{"id":"ec1a6dc9-8719-4397-a3b3-d4645bc28f9e","arxiv_id":"1908.01749","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"For every k, the k-torsion freeness of the nth-order Kähler differential module of a hypersurface point is equivalent to the singular locus having codimension at least k+1.","lead":"This paper proves that a module of high-order Kähler differentials of a hypersurface is k-torsion-free exactly when the hypersurface is smooth away from a singular locus of codimension at least k. It generalizes Lipman's classical criterion for ordinary differentials to arbitrary k and high order.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central theorem rests entirely on the imported bound projdim(Ω^(n)_R/K) ≤ 1 from [4, Thm 4.3]; the paper does not reprove it, so a failure of that bound would invalidate Theorem 4.5 and Corollary 4.4.","rationale":"Read in good faith, the paper's main theorem is a clean reduction: once the module of high-order differentials is known to have projective dimension at most one, Theorem 3.4 together with the support computation in Corollary 4.4 yields the codimension criterion for every k. The internal steps are coherent and no contradiction with Lipman's theorem or the k=1 case in [4] is apparent. The single least secure point is exactly the imported bound, because the entire structure collapses without it and no proof or computational cross-check is supplied in this preprint. The secondary issue about Q being non-local in Theorem 3.4 does not threaten Theorem 4.5, where R is a domain and Q is a field. Since the missing part is in a published source and the current proofs are otherwise sound, the appropriate verdict remains CONDITIONAL rather than ACCEPT.","tokens_in":7372,"tokens_out":25298,"duration_ms":256478,"concrete_test":"Verify [4, Theorem 4.3] directly for a nontrivial case: using the presentation of Ω^(n)_R/K from [4, Theorem 2.8], compute the projective dimension of Ω^(n)_R/K for a singular hypersurface with s = 2, e.g., R = k[[x,y]]/(x^2 - y^3), for n = 2 and n = 3, using Macaulay2 or a direct free-resolution computation. If projdim > 1 for either n, the central theorem 4.5 is false; if the bound holds and the proof in [4] is sound, the conditional verdict can be upgraded.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The proof of Theorem 4.5 is a reduction to two external results from [4]: the projective-dimension bound projdim_R(Ω^(n)_R/K) ≤ 1 and, in Proposition 4.1, the regularity criterion [4, Thm 3.1]. The bound is the more load-bearing: it is what makes Ω^(n)_R/K admit a length-one free resolution, identifies D(Ω^(n)_R/K) with Ext^1_R(Ω^(n)_R/K,R), licenses Lemma 4.3 to convert vanishing of Ext^1_Rp(Ω^(n)_Rp/K,Rp) into freeness in Corollary 4.4, and is a hypothesis of Theorem 3.4. The paper cites [4, Thm 4.3] without reproducing or even sketching the proof, and Remark 4.7 reports that the authors could not compute any example for n > 1. If the bound failed for some n, the equivalence in Theorem 4.5 would have no support. A secondary but related gap is that Theorem 3.4's proof applies the Auslander–Buchsbaum formula to the total quotient ring Q, which is not local in general; for the hypersurface-domain case Q is a field, so Theorem 4.5 itself is not endangered by this, but the general theorem as stated is not fully justified.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the R-module Ω^(n)_R/K of Kähler differentials of order n for the local ring R of a closed point on an irreducible hypersurface W over a perfect field K. Its main theorem (Theorem 4.5) asserts that Ω^(n)_R/K is k-torsion free if and only if codim(R/p) ≥ k+1 for every prime p in the singular locus Sing(R). The proof follows Lipman's strategy: Proposition 4.1 gives a regularity criterion in terms of freeness of high-order differentials; Corollary 4.4 identifies the support of Ext^1_R(Ω^(n)_R/K,R) with Sing(R); and Theorem 3.4 converts vanishing of Ext^i_R(D(M),R) into grade conditions, using the imported bound projdim(Ω^(n)_R/K)≤1 from [4]. Section 3 develops a general k-torsion freeness criterion for modules of projective dimension at most one.","tokens_in":7639,"tokens_out":15355,"duration_ms":156994,"significance":"If the main theorem is correct, it gives a clean and elegant characterization: k-torsion freeness of high-order differentials of a hypersurface is controlled entirely by codimensions of singular primes, generalizing Lipman's classical theorem and the n=1 result of [4] to all n and k. The paper is clearly written and honest about its limitations, with Remark 4.7 stating that no example for n>1 could be computed. It also gives self-contained module-theoretic tools (Lemma 3.3, Theorem 3.4) that may be useful beyond hypersurfaces. The central argument, conditional on the cited projective dimension bound, is coherent. The main reservations are a proof gap in the general Theorem 3.4 and an inconsistent citation for the projective dimension bound.","major_comments":[{"comment":"The proof applies the Auslander-Buchsbaum formula to the total quotient ring Q as if Q were a local ring, writing 0 = depth(Q) = projdim(M_Q)+depth(M_Q). For a general Noetherian local ring R the total quotient ring Q need not be local; for example, if R is the local ring of a reducible hypersurface such as K[x,y]/(xy) at a maximal ideal, Q is a nontrivial product of fields and not local. Depth is not defined globally for such Q and the formula as stated is not justified. The argument can be repaired by localizing at the maximal ideals of Q and using that every such localization has depth 0, but as written the proof of the general theorem is incomplete. Since Theorem 4.5 is applied only to a domain, where Q is a field, the main theorem is not endangered, but Section 3 states a broader result and should be corrected.","section":"Section 3, proof of Theorem 3.4"},{"comment":"The proof of Theorem 4.5 rests on the bound projdim(Ω^(n)_R/K)≤1, quoted in Corollary 4.4 as '[4, Theorem 4.3]'. The same reference is used in Theorem 2.3 for the assertion that Ω^(n)_R/K is torsion free if and only if W is normal at P. These cannot both be the same statement unless [4, Theorem 4.3] contains both claims, which the manuscript does not state. Since every application of Lemma 4.3 and Theorem 3.4 requires projdim≤1, the equivalence in Theorem 4.5 has no support if this bound fails; a precise and correct citation of the source of the bound is therefore load-bearing. The authors should identify exactly which theorem in [4] supplies the bound and, if space permits, give a short indication of why it holds.","section":"Section 4, Corollary 4.4 and Theorem 4.5"},{"comment":"Proposition 4.1 uses the regularity criterion [4, Theorem 3.1] as a black box for the converse direction. This is acceptable for a published theorem, but since it is a second external result on which the main theorem depends, the statement of [4, Theorem 3.1] should be reproduced verbatim in the paper, and the authors should clearly record its exact hypotheses (e.g., whether it requires the maximal ideal to be closed and the field to be perfect).","section":"Section 4, Proposition 4.1"}],"minor_comments":[{"comment":"The manuscript contains numerous typographical and OCR artifacts (e.g., 'HYPERSURF ACES' in the title, 'OBject', 'equiva lent', 'differentials'); the authors should proofread the final version carefully.","section":"Throughout"},{"comment":"In the converse direction, the sentence 'there exists m ⊂ Ag a maximal ideal such that p ⊂ m' is correct but could be clarified: any prime ideal of Ag is contained in a maximal ideal, and since U = Spec(Ag), the chosen m automatically lies in U.","section":"Section 4, proof of Proposition 4.1"},{"comment":"The theorem statement uses the phrase 'W is non-singular in codimension k at P', which is defined in Section 2 in terms of codim(R/p) ≥ k+1 for all p ∈ Sing(R). Repeating this definition in Theorem 4.5 would make the statement more self-contained.","section":"Section 4, Theorem 4.5"},{"comment":"Equation (2) asserts grade(M) = min{depth(Rp) : p ∈ Supp(M)}. This requires M to be finite and R Noetherian, which is assumed, but the passage would be clearer if it noted explicitly that D(M) is finite whenever M is finite with a length-one projective resolution.","section":"Section 3, equation (2)"}],"recommendation":"major_revision","confidential_remarks":"The manuscript leans heavily on [4], co-authored by one of the present authors, for both the projective dimension bound and the regularity criterion. This is legitimate if [4] is published, but the inconsistent citation of [4, Theorem 4.3] in Theorem 2.3 versus Corollary 4.4 should be corrected. The novelty over [4] is a genuinely broader k-torsion statement, though the module-theoretic machinery in Section 3 is standard. The main theorem is plausible and the proof is likely repairable; the issues identified above do not appear to require a change in the central claim, so I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear —,\n\nThe headline is that Theorem 4.5 looks right: for a hypersurface local ring, Ω^(n)_R/K is k-torsion free iff the singular locus has codimension at least k+1, for every n and every k. This genuinely extends Lipman's classical criterion and the k=1 version previously proved for high-order differentials. The proof is a clean reduction: Theorem 3.4 gives a module-level criterion, Corollary 4.4 identifies the support of Ext^1(Ω^(n)_R/K, R) with Sing(R), and the main equivalence follows directly. The paper does not just re-coordinatize existing results; Theorem 3.4 is a genuinely more general statement about modules of projective dimension at most one.\n\nThe main soft spot is in the proof of Theorem 3.4. The authors apply Auslander–Buchsbaum to the total quotient ring Q, treating it as a local ring of depth zero. That is automatic when R is a domain, which covers the hypersurface case, but the theorem as stated for arbitrary Noetherian local rings is not fully justified. A referee should ask them to either restrict the statement to domains or handle the semi-local Artinian case properly. This gap does not endanger Theorem 4.5.\n\nThe other caveat is the heavy reliance on [4] for the projective dimension bound projdim(Ω^(n)_R/K) ≤ 1 and for the regularity criterion in Proposition 4.1. This is a published result by one of the present authors, so citing it is legitimate, but it is genuinely load-bearing. If that bound failed for some n, the equivalence in Theorem 4.5 would have no support. The paper does not reprove or sketch the bound, and Remark 4.7 honestly reports that the authors could not compute any examples for n > 1. This makes the final theorem conditional on [4] being correct, not circular.\n\nI think the paper deserves a serious referee. The main theorem is plausible, the strategy is sound, and the writing is clear. The gaps are fixable. I would send it to review and ask the authors to tighten Theorem 3.4 and state the relevant results from [4] in enough detail that a reader can check the hypotheses. I would cite this in work on differential modules and singularity theory.","headline":"A solid generalization of Lipman's k-torsion criterion to high-order differentials of hypersurfaces, with a small gap in the general module-level theorem.","tokens_in":8209,"tokens_out":3556,"would_cite":true,"duration_ms":35215,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["13N05","13D07","13C12"],"pacs":[],"model":"deepseek-v4-flash","headline":"For hypersurfaces, the k-torsion freeness of every order-n differential module is determined by the codimension of the singular locus.","keywords":["k-torsion","high order differentials","hypersurfaces","module of differentials","singular locus","projective dimension","Cohen-Macaulay","reflexive modules"],"falsifier":"Take the cusp ring $R=K[x,y]/(y^2-x^3)$ localized at the origin. Its singular prime has codimension 1, so the theorem predicts that $\\Omega^{(n)}_{R/K}$ is not torsion-free for any $n$; showing that some $n$ gives a torsion-free module would refute the central claim.","tokens_in":7116,"feed_emoji":"📐","tokens_out":13712,"duration_ms":119842,"temperature":0.7,"pith_summary":"The paper establishes that, for the local ring of a closed point on an irreducible hypersurface over a perfect field, the module of differentials of order $n$ is $k$-torsion free exactly when every singular prime has codimension at least $k+1$. This extends the classical torsion-free versus codimension-one criterion and the reflexive versus codimension-two criterion from order $1$ to all orders $n$ and all positive integers $k$. A sympathetic reader should care because it turns a homological property of every high-order differential module into a single geometric condition on the singular locus, with no module computation needed.","feed_headline":"A codimension bound decides k-torsion in hypersurface differentials","feed_subtitle":"High-order differentials are k-torsion free exactly when singular primes sit in codimension at least k+1.","key_machinery":"The transpose module $D(M)$, defined as the cokernel of the dual of a presentation map $P_1 \\to P_0 \\to M$, carries the argument: $k$-torsion freeness is the vanishing of $\\operatorname{Ext}^i_R(D(M),R)$ for $i=1,\\dots,k$. For any module with projective dimension at most one, Theorem 3.4 converts this into a depth condition on the support of $D(M)$, and Corollary 4.4 converts that support into the singular locus of the hypersurface.","core_discovery":"Theorem 4.5 states that, with $R$ the local ring of a closed point on an irreducible hypersurface over a perfect field, the module $\\Omega^{(n)}_{R/K}$ is $k$-torsion free if and only if $\\operatorname{codim}(R/\\mathfrak{p}) \\geq k+1$ for every prime $\\mathfrak{p}$ in the singular locus of $R$. The proof identifies the support of $\\operatorname{Ext}^1_R(\\Omega^{(n)}_{R/K}, R)$ with the singular locus, then applies a general criterion for modules of projective dimension at most one: $k$-torsion freeness is equivalent to $\\operatorname{depth}(R_{\\mathfrak{p}}) \\geq k+1$ on the support of the transpose module $D(M)$, which becomes the codimension condition because $R$ is Cohen-Macaulay.","pith_inferences":["Editorial inference: the same strategy would prove the analogous statement for reduced complete intersections if the projective dimension bound carries over; the authors state this conditional in Remark 4.7 but do not settle it.","Editorial inference: the theorem implies that the largest $k$ for which $\\Omega^{(n)}_{R/K}$ is $k$-torsion free is $\\min_{\\mathfrak{p}\\in\\operatorname{Sing}(R)} \\operatorname{codim}(R/\\mathfrak{p}) - 1$, a single integer invariant of the singularity that the paper does not name.","Editorial inference: the explicit presentation of $\\Omega^{(n)}_{R/K}$ available for general finitely generated algebras could be used to test the projective dimension bound computationally; the authors report that the matrices become too large for examples with $n>1$."],"forward_implications":["Checking $k$-torsion freeness of $\\Omega^{(n)}_{R/K}$ for any $n$ reduces to checking the codimension of singular primes; no direct computation of Ext modules is required.","Setting $k=1$ recovers the statement that the module is torsion free exactly when the hypersurface is normal at the point, and $k=2$ gives reflexivity exactly when it is non-singular in codimension $2$.","Because the codimension condition does not depend on $n$, for a fixed hypersurface and fixed $k$ the property holds for every order $n$ or for none.","The general Theorem 3.4 provides a $k$-torsion criterion for any finite module of projective dimension at most one, independent of differentials."],"supporting_citations":[{"why":"Defines k-torsion freeness as the vanishing of Ext^i_R(D(M),R) for i=1,...,k, the notion the paper characterizes.","marker":"[3]"},{"why":"Supplies the projective dimension bound projdim(Omega^(n)_{R/K}) <= 1 and the maximal-ideal regularity criterion used in Proposition 4.1 and Corollary 4.4.","marker":"[4]"},{"why":"Provides the grade and depth formulas that turn the vanishing of Ext modules into depth and codimension conditions in Theorem 3.4.","marker":"[6]"},{"why":"Gives the freeness and rank computation for Omega^(n) at regular localizations, used in Proposition 4.1.","marker":"[9]"},{"why":"States the n=1 torsion and reflexivity theorems and supplies the proof strategy of identifying the support of Ext^1 with the singular locus.","marker":"[10]"},{"why":"Shows that high-order differentials commute with localization, used in Corollary 4.4 to pass from R to R_p.","marker":"[12]"}],"fun_headline_variants":["Codim ≥ k+1: the torsion-free threshold for differentials","Hypersurface differentials: torsion-free iff codim of singular primes ≥ k+1","k-torsion freeness of differentials: exactly codim ≥ k+1 on singular locus","Order n differentials: k-torsion-free iff singular codim ≥ k+1","k-torsion in differentials vanishes if singular locus sits in codim k+1"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof depends on the imported bound that $\\Omega^{(n)}_{R/K}$ has projective dimension at most one; if that bound ever fails for a hypersurface, the support computation and the $k$-torsion criterion no longer apply.","fun_headline_variants_meta":{"raw":{"variants":["Codim ≥ k+1: the torsion-free threshold for differentials","Hypersurface differentials: torsion-free iff codim of singular primes ≥ k+1","k-torsion freeness of differentials: exactly codim ≥ k+1 on singular locus","Order n differentials: k-torsion-free iff singular codim ≥ k+1","k-torsion in differentials vanishes if singular locus sits in codim k+1"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00109,"raw_usage":{"total_tokens":4456,"prompt_tokens":748,"completion_tokens":3708,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":364,"completion_tokens_details":{"reasoning_tokens":3594}},"tokens_in":364,"tokens_out":3708,"duration_ms":24439,"temperature":1.0,"reasoning_tokens":3594,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:04:37.706990+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the cusp ring $R=K[x,y]/(y^2-x^3)$ localized at the origin. Its singular prime has codimension 1, so the theorem predicts that $\\Omega^{(n)}_{R/K}$ is not torsion-free for any $n$; showing that some $n$ gives a torsion-free module would refute the central claim.","supporting_citations":[{"cited_title":"Auslander, M","cited_arxiv_id":null,"evidence_quote":"Defines k-torsion freeness as the vanishing of Ext^i_R(D(M),R) for i=1,...,k, the notion the paper characterizes."},{"cited_title":"Barajas, D","cited_arxiv_id":null,"evidence_quote":"Supplies the projective dimension bound projdim(Omega^(n)_{R/K}) <= 1 and the maximal-ideal regularity criterion used in Proposition 4.1 and Corollary 4.4."},{"cited_title":"Bruns, J","cited_arxiv_id":null,"evidence_quote":"Provides the grade and depth formulas that turn the vanishing of Ext modules into depth and codimension conditions in Theorem 3.4."},{"cited_title":"Laksov, A","cited_arxiv_id":null,"evidence_quote":"Gives the freeness and rank computation for Omega^(n) at regular localizations, used in Proposition 4.1."},{"cited_title":"Lipman; Free derivation modules on algebraic varieties , Amer","cited_arxiv_id":null,"evidence_quote":"States the n=1 torsion and reflexivity theorems and supplies the proof strategy of identifying the support of Ext^1 with the singular locus."},{"cited_title":"Nakai; High order derivations I , Osaka J","cited_arxiv_id":null,"evidence_quote":"Shows that high-order differentials commute with localization, used in Corollary 4.4 to pass from R to R_p."}],"review_version":1}