{"id":"61a2629b-734a-40fc-9add-d6870977fbdd","arxiv_id":"2412.12860","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Nearly Gorenstein Stanley-Reisner rings of dimension at least three are Gorenstein, and canonical traces of punctured-Gorenstein Stanley-Reisner rings are exactly the ring, the maximal ideal, or its square.","lead":"The paper proves that nearly Gorenstein Stanley-Reisner rings of dimension at least 3 are actually Gorenstein, and it classifies the canonical trace for the broader class of rings that are Gorenstein away from the maximal ideal. The result closes an open question about when a weak Gorenstein condition forces the strong one in this family.","discovery_kind":"unification","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central trace classification rests on Lemma 3.2(2), an injectivity claim imported from Gräbe 1984; if that injectivity fails for a nonorientable homology manifold, Propositions 3.4 and 3.9, and hence Theorem A, would not follow.","rationale":"I read the paper as a serious and largely coherent classification of the canonical trace for Cohen-Macaulay Stanley-Reisner rings that are Gorenstein on the punctured spectrum. The main structural steps are plausible: Proposition 2.8 gives a useful dichotomy between cone points and homology manifolds; Propositions 3.4 and 3.9 use Gräbe's description of the canonical module to bound and then realize the trace; and the low-dimensional cases are delegated to the first author's prior work. I did not find an internal contradiction or a clear counterexample in the text. The most load-bearing point is indeed Lemma 3.2(2), exactly as the reader identified: the proof of Proposition 3.4 needs to know that the relative homology groups are at most one-dimensional and that all inclusions between nonzero ones induce isomorphisms, and Proposition 3.9 needs the same injectivity to compare x_i·1_j with x_j·1_i. The paper cites Gräbe's Hauptlemma 3.2 without proof. This is a legitimate external dependency rather than an evident error, but it is not machine-checked and is not re-derived. A concrete computational check on the smallest nonorientable homology manifold, RP^2, would settle whether the injectivity holds in the non-manifold and nonorientable regimes that the propositions require. Because the dependency is real but not demonstrably false, the appropriate verdict remains CONDITIONAL, matching the reader's assessment. I have not seen a reason to move to ACCEPT, REJECT, or UNVERDICTED: the concern is a verification gap, not a disproof. My agreement with the reader is full on the weakest assumption, since the same imported injectivity lemma was already singled out as the fragility point.","tokens_in":16,"tokens_out":51478,"duration_ms":1034464,"concrete_test":"Implement Gräbe's chain description for the minimal 6-vertex triangulation of RP^2 over Q, a 2-dimensional non-k-orientable k-homology manifold. For every inclusion τ ⊆ σ of faces, compute H_2(∆, cost(τ); Q) and the induced map ι_* by direct linear algebra. Verify that whenever the domain is nonzero, ι_* is an isomorphism onto H_2(∆, cost(σ); Q). If some map is zero or non-injective, Lemma 3.2(2) fails in exactly the class used by Propositions 3.9 and 3.4, and Theorem A would require a different proof. Independently re-derive the cited Hauptlemma 3.2 from the long exact sequence of the triple; if it cannot be reproduced without extra connectivity hypotheses, treat the lemma as unverified.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Lemma 3.2(2) is the pivot of the trace computations. It asserts that for a normal pseudomanifold ∆, whenever H_d(∆, cost(τ); k) ≠ 0, the natural map ι_* : H_d(∆, cost(τ); k) → H_d(∆, cost(σ); k) is an isomorphism for every τ ⊆ σ ∈ ∆. The proof delegates the injectivity half to Gräbe's Hauptlemma 3.2 in [14], a 1984 paper in German, without stating or proving the precise hypotheses in the present text. Propositions 3.4 and 3.9 rely on this to control which degree components of ω_R occur and to show that multiplication by x_i identifies components; without injectivity, the relation x_i·1_j = x_j·1_i can fail, and the conclusion tr(ω_R) = M_R^2 for nonorientable homology manifolds is unsupported. Since Theorem A(X), Theorem A(Z), and Corollary 3.5 all pass through these propositions, the central classification is only as secure as this imported lemma. The paper gives no independent derivation, and the reader's own verification cannot access [14] directly.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the canonical trace of the canonical module of Stanley–Reisner rings that are Gorenstein on the punctured spectrum. The main theorem (Theorem A) states that for a Cohen–Macaulay Stanley–Reisner ring R over a field k, R is Gorenstein on the punctured spectrum if and only if tr(ω_R) equals M_R^i for i∈{0,1,2}; moreover, the intermediate case tr(ω_R)=M_R occurs exactly for two low-dimensional families (disjoint unions of at least three vertices or paths of length at least three), and the case tr(ω_R)=M_R^2 occurs exactly for non-k-orientable k-homology manifolds. A corollary resolves a question from [28]: every nearly Gorenstein Stanley–Reisner ring of dimension at least three is Gorenstein. Theorem B shows that every Stanley–Reisner ring Gorenstein on the punctured spectrum is level, and it characterizes when such a ring is simultaneously almost Gorenstein. The proofs use Gräbe's description of canonical modules of Stanley–Reisner rings, a lemma on relative homology maps, and an analysis of the local structure of pseudomanifolds.","tokens_in":13901,"tokens_out":28561,"duration_ms":260033,"significance":"If correct, the paper gives a complete and clean trichotomy for the canonical trace of Cohen–Macaulay Stanley–Reisner rings that are Gorenstein on the punctured spectrum, and it settles the nearly-Gorenstein-vs-Gorenstein question for this class. The criterion is concrete and falsifiable, and the connection to non-orientable homology manifolds is a pleasing combinatorial characterization. The paper also strengthens the link between levelness, nearly Gorensteinness, and almost Gorensteinness for monomial rings. The authors use established structural results (Gräbe's theorem) as a black box, which is appropriate, though it places a burden on the reader to verify the exact hypotheses of the imported lemma.","major_comments":[{"comment":"The proof begins with the assertion that a connected simplicial complex whose Stanley–Reisner ring is Gorenstein on the punctured spectrum is necessarily normal. This assertion is load-bearing because normality is used immediately to conclude that the complex is strongly connected, which is then used in the distance argument. No proof or reference is provided for this implication. Since normality of ∆ is equivalent to k[∆] satisfying Serre's condition (S2), the authors should either prove that Gorensteinness at every non-maximal graded prime forces (S2) in this setting or give a precise reference for this fact.","section":"Section 2, Proposition 2.8"},{"comment":"Part (2) of Lemma 3.2 is the pivotal injectivity statement for the relative homology maps ι_*. The proof delegates the injectivity to Gräbe's Hauptlemma 3.2 in [14] without stating the exact hypotheses of that lemma or verifying that a normal pseudomanifold satisfies them. Since Propositions 3.4, 3.9, Corollary 3.10, and ultimately Theorem A all rely on this injectivity, the authors should state the precise result used, confirm that every normal pseudomanifold is a connected quasi-manifold in the sense of [14], and explain why injectivity plus the one-dimensionality from part (1) yields the asserted isomorphism.","section":"Section 3, Lemma 3.2"},{"comment":"In the proof, the sentence 'Since the annihilator of xi must be contained in that of 1_a' is not justified. From φ(1_a)=x_i one obtains immediately the reverse inclusion, Ann(1_a) ⊆ Ann(x_i). The subsequent conclusion s(a)={i} is crucial for the argument that tr(ω_R) ⊆ M_R^2. Please give a detailed derivation of s(a)={i} from the graded homomorphism φ and the Gräbe structure theorem. In particular, the proof should address the possibility that variables are zero divisors in a Stanley–Reisner ring; the later cancellation step 'φ(1_b)=1' also needs a justification that does not silently assume x_i is regular.","section":"Section 3, Proposition 3.4"}],"minor_comments":[{"comment":"There are several typos: 'Propoition 3.9' in the Section 3 heading, 'non-oriantable' in Corollary 3.10, and 'n /greaterorequalslant3' in the abstract due to a LaTeX rendering issue.","section":"Throughout"},{"comment":"In the definition of δ, the notation 'σ_i ∈ Fd(∆)' should be 'σ_i ∈ F(∆)'.","section":"Proposition 2.8"},{"comment":"The step in (1)⇒(4) invoking [27, Theorem 5.1] to conclude that the socle degree s(R) equals 1, and the subsequent equality between codimension and Cohen–Macaulay type, is quite compressed. Please expand this argument so that a reader can follow the logic.","section":"Theorem B, proof of (Y)"}],"recommendation":"major_revision","confidential_remarks":"The central results are likely correct and the paper is a solid contribution, but the proof has a few load-bearing points where the text is either too terse or relies on an imported lemma whose exact hypotheses are not stated. In particular, the normality assertion in Proposition 2.8 and the injectivity in Lemma 3.2(2) deserve careful treatment. The authors should also revisit the annihilator argument in Proposition 3.4, which as written seems to assert the wrong inclusion. These issues are fixable within the paper's scope, so I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nYou should know about this paper: it settles an open question from Miyashita's earlier work, showing that nearly Gorenstein Stanley–Reisner rings of dimension at least 3 are actually Gorenstein, and it classifies the possible canonical traces for Cohen–Macaulay Stanley–Reisner rings that are Gorenstein on the punctured spectrum. The classification is clean: the trace is the whole ring, the maximal ideal, or its square, and the M_R^2 case occurs exactly for non-k-orientable k-homology manifolds.\n\nWhat's genuinely good: the main theorem is new, the proof is structured and mostly transparent, and the use of Gräbe's canonical module description is appropriate. Proposition 2.8, the cone-point dichotomy, is a nice combinatorial argument. The paper also connects the result to level rings and almost Gorenstein rings, giving Theorem B, which is a reasonable bonus. The authors are honest about dependencies, including citing the first author's prior low-dimensional classification.\n\nSoft spots, in proportion: the load-bearing injectivity lemma (Lemma 3.2(2)) is imported from Gräbe's 1984 paper without restating the Hauptlemma. That is a bit annoying, and a referee should verify that the stated hypotheses (connected quasi-manifold) match Gräbe's result. But I do not think this is a fatal gap: the paper gives the exact condition, and the rest of the argument only needs injectivity to force isomorphisms between one-dimensional vector spaces. The stress-test worry that the classification would collapse if injectivity fails is true but hypothetical; nothing in the text suggests the lemma is misstated. A second soft spot: a few steps in Propositions 3.4 and 3.9 are compressed, particularly the extension from adjacent vertices to all vertices. They can be filled, but they'll take an expert reader some time. The proof of Proposition 2.8 is also terse in places.\n\nOverall, the central argument holds up. The paper deserves a serious referee and, likely, publication in a good commutative algebra journal. I would take it to reading group and would cite it.\n\nRecommendation: send to peer review.","headline":"A solid, publishable classification of canonical traces for CM Stanley-Reisner rings; the main theorem answers a real open question, and the imported injectivity lemma is a standard citation rather than a fatal gap.","tokens_in":14465,"tokens_out":3683,"would_cite":true,"duration_ms":30875,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["13H10","13A02","05E40"],"pacs":[],"model":"deepseek-v4-flash","headline":"A Cohen-Macaulay Stanley-Reisner ring is Gorenstein on the punctured spectrum exactly when the canonical trace is the ring, its maximal ideal, or the square of that ideal.","keywords":["Stanley-Reisner rings","canonical trace","nearly Gorenstein","Gorenstein on the punctured spectrum","level rings","homology manifolds","simplicial complexes","Cohen-Macaulay rings"],"falsifier":"The classification would be falsified by a Cohen-Macaulay Stanley-Reisner ring of dimension at least three that is nearly Gorenstein but not Gorenstein, or by a non-$k$-orientable $k$-homology manifold whose canonical trace is not $M_R^2$.","tokens_in":13462,"feed_emoji":"🔺","tokens_out":12181,"duration_ms":94658,"temperature":0.7,"pith_summary":"The paper aims to locate exactly where non-Gorenstein behavior can survive in a Stanley-Reisner ring. For a Cohen-Macaulay ring $R=k[\\Delta]$, it proves that being Gorenstein on the punctured spectrum is equivalent to the trace of the canonical module $\\operatorname{tr}(\\omega_R)$ being one of three ideals: $R$, the irrelevant maximal ideal $M_R$, or $M_R^2$. This trichotomy resolves an open question: every nearly Gorenstein Stanley-Reisner ring of dimension at least three is actually Gorenstein. The only non-Gorenstein nearly Gorenstein examples lie in dimension at most one, as a disjoint union of at least three vertices or a path of length at least three, and the $M_R^2$ case occurs exactly for non-$k$-orientable $k$-homology manifolds. Since $\\operatorname{tr}(\\omega_R)$ measures the non-Gorenstein locus, the result is a complete combinatorial description of that locus for this class.","feed_headline":"Nearly Gorenstein Stanley-Reisner rings are Gorenstein in dimension 3+","feed_subtitle":"Their canonical trace takes only three values, pinning down the non-Gorenstein locus.","key_machinery":"The load-bearing object is the canonical trace $\\operatorname{tr}(\\omega_R)$, the sum of the images of all graded module maps from the canonical module $\\omega_R$ into $R$; its radical detects the non-Gorenstein locus. The paper controls it through the structure theorem for canonical modules of Stanley-Reisner rings, which realizes $\\omega_R$ as a direct sum, indexed by faces, of relative homology groups $H_d(\\Delta,\\operatorname{cost}_\\Delta(s(a)))$, with multiplication by a variable acting by maps $\\iota_*$ between such groups. An imported lemma says those maps are injective for connected quasi-manifolds, and since the relevant homology groups have dimension at most one, the trace is forced to be generated in low degree. The final piece is the dichotomy of Proposition 2.8, which confines a connected complex of dimension at least two to having a cone point or being a $k$-homology manifold, so the trace cannot be anything beyond $R$, $M_R$, or $M_R^2$.","core_discovery":"The central claim is Theorem A of the paper: if $R=k[\\Delta]$ is Cohen-Macaulay, then $R$ is Gorenstein on the punctured spectrum if and only if $\\operatorname{tr}(\\omega_R)=M_R^i$ for some $i\\in\\{0,1,2\\}$. The proof is driven by a topological dichotomy: a connected complex of dimension at least two whose Stanley-Reisner ring is Gorenstein on the punctured spectrum either has a cone point or is a $k$-homology manifold. In the homology-manifold case, the structure theorem for canonical modules shows the trace is $R$ when $\\Delta$ is $k$-orientable and $M_R^2$ when it is not. In dimension at least three this forces nearly Gorenstein rings to be Gorenstein, and the remaining $M_R$ case is confined to the two low-dimensional families listed in the theorem.","pith_inferences":["Inference: the same three-value trace dichotomy may hold for other standard graded Cohen-Macaulay rings whose canonical modules come with a combinatorial or homological description; the missing ingredient would be an analogue of the injectivity lemma.","Inference: Proposition 2.8 offers a purely combinatorial test for whether a connected complex whose Stanley-Reisner ring is Gorenstein on the punctured spectrum is a homology manifold: look for a cone point. The test does not require the Cohen-Macaulay assumption in that statement.","Inference: the levelness result for this class could be tested against other families of combinatorial rings, such as normal affine semigroup rings or Ehrhart rings, where the canonical trace is studied by similar methods."],"forward_implications":["A Cohen-Macaulay Stanley-Reisner ring is Gorenstein on the punctured spectrum exactly when $\\operatorname{tr}(\\omega_R)$ is $R$, $M_R$, or $M_R^2$.","Every nearly Gorenstein Stanley-Reisner ring of dimension at least three is Gorenstein, so the question of whether nearly Gorenstein equals Gorenstein is answered affirmatively above dimension two.","The non-Gorenstein nearly Gorenstein Stanley-Reisner rings are completely classified: disjoint unions of at least three vertices and paths of length at least three.","The case $\\operatorname{tr}(\\omega_R)=M_R^2$ occurs exactly for non-$k$-orientable $k$-homology manifolds; in characteristic 2 this case disappears, so Gorenstein on the punctured spectrum forces nearly Gorensteinness.","Every Stanley-Reisner ring that is Gorenstein on the punctured spectrum is level, and in the non-Gorenstein case, nearly Gorenstein plus almost Gorenstein is equivalent to nearly Gorenstein alone."],"supporting_citations":[{"why":"Supplies the structure theorem identifying the canonical module of a Stanley-Reisner ring with a direct sum of relative homology groups.","marker":"[13]"},{"why":"Provides the injectivity of the multiplication maps between relative homology groups for connected quasi-manifolds, used to control the trace.","marker":"[14]"},{"why":"Defines nearly Gorenstein rings and gives the basic identities for the canonical trace, including the square-root criterion for the punctured spectrum.","marker":"[17]"},{"why":"Establishes the low-dimensional classification of nearly Gorenstein Stanley-Reisner rings and poses the question answered here.","marker":"[28]"},{"why":"Supplies the obstruction used to rule out the possibility that the trace is the maximal ideal in the homology-manifold cases.","marker":"[11]"},{"why":"Provides standard Cohen-Macaulay and Gorenstein facts for Stanley-Reisner rings and homology spheres used throughout.","marker":"[3]"},{"why":"Gives the definition of level rings and foundational canonical-module facts needed for Theorem B.","marker":"[34]"}],"fun_headline_variants":["Dim 3+ forces nearly Gorenstein SR rings to be Gorenstein","In dim 3+, nearly Gorenstein SR rings are Gorenstein","Canonical trace takes three values, pinning Gorenstein locus","Three trace values characterize punctured Gorenstein SR rings"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof rests on an imported structural fact: for a normal pseudomanifold, the multiplication maps between relative homology groups are injective, and if that failed the trace computations would not force the three-value classification.","fun_headline_variants_meta":{"raw":{"variants":["Dim 3+ forces nearly Gorenstein SR rings to be Gorenstein","In dim 3+, nearly Gorenstein SR rings are Gorenstein","Canonical trace takes three values, pinning Gorenstein locus","Three trace values characterize punctured Gorenstein SR rings"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001597,"raw_usage":{"total_tokens":6307,"prompt_tokens":827,"completion_tokens":5480,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":443,"completion_tokens_details":{"reasoning_tokens":5401}},"tokens_in":443,"tokens_out":5480,"duration_ms":33258,"temperature":1.0,"reasoning_tokens":5401,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T13:40:56.897355+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"The classification would be falsified by a Cohen-Macaulay Stanley-Reisner ring of dimension at least three that is nearly Gorenstein but not Gorenstein, or by a non-$k$-orientable $k$-homology manifold whose canonical trace is not $M_R^2$.","supporting_citations":[{"cited_title":"Gr¨ abe,The canonical module of a Stanley–Reisner ring , Journal of Algebra 86 (1984), no","cited_arxiv_id":null,"evidence_quote":"Supplies the structure theorem identifying the canonical module of a Stanley-Reisner ring with a direct sum of relative homology groups."},{"cited_title":"Gr¨ abe, ¨Uber den Stanley–Reisner-Ring von Quasimannigfaltigkeit en, Math","cited_arxiv_id":null,"evidence_quote":"Provides the injectivity of the multiplication maps between relative homology groups for connected quasi-manifolds, used to control the trace."},{"cited_title":"Herzog, T","cited_arxiv_id":null,"evidence_quote":"Defines nearly Gorenstein rings and gives the basic identities for the canonical trace, including the square-root criterion for the punctured spectrum."},{"cited_title":"Gasanova, J","cited_arxiv_id":null,"evidence_quote":"Supplies the obstruction used to rule out the possibility that the trace is the maximal ideal in the homology-manifold cases."},{"cited_title":"Bruns and J","cited_arxiv_id":null,"evidence_quote":"Provides standard Cohen-Macaulay and Gorenstein facts for Stanley-Reisner rings and homology spheres used throughout."},{"cited_title":"P Stanley, Combinatorics and Commutative Algebra , V ol","cited_arxiv_id":null,"evidence_quote":"Gives the definition of level rings and foundational canonical-module facts needed for Theorem B."}],"review_version":1}