{"id":"2250d632-57b2-44da-ba63-dc792e5b47e8","arxiv_id":"1908.02167","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Over a hypersurface ring, if M has finite projective dimension and N is locally free at primes of height at most one, then reflexivity of M tensor N implies both M and N are reflexive.","lead":"This paper proves a new rigidity theorem over hypersurface rings: if one module has finite projective dimension and the other is locally free at all height-one primes, then a reflexive tensor product forces both modules to be reflexive. It also sharpens a known counterexample, showing exactly what goes wrong when the local freeness condition is dropped.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified; the only defect is a fixable typo in Proposition 4.3(i).","rationale":"The reader identified the connectedness of the Hochster–Huneke graph as the weakest assumption, and I agree that Proposition 4.3 is the pivotal lemma. However, the connectedness statement itself is true and its proof is repairable by a one-line correction: the typo 'u is R-regular' should read 'v is R-regular'. The rest of the proof of Theorem 1.2 is coherent: the use of finite CI-dimension over a hypersurface is standard, the local freeness on Y1 follows from finite projective dimension together with Serre's condition, and the induction in Theorem 3.1 is internally consistent once the depth inequalities for CI-dimension are supplied. I therefore find no load-bearing concern that would change the ACCEPT verdict. The independent support of the argument is cumulative: the theorem is a sharpening of a known rigidity result, the construction is explicit, and the proof checks out apart from the typographical error noted above.","tokens_in":9128,"tokens_out":31805,"duration_ms":356169,"concrete_test":"Rewrite the final algebra of Proposition 4.3(i) with the regular element v in place of u: from v = a(u+v) with a a unit, conclude v is R-regular; then for every u in I, uv = 0, so u = 0 and I = 0, a contradiction. If this amended line yields the contradiction without assuming u is regular, the connectedness lemma stands as a theorem, with only a typo corrected.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim survives scrutiny. The truly load-bearing step is Proposition 4.3(i), since Corollary 4.4 and hence Theorem 1.2 depend on connectedness of the Hochster–Huneke graph. In that proof, the equation v'(u+v) - v(u'+v') = 0 correctly implies, via regularity of the sequence {u+v, u'+v'}, that v = a(u+v) and u = b(u+v). From a+b = 1, at least one of a,b is a unit; but if a is a unit, the regular element obtained is v, not u as printed. The sentence 'Then u is R-regular, and the equality uJ = 0 shows that J = 0' is therefore not literally correct, e.g. when a = 1 and b = 0. The contradiction still follows: v = a(u+v) is R-regular and lies in J, so for every u in I, uv = 0 forces u = 0, hence I = 0, contradicting non-nilpotence of I. This is a typographical slip in an otherwise valid argument, not a substantive gap. A second understated input is the step in Theorem 3.1 where depth(N_q) ≥ depth(R_q) is used to conclude CI-dim(N_q) = 0; this relies on the standard inequality depth ≤ depth R for nonzero modules of finite CI-dimension, which is valid but not explicitly cited. Neither issue threatens Theorem 1.2.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper revisits an example of Celikbas and Takahashi concerning the failure of both modules in a tensor product to be reflexive in the Second Rigidity Theorem of Huneke and Wiegand. The authors prove a general result, Theorem 3.1, stating that if M is Tor-rigid, N has finite complete intersection dimension, M⊗N satisfies the condition (~S_n), and Tor_i(M,N) is torsion for all i≫0, then all positive Tor modules vanish and N satisfies (~S_n). Under the additional hypotheses that R satisfies (S2) and N is locally free on Y_1(R), Corollary 4.4 strengthens the conclusion to both M and N satisfying (~S_n). This yields Theorem 1.2 for hypersurface rings: if M has finite projective dimension and N has finite projective dimension after localization at every prime of height at most one, then reflexivity of M⊗N forces both M and N to be reflexive. The paper also analyzes the Hochster–Huneke graph and proves connectedness without a completeness assumption, and it revisits the Celikbas–Takahashi example to demonstrate sharpness.","tokens_in":9403,"tokens_out":26546,"duration_ms":257539,"significance":"If correct, the paper gives a clean positive result that rules out the known counterexample to the Second Rigidity Theorem under a mild local finiteness condition on N. The main theorem is more general than the hypersurface setting, working with complete intersection dimension and the (~S_n) conditions, and the proof is largely self-contained, with useful preliminary lemmas on transposes, pushforwards, and Tor-rigidity. The paper also supplies a proof of connectedness of the Hochster–Huneke graph in the non-complete case, and it provides an explicit presentation of the tensor product in the Celikbas–Takahashi example. These are genuine contributions to the homological algebra of tensor products and rigidity.","major_comments":[{"comment":"The definition of Tor-rigidity in §2.1 is the weak one: Tor^R_1(M,N)=0 implies Tor^R_2(M,N)=0. However, the proof of Theorem 3.1 uses the standard strong form of Tor-rigidity: after showing Tor^R_1(M,N_1)=0, the text concludes 'As M is Tor-rigid, we have Tor^R_i(M,N)=0 for each i≥1.' This inference requires the strong property Tor^R_i=0 ⇒ Tor^R_{i+1}=0 for all i≥1, not merely the i=1 case stated in §2.1. Since the examples cited in §2.1 (Lichtenbaum, Huneke–Wiegand) do satisfy the strong form, the gap is easily fixed by amending the definition, but as written the proof is not valid under the stated definition.","section":"§2.1 and proof of Theorem 3.1"},{"comment":"In the final paragraph of the proof, after assuming that a is a unit, the element that is shown to be regular is v=a(u+v), not u. The sentence 'Then u is R-regular, and the equality uJ=0 shows that J=0' is therefore not literally correct. The contradiction still follows because v is R-regular and v∈J, so for every u∈I the relation uv=0 forces u=0, giving I=0. This is a typographical slip rather than a substantive gap, but it should be fixed.","section":"§4, Proposition 4.3(i)"}],"minor_comments":[{"comment":"In the proof of part (ii), the module is N, but several occurrences read 'M_{p_i}', 'M_{q_{i+1}}', and 'rank_{R_p}(M_p)'. These should be N throughout to avoid confusion.","section":"§4, Proposition 4.3(ii)"},{"comment":"The displayed exact sequence is hard to parse because the matrices and the module M⊗N are not clearly connected by arrows, and the phrase 'rightmost matrix' is ambiguous. Please rewrite the sequence with labeled maps and indicate explicitly which cokernel is M⊗R N. Also, since the verification relies on unpublished Macaulay2 code [21], it would be helpful to include the code or a reproducible computation.","section":"Example 4.5"},{"comment":"The reduction to CI-dim_{R_p}(N_p)=0 for p∈Y_0(R) uses the fact that finite CI-dimension satisfies the Auslander–Buchsbaum type formula CI-dim(N_p)=depth(R_p)-depth(N_p). This is standard, but it would be good to cite it explicitly, since the argument in that lemma depends on it.","section":"Lemma 3.3"},{"comment":"The examples of Tor-rigid modules listed in §2.1 should state explicitly that they satisfy the strong Tor-rigidity property used in Theorem 3.1, not merely the weak form given in the definition.","section":"§2.1"},{"comment":"There are a few typographical issues: 'syzgy' should be 'syzygy' in the text after 2.2, 'The inequality in (3.1.5) are due to' should be 'is due to', and 'enviroment' in the acknowledgments should be 'environment'.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The paper is mathematically sound in its main lines, and the issues I raise are local and fixable. The most important point is the definition of Tor-rigidity in §2.1, which needs to be strengthened to the standard strong form; otherwise the proof of Theorem 3.1 does not go through as written. The Example 4.5 verification also depends on unpublished code, but this is not load-bearing for the main theorem. I recommend major revision because a proof step in the central theorem needs a nontrivial clarification, even though I expect the authors can resolve it quickly."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a solid paper with a clean new criterion. It doesn't settle the reflexivity question for M in Huneke–Wiegand's Second Rigidity Theorem, but it shows that the Celikbas–Takahashi counterexample cannot occur once N has finite projective dimension locally on the height-one primes. The main theorem holds up.\n\nWhat is new: Theorem 3.1 is a genuinely general statement—Tor-rigid M, finite CI-dimension N, tensor product satisfying (~S_n), and torsion high Tor force all positive Tor to vanish and N to satisfy (~S_n). Corollary 4.4 upgrades the conclusion to both M and N when N is locally free on Y1(R), using connectedness of the Hochster–Huneke graph. Theorem 1.2 is the hypersurface packaging, and it is exactly the right boundary condition to rule out the [10] example. The new explicit presentation of M ⊗ N in Example 4.5 is a useful addition. The proof strategy is coherent, and the graph-connectedness lemma is proven in the text rather than taken for granted.\n\nSoft spots are minor. The stress-test note is right: in Proposition 4.3(i), the sentence 'Then u is R-regular' is a typo—when a is the unit in the argument, the regular element is v, not u. The contradiction still goes through (v regular and in J forces I = 0), so this is cosmetic, not load-bearing. Also, in Theorem 3.1 the step where depth(R_q) ≥ depth(N_q) is used to conclude CI-dim(N_q)=0 relies on the standard inequality depth ≤ depth R for modules of finite CI-dimension; it is valid but not explicitly cited. Finally, Example 4.5's verification depends on a Macaulay2 computation via an unpublished note [21]; that is acceptable for an example, and the main theorem does not depend on it. The citation pattern is clean: several references are by overlapping authors, but they are distinct published results and the central argument does not lean on the example.\n\nWho it's for: commutative algebraists working on tensor products, reflexivity, and rigidity of Tor. It deserves a serious referee; I would send it out. After a fix of the typo, I'd be happy with acceptance.","headline":"A solid, honest paper that sharpens the boundary of Huneke–Wiegand's Second Rigidity Theorem, with only a fixable typo in the graph-connectedness proof.","tokens_in":9947,"tokens_out":3760,"would_cite":true,"duration_ms":33742,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["13D07","13H10","13D05","13C12"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that, for a hypersurface ring, reflexivity of a tensor product forces both factors to be reflexive once the second factor is locally free in codimension one.","keywords":["reflexivity of tensor products of modules","Tor-rigidity","condition (S2)","syzygy","vanishing of Tor","minimal-prime graph","complete intersection dimension","hypersurface ring"],"falsifier":"Localize the known counterexample at the height-one prime $\\mathfrak q=(x,y)$ and resolve $N_{\\mathfrak q}$: the minimal free resolution alternates multiplication by $x$ and $y$ forever, so $\\operatorname{pd}_{R_{\\mathfrak q}}(N_{\\mathfrak q})=\\infty$, verifying that this is precisely where the added hypothesis excludes the example. A single example satisfying all hypotheses of Theorem 1.2 with $M$ not reflexive would refute the claim; a computational search over small hypersurface presentations, checking the biduality map $M \\to M^{**}$ explicitly, could be used to look for one.","tokens_in":8942,"feed_emoji":"🧮","tokens_out":10580,"duration_ms":94713,"temperature":0.7,"pith_summary":"This paper establishes a complement to the Second Rigidity Theorem: a reflexive tensor product can force both factors to be reflexive, not merely the factor with finite projective dimension. Over a hypersurface ring, if $M$ has finite projective dimension and $N$ is locally of finite projective dimension at every prime of height at most one, then reflexivity of $M \\otimes_R N$ implies reflexivity of both $M$ and $N$. The proof is carried by a connectedness theorem for the graph of minimal prime ideals, which gives $N$ a well-defined rank under the condition $(S_2)$, together with the depth formula for modules of finite complete intersection dimension. It also shows that a previously constructed counterexample, where the tensor product and $N$ are reflexive but $M$ is not, escapes exactly because $N$ has infinite projective dimension at a height-one prime. A broader theorem replaces reflexivity by the Serre-type condition $(\\tilde S_n)$ and replaces finite projective dimension of $M$ by Tor-rigidity, yielding vanishing of all positive Tor modules.","feed_headline":"Reflexive tensor products force both factors reflexive","feed_subtitle":"A new proof rules out a known counterexample when the second module is locally free in height one.","key_machinery":"The load-bearing object is the graph $G(R)$ whose vertices are the minimal prime ideals of $R$, with an edge between $\\mathfrak p$ and $\\mathfrak q$ exactly when $\\operatorname{height}(\\mathfrak p+\\mathfrak q) \\leq 1$. Under condition $(S_2)$, this graph is connected, and connectedness is used to show that a module which is free at every prime of height at most one has constant rank across all minimal primes. That rank, combined with the depth formula for modules of finite complete intersection dimension, lets the authors upgrade local freeness to the Serre-type condition on both modules. The other mechanical ingredient is Tor-rigidity: if $M$ is Tor-rigid, meaning that vanishing of $\\operatorname{Tor}_1^R(M,N)$ forces vanishing of $\\operatorname{Tor}_2^R(M,N)$, then vanishing of $\\operatorname{Ext}^1_R(\\operatorname{Tr}N,M)$ propagates to vanishing of all positive Tor groups through the exact sequences of section 2.2; this converts reflexivity of the tensor product into a rank statement.","core_discovery":"Theorem 1.2 asserts: let $R$ be a hypersurface ring, that is, a quotient of an unramified regular local ring by a regular element, and let $M$ and $N$ be nonzero finitely generated modules with $\\operatorname{pd}_R(M)<\\infty$ and $\\operatorname{pd}_{R_{\\mathfrak p}}(N_{\\mathfrak p})<\\infty$ for every prime $\\mathfrak p$ of height at most one. If $M \\otimes_R N$ is reflexive, then both $M$ and $N$ are reflexive. The main Theorem 3.1 is more general: for a Noetherian local ring, if $M$ is Tor-rigid, $N$ has finite complete intersection dimension, $M \\otimes_R N$ satisfies $(\\tilde S_n)$, and $\\operatorname{Tor}_i^R(M,N)$ is torsion for all sufficiently large $i$, then $\\operatorname{Tor}_i^R(M,N)=0$ for all $i \\geq 1$ and $N$ satisfies $(\\tilde S_n)$. The known counterexample with reflexive tensor product but non-reflexive $M$ fails the height-one local finiteness hypothesis, so Theorem 1.2 removes precisely that escape.","pith_inferences":["The connectedness-of-graph argument is not tied to reflexivity: the same graph should govern other statements where local freeness in codimension one is promoted to global rank, such as questions about higher syzygies of tensor products.","A natural extension is to replace the height-one condition by a higher-codimensional analogue: if $N$ is locally free outside a closed set of codimension at least $c$, one might expect reflexivity of $M \\otimes_R N$ to force both factors to satisfy higher $(\\tilde S_n)$ conditions.","The explicit presentation of the tensor product in the counterexample could be used to search for similar examples in higher dimensions where the local finiteness condition holds; the paper's method predicts that none exist over $(S_2)$ hypersurfaces."],"forward_implications":["The known counterexample cannot be adjusted to satisfy the height-one local finiteness condition; the height-one prime with infinite projective dimension is essential to the counterexample.","Under the hypotheses of Theorem 3.1, the tensor product being an $n$-th syzygy forces $N$ to be an $n$-th syzygy and kills all positive Tor groups, so the conclusion applies not only to reflexivity but to higher syzygy behavior.","In Corollary 4.4, the same hypotheses plus local freeness on $Y_1(R)$ upgrade the conclusion to both $M$ and $N$ satisfying $(\\tilde S_n)$; in the case $n=2$, both modules are reflexive.","The sharpness remark shows that the torsion hypothesis on high Tor groups alone cannot replace Tor-rigidity: a 2-Tor-rigid example with torsion high Tor has a non-reflexive conclusion.","The paper leaves open whether, with only torsion of high Tor assumed and no Tor-rigidity, a reflexive tensor product forces at least one factor to be reflexive; it notes that for domains of dimension at least two the answer is affirmative."],"supporting_citations":[{"why":"Supplies the earlier counterexample, with reflexive tensor product and $N$ reflexive but $M$ not, that Theorem 1.2 rules out; the paper computes a new presentation of its tensor product.","marker":"[10]"},{"why":"States the Second Rigidity Theorem for hypersurface rings that this paper corroborates and extends.","marker":"[16]"},{"why":"Defines the graph of minimal primes and proves its connectedness under $(S_2)$, the key step in giving $N$ a rank.","marker":"[15]"},{"why":"Supplies the depth formula for modules of finite complete intersection dimension, used to promote vanishing of Tor to reflexivity of both modules.","marker":"[1]"},{"why":"Provides the exact sequences connecting $\\operatorname{Ext}^1(\\operatorname{Tr}N,-)$ with Tor and the syzygy/transpose identifications used in Lemmas 3.2 through 3.4.","marker":"[2]"},{"why":"Establishes the equivalence between vanishing of $\\operatorname{Ext}^i(\\operatorname{Tr}N,M)$ and $\\operatorname{Tor}_i(M,N)$ when complete intersection dimension is zero, used to reduce torsion assumptions to full vanishing.","marker":"[9]"},{"why":"Proves Tor-rigidity for modules of finite projective dimension over hypersurfaces, which supplies hypothesis (i) in Theorem 1.2.","marker":"[19]"},{"why":"Used in Corollary 4.4 to conclude $M$ satisfies the Serre-type condition once $N$ has rank and the depth formula applies.","marker":"[8]"}],"fun_headline_variants":["Tor-rigid reflexive tensor forces both modules reflexive","Height-one local freeness blocks reflexive counterexample","Reflexive tensor with Tor-rigid makes both reflexive","Hypersurface rings: reflexive tensor implies reflexive factors","Second rigidity: missing height-one hypothesis is key"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument depends on the graph of minimal primes being connected under condition $(S_2)$; if that graph were disconnected, a module locally free in height one would not have to have constant rank, and the step forcing $M$ to be reflexive would break.","fun_headline_variants_meta":{"raw":{"variants":["Tor-rigid reflexive tensor forces both modules reflexive","Height-one local freeness blocks reflexive counterexample","Reflexive tensor with Tor-rigid makes both reflexive","Hypersurface rings: reflexive tensor implies reflexive factors","Second rigidity: missing height-one hypothesis is key"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000564,"raw_usage":{"total_tokens":2621,"prompt_tokens":840,"completion_tokens":1781,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":456,"completion_tokens_details":{"reasoning_tokens":1705}},"tokens_in":456,"tokens_out":1781,"duration_ms":13101,"temperature":1.0,"reasoning_tokens":1705,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:02:09.115876+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Localize the known counterexample at the height-one prime $\\mathfrak q=(x,y)$ and resolve $N_{\\mathfrak q}$: the minimal free resolution alternates multiplication by $x$ and $y$ forever, so $\\operatorname{pd}_{R_{\\mathfrak q}}(N_{\\mathfrak q})=\\infty$, verifying that this is precisely where the added hypothesis excludes the example. A single example satisfying all hypotheses of Theorem 1.2 with $M$ not reflexive would refute the claim; a computational search over small hypersurface presentations, checking the biduality map $M \\to M^{**}$ explicitly, could be used to look for one.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the earlier counterexample, with reflexive tensor product and $N$ reflexive but $M$ not, that Theorem 1.2 rules out; the paper computes a new presentation of its tensor product."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"States the Second Rigidity Theorem for hypersurface rings that this paper corroborates and extends."},{"cited_title":"(1994), no","cited_arxiv_id":null,"evidence_quote":"Defines the graph of minimal primes and proves its connectedness under $(S_2)$, the key step in giving $N$ a rank."},{"cited_title":"Algebra 26 (1998), no","cited_arxiv_id":null,"evidence_quote":"Supplies the depth formula for modules of finite complete intersection dimension, used to promote vanishing of Tor to reflexivity of both modules."},{"cited_title":"94, American Mathematical Society, Providence, R.I., 1 969","cited_arxiv_id":null,"evidence_quote":"Provides the exact sequences connecting $\\operatorname{Ext}^1(\\operatorname{Tr}N,-)$ with Tor and the syzygy/transpose identifications used in Lemmas 3.2 through 3.4."},{"cited_title":"5, 1670–1684","cited_arxiv_id":null,"evidence_quote":"Establishes the equivalence between vanishing of $\\operatorname{Ext}^i(\\operatorname{Tr}N,M)$ and $\\operatorname{Tor}_i(M,N)$ when complete intersection dimension is zero, used to reduce torsion assumptions to full vanishing."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Proves Tor-rigidity for modules of finite projective dimension over hypersurfaces, which supplies hypothesis (i) in Theorem 1.2."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Used in Corollary 4.4 to conclude $M$ satisfies the Serre-type condition once $N$ has rank and the depth formula applies."}],"review_version":1}