollary 2 specializes to ordinary singularities and is compared with a formula of Yoon. The proof uses Deligne extensions, twisted logarithmic complexes, weighted blowups, and a mu-constant deformation argument; Part 3 contains Singular computations for several examples.","tokens_in":20921,"tokens_out":6516,"duration_ms":65536,"significance":"If correct, Theorem 2 is a substantial and useful generalization of the Budur--Saito formula for reduced line arrangements to non-reduced plane curves with semi-weighted-homogeneous singularities. The formulas are explicit, depend only on local and global discrete data, and do not require resolving the global configuration. The paper recovers known results, gives numerical consistency checks with Singular, and the formulas are derived rather than fitted. The main risk is the deformation-invariance step in Section 4: the first equality of Theorem 2 is reduced to a dimension count at the weighted-homogeneous limit c=0, and the proof that this count propagates to c=1 is not fully written out. If that step can be completed, the paper would be a solid contribution to the computation of spectra of cones.","major_comments":[{"comment":"The first equality of Theorem 2 is load-bearing and depends on the assertion after (4.10) that the cokernel E_{c,i} of the injection epsilon_{c,i} is independent of c in S and that the snake-lemma quotient is locally free over S. This is exactly the step that transfers the formula from the weighted-homogeneous model c=0, where (4.11) computes dim E_{0,i}, to the original semi-weighted-homogeneous case c=1. If dim E_{1,i} differs from dim E_{0,i}, the formula for n_{f,i/d} is off by that difference. The text only says 'using the local C*-action as in (3.3)' without proving that the C*-action identifies E_{1,i} with E_{0,i} while preserving the Deligne-extension twists eL_{c,i} that carry the residues beta_{j,l,i}. Please supply a complete proof of local freeness and constancy of the cokernel, or state and prove this as a separate lemma with explicit base-change and equivariance arguments.","section":"Section 4, Eqs. (4.8)--(4.11)"},{"comment":"The Nakayama argument is used twice: first to deduce vanishing of R^n e-pi_* from the c=0 fiber, and then to conclude vanishing for all c in S and for c=1. The first use is standard, but the passage from a sufficiently small neighborhood of 0 in S to c=1 is not. A disk around 0 that is small enough for Nakayama need not contain 1, and the sentence 'Using the local C*-action as in (3.3), the last vanishing holds also for c=1' requires a precise description of the C*-action on the family and on the sheaves. For a semi-weighted-homogeneous h that is not weighted homogeneous, the fibers h_c are not related to h_1 by the coordinate scaling of (3.3) alone. Please state the exact equivariance that gives the c=1 statement, or restrict the argument to a neighborhood and use a separate continuity or monodromy argument.","section":"Section 4, paragraph after (4.8)"},{"comment":"The local computation of E_{0,i} via (4.12) is plausible but is written as a one-line verification. In particular, the claim that a logarithmic form h'^{-1} u^{m_1-1} v^{m_2-1} du wedge dv lies in the image of epsilon_{0,i} exactly when (d_j - gamma_{j,i}) - d_j + w_j m_1 + w'_j m_2 >= 0 needs a justification that the vanishing order along a union of general C*-orbits in the weighted blowup controls membership in the twisted Deligne extension eL_i. Since this is the only local computation feeding the dimension formula (4.11), please expand it into a clear argument with the relevant filtration or vanishing-order estimate.","section":"Section 4, Eq. (4.12) and the local counting"}],"minor_comments":[{"comment":"The inclusion is written as 'e-pi_* Omega^2_{rel,log}(eL_{S,i}) ,-> Omega^2_{rel,log}(L_{S,i}' with a missing closing parenthesis; please correct the notation.","section":"Section 4, Eq. (4.10)"},{"comment":"The morphism e-rho' is used without definition; please define it explicitly as the restriction of e-rho to the exceptional divisor eE.","section":"Section 2, diagram in (2.1)--(2.2)"},{"comment":"The Singular code is hard to follow because the meaning of the variables GlCmp, Si, OD, and LG is explained only inline. A short table or comment block describing the input format and the meaning of the output would improve reproducibility.","section":"Sections 9--12"},{"comment":"The notation 'q' vs 'q'' ' is used for singular points that are not ordinary double points vs all singular points; a sentence explicitly recalling this distinction near Theorem 2 would help the reader.","section":"Theorem 2"},{"comment":"The strict versus non-strict inequalities in (4.12) and (5.6) are central to obtaining the correct ceiling terms, but the contrast is only implicit. A short remark explaining why the residue convention [0,1) versus (0,1] forces the strict inequality would be useful.","section":"Section 5, Eqs. (5.5)--(5.7)"}],"recommendation":"major_revision","confidential_remarks":"The paper fits the journal's scope and the main theorems are plausible, but Section 4 contains a load-bearing deformation-invariance assertion that is not proved. I do not see grounds for rejection; a complete proof of the constancy of E_{c,i} and a precise statement of the C*-action argument should be obtainable within the manuscript's framework. The numerical examples and the recovery of known formulas are valuable supporting evidence."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nWhat you should know first: this is a genuine extension of the Budur–Saito and Yoon formulas. Theorem 2 gives the full spectrum of the cone of a non-reduced plane curve whose reduction has semi-weighted-homogeneous singularities, in terms of local weights, local weighted degrees, multiplicities, and global degrees. That is new; the ordinary case in Corollary 2 recovers Yoon, and the reduced line arrangement case recovers Budur–Saito. The paper is honest about those overlaps, which is refreshing.\n\nThe machinery is solid mixed Hodge module stuff. The numerical checks with Singular are consistent, and the formulas have enough cross-checks that I'd bet on them being right.\n\nThe soft spot is exactly where the reader flags: getting from weighted-homogeneous to semi-weighted-homogeneous in Theorem 2. Section 4 shows vanishing of higher direct images for c=0 (weighted-homogeneous), uses Nakayama to propagate that vanishing to nearby c, then claims the cokernel E_{c,i} of the injection ε_{c,i} is independent of c, using the local C*-action as in (3.3). The Nakayama half is fine. But the constancy of the cokernel dimension is more delicate: it requires the snake-lemma quotient to be locally free and the C*-action to preserve the Deligne-extension twists that carry the residue data. The text doesn't write that out. If E_{1,i} had larger dimension than E_{0,i}, the first formula in (4) would overcount n_{f,i/d}. I don't see a counterexample, and the recovered special cases suggest the constancy holds, but a referee should ask for a full proof of that step rather than taking it on faith.\n\nMinor point: the Singular code is embedded in the text; not a real problem, but a separate artifact would be better.\n\nAll in all, this is a specialist's paper, for people computing spectra of cones, Bernstein–Sato polynomials, and testing the monodromy conjecture. It deserves a serious referee. I'd send it out, with the explicit request that the deformation-invariance argument be pinned down. If the authors can close that gap, it's publishable as is; if not, the theorem stands but the proof needs a repair.\n\nBring it to the reading group if you want to see what current mixed-Hodge-module arguments look like in action. I would likely cite it if I were working in that area.","headline":"Genuine extension of Budur–Saito/Yoon to semi-weighted-homogeneous non-reduced curves; the result looks right but the key deformation step needs a fuller proof.","tokens_in":21450,"tokens_out":2654,"would_cite":true,"duration_ms":25095,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14B05","32S25","32S40","14C30"],"pacs":[],"model":"deepseek-v4-flash","headline":"The spectrum of the cone over a projective hypersurface is determined by the spectral numbers at the singular points of its underlying reduced hypersurface and the global degree.","keywords":["spectrum","cone of a projective hypersurface","semi-weighted-homogeneous singularity","non-reduced plane curves","mixed Hodge modules","local systems","vanishing cohomology","plane curve singularities"],"falsifier":"Take a non-reduced plane curve whose reduced curve has a semi-weighted-homogeneous but not weighted-homogeneous singularity, for instance one of the families computed in the examples, compute its spectrum by an independent embedded resolution or by a certified computer algebra routine, and compare every coefficient with formula (4); a single mismatch at any spectral number would refute the theorem.","tokens_in":20448,"feed_emoji":"📐","tokens_out":8400,"duration_ms":77654,"temperature":0.7,"pith_summary":"This paper tries to establish that the spectrum of the cone over a projective hypersurface—a rational-number-valued invariant encoding the Hodge numbers and monodromy of the Milnor fiber—can be recovered from local data at the singular points of the underlying reduced hypersurface together with the global degree. For a reduced hypersurface with only isolated singularities, Theorem 1 gives such a formula, and when the irreducible components carry constant multiplicities (automatically true in dimension at least two), it shows that non-reduced multiplicities only shift the spectral numbers in a prescribed way. For non-reduced plane curves, Theorem 2 goes further under the assumption that every singularity of the reduced curve is semi-weighted-homogeneous, meaning that each local branch has a weighted homogeneous lowest-degree part: the full spectrum is expressed through local weights, local weighted degrees and multiplicities, and global degrees and multiplicities. The formulas count lattice points in triangles and reduce to the known formula for reduced line arrangements. If correct, this makes the spectrum of such cones a combinatorial quantity rather than something requiring a global resolution.","feed_headline":"Cone spectra reduce to local singularity data","feed_subtitle":"New closed formulas compute the spectrum of non-reduced plane-curve cones from local weights and degrees alone.","key_machinery":"The central object is the Deligne extension $L_i$ of a rank-one local system on the complement of the hypersurface, whose monodromy around each component is governed by the multiplicity of that component. The proof computes the cohomology of the sheaf of logarithmic differential two-forms on the plane twisted by $L_i$; after a weighted blowup at each singular point, it shows that the higher direct images vanish and that the remaining cokernel has dimension equal to the lattice-point count $N_j(\\lceil\\gamma_{j,i}\\rceil-1)$. The semi-weighted-homogeneous case is reached from the weighted-homogeneous case through a $\\mu$-constant deformation together with Nakayama's lemma, which propagates the vanishing of the higher direct images across the deformation. Theorem 1 is proved separately through a $d$-fold cyclic covering and a Thom–Sebastiani-type decomposition of the spectrum.","core_discovery":"The central claim, stated in the paper's own terms, is that the spectrum of a homogeneous defining polynomial of a cone is determined by the spectrum of the reduced polynomial together with simple combinatorial bookkeeping. Theorem 1 states that for a reduced projective hypersurface $Z'$ with isolated singularities, the coefficient $n_{f',i/d'}$ is a binomial count minus the sum, over singular points, of the number of spectral numbers at that point lying in the interval $[i/d'-1,i/d')$. When the hypersurface is non-reduced with constant multiplicities, $f=f'^m$, the coefficients of the cone are obtained from those of $f'$ by shifting the spectral parameter and adding $(-1)^n$ only at one endpoint. In the plane curve case with $f=\\prod_{k=1}^r f_k^{a_k}$ and $C_{\\rm red}$ having only semi-weighted-homogeneous singularities, Theorem 2 gives explicit closed formulas: for $i\\in[1,d]$, the coefficient $n_{f,i/d}$ equals $\\binom{\\iota_i-1}{2}-\\sum_j N_j(\\lceil\\gamma_{j,i}\\rceil-1)$, and the coefficient $n_{f,i/d+2}$ equals $\\binom{d'-\\iota_i-1}{2}-\\sum_j N_j(d_j-\\lceil\\gamma_{j,i}\\rceil)-\\delta_{i,d}$, with the middle coefficient $n_{f,i/d+1}$ determined by an Euler-characteristic identity. The local data involved are the weights $w_j,w'_j$, the weighted degrees $d_{j,l}$ and multiplicities $a_{j,l}$ of local irreducible components, and the global degrees $d'_k$ and multiplicities $a_k$; no global embedded resolution is needed.","pith_inferences":["Because each $N_j(n)$ counts pairs $(m_1,m_2)$ with $w_jm_1+w'_jm_2\\le n$, the spectral coefficients should behave quasi-polynomially as the global degrees and multiplicities vary; this periodicity is a testable prediction not drawn out in the paper.","The same twisted-logarithm computation may extend to non-reduced higher-dimensional hypersurfaces whose components have nonconstant multiplicities, where the present constant-multiplicity assumption is not available.","Since the formula reduces $n_{f,3/d}$ to a lattice-point comparison, it turns a piece of the strong monodromy conjecture in this class into a finite combinatorial check, potentially enabling a systematic search for counterexamples or confirming the conjecture on larger families.","The deformation-invariance step, if it holds more generally, suggests that the local contribution of each semi-weighted-homogeneous singularity depends only on its weights and weighted degree, not on the global position of the curve; this would make the formula robust under perturbations of the global components."],"forward_implications":["For any reduced plane curve, the spectrum of its cone is determined by the spectral numbers at the singular points and the degree; in particular, the number of ordinary double points does not affect the coefficients $n_{f,i/d}$, $n_{f,i/d+1}$, and $n_{f,i/d+2}$.","For non-reduced plane curves whose reduced curve has only semi-weighted-homogeneous singularities, the full spectrum can be computed from local weights, local weighted degrees and multiplicities, and global degrees and multiplicities, without resolving the whole configuration.","When all singularities are ordinary, the formula for $n_{f,i/d+1}$ is equivalent to the previously known characteristic-class computation in the ordinary singularity case.","The reduced line arrangement formula is recovered as the special case where all multiplicities and degrees equal one and the local weights are $w_j=w'_j=1$.","For higher-dimensional hypersurfaces with constant component multiplicities, the spectrum of the cone is obtained from the reduced spectrum by a simple shift and endpoint correction, thanks to Theorem 1."],"supporting_citations":[{"why":"It defines the mixed Hodge structure on vanishing cohomology and the spectrum used throughout the paper.","marker":"[St 77]"},{"why":"It supplies the reduced line arrangement formula that Theorem 2 and Corollary 1 generalize.","marker":"[BuSa 10]"},{"why":"It gives the ordinary singularity case formula that Theorem 2 and Corollary 2 reproduce.","marker":"[Yo 19]"},{"why":"It provides the mixed Hodge module machinery for nearby cycles, Deligne extensions, and direct images used in the proofs.","marker":"[Sa 90]"},{"why":"It is used together with $\\mu$-constant deformations to reduce semi-weighted-homogeneous spectra to weighted-homogeneous ones.","marker":"[DMST 06]"},{"why":"It supplies the $\\mu$-constant invariance of the spectrum that lets the weighted-homogeneous vanishing propagate.","marker":"[Va 82b]"},{"why":"It fixes the relation between the geometric monodromy and the monodromy action underlying the local systems $L_i$.","marker":"[DiSa 14]"},{"why":"It provides the intersection-number identity used to pass from local to global multiplicities in Corollary 2.","marker":"[Fu 84]"},{"why":"It underlies the twisted logarithmic complex comparison for positively weighted homogeneous divisors used in the weighted blowup calculation.","marker":"[BaSa 24]"}],"fun_headline_variants":["Local singularity data decides cone spectrum","No global resolution needed for cone spectrum","Cone spectrum from local spectral numbers","Non-reduced plane-curve cones: explicit spectra"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof assumes that the vanishing of the higher direct images of the twisted logarithmic sheaf, verified in the weighted-homogeneous case, persists under a $\\mu$-constant deformation to any semi-weighted-homogeneous singularity; if that deformation-invariance fails, the dimension count behind the formula breaks.","fun_headline_variants_meta":{"raw":{"variants":["Local singularity data decides cone spectrum","No global resolution needed for cone spectrum","Cone spectrum from local spectral numbers","Non-reduced plane-curve cones: explicit spectra"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.002128,"raw_usage":{"total_tokens":8317,"prompt_tokens":1059,"completion_tokens":7258,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":675,"completion_tokens_details":{"reasoning_tokens":7205}},"tokens_in":675,"tokens_out":7258,"duration_ms":50855,"temperature":1.0,"reasoning_tokens":7205,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T10:56:59.167723+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a non-reduced plane curve whose reduced curve has a semi-weighted-homogeneous but not weighted-homogeneous singularity, for instance one of the families computed in the examples, compute its spectrum by an independent embedded resolution or by a certified computer algebra routine, and compare every coefficient with formula (4); a single mismatch at any spectral number would refute the theorem.","supporting_citations":[],"review_version":1}