{"id":"b4f2e3d1-56f0-4cdd-a1df-4b7b5b56d35b","arxiv_id":"2411.14963","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Generalized cluster algebras that are Krull domains can have any finitely generated abelian group as their class group, including finite ones with torsion.","lead":"This paper proves that every finitely generated abelian group can appear as the class group of a generalized cluster algebra, a class of rings related to cluster algebras. It also shows generalized cluster algebras are finite-factorization domains and analyzes factorization in Laurent phenomenon algebras.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 4.14's proof is internally inconsistent: the displayed B formula, the exchange polynomials, and the map φ use incompatible indices, so the proof as written does not realize the claimed class group.","rationale":"The central strategy — realize Z^{m-1} by m distinct primes from x_N^m+1 with one relation summing them, and torsion by (x+1)^{n_i} — is sound and consistent with Example 4.15. But the written proof has two independent index mismatches. First, the B formula produces entries that change the exchange polynomial for x1, so the asserted list of exchange polynomials cannot be derived from the seed defined there. Second, the φ formula's kernel does not contain a1 and has too large a free image, so the isomorphism statement in the proof fails even for the paper's own example. The reader's CONDITIONAL verdict is appropriate: these are fixable by adopting Example 4.15's index convention, and the broader framework (FF-domains, Krull/class-group criteria) is plausible. I would not recommend rejection because the intended construction is explicit and likely correct, but the theorem needs a corrected B, corrected φ, and an explicit characteristic-zero (or char ∤ m) statement to be rigorous. The reader's characteristic concern is noted but is less decisive if the global characteristic-zero convention from Section 2 is retained.","tokens_in":22045,"tokens_out":21322,"duration_ms":199230,"concrete_test":"Take the G=Z×Z/3 case (m=2, k=1, n1=3, N=4) and run both variants. Using Theorem 4.14's displayed B, compute f1: the rule gives b31=3, so f1=1+x_3^3 x_4^2, not x_4^2+1; using its displayed φ, φ(1,1,0,0,0)=(0,1,-1,0,0)≠0, so a1 is absent from the kernel. Using Example 4.15's indexing (b32=3, φ(u)=(u1-u2, u2-u1, [u3]_3, 0, 0)), f1=x_4^2+1 and ker φ=⟨a1,a2,a3,a4⟩, so the class group computation goes through. This settles that the theorem's text needs the example's index convention.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Most load-bearing concern: internal indexing inconsistency in Theorem 4.14's construction. The defining formula for B (Section 4.2) gives b_{ij}=n_j for N/2<i≤N-1, j=N-i. For the paper's own Example 4.15 (m=2, k=1, n1=3, N=4), this yields b_{31}=3, so f1=1+x_3^3 x_4^2; the example instead uses b_{32}=3 and computes f1=x_4^2+1. Hence the matrix from which the exchange polynomials are computed is not the matrix defined in the theorem, and the appeals to Corollary 4.10 (m distinct primes from f1, and the resulting relations) do not follow from the displayed seed. The homomorphism φ has a similar mismatch: the printed formula has m+1 difference coordinates summing to zero, so its free image has rank m, while G has free rank m-1; moreover φ(a1) is not zero for the theorem's a1, so a1 is not in the claimed kernel. Example 4.15 repairs both issues, but the theorem itself does not. These are repairable indexing errors, not a flawed strategy, but they mean the proof as written does not establish the central claim. The characteristic concern is secondary because Section 2 fixes R to be Z or a field of characteristic 0; under that standing assumption x^m+1 is squarefree.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies divisor class groups and factorization properties of generalized cluster algebras. Its main results are: (i) generalized cluster algebras and generalized upper cluster algebras are FF-domains and their cluster variables are strong atoms (Propositions 2.12 and 2.13); (ii) a structural theorem computing the class group of a Krull generalized cluster algebra as Z^r modulo the exponent vectors of the cluster variables (Theorem 4.2, following GELS19); (iii) the realization theorem claiming that every finitely generated abelian group occurs as the class group of an acyclic coprime generalized cluster algebra, with exactly |R| height-one primes in every class (Theorem 4.14); and (iv) analogous factoriality and FF-domain statements for LP algebras, with examples distinguishing LP algebras from cluster algebras. The paper is largely an application of the GELS19 class-group formula and the BCDX20 upper/lower bound theorem, supplemented by explicit worked examples.","tokens_in":22295,"tokens_out":28449,"duration_ms":262988,"significance":"If Theorem 4.14 is correct, it establishes a striking new phenomenon: unlike classical cluster algebras, whose Krull class groups are always free abelian, generalized cluster algebras can realize arbitrary finitely generated abelian groups, including finite groups with nontrivial torsion. This has consequences for factorization theory, since a finite class group can make arithmetic invariants finite. The paper's examples (4.11, 4.13, 4.15) are explicit and useful, and the overall strategy is plausible: the class-group computation is a direct application of established external theorems, and the new construction is demonstrated in the worked example. However, the proof of Theorem 4.14 as written contains systematic indexing inconsistencies in the definition of the exchange matrix, in the exchange polynomials, and in the homomorphism used to compute the class group. These are load-bearing but appear repairable.","major_comments":[{"comment":"The displayed formula for B is not the matrix used to compute the exchange polynomials. For the paper's own Example 4.15 (m=2, k=1, n1=3, N=4), the formula gives b_{3,1}=n_1=3 and b_{4,1}=m=2, so column 2 is zero; this matrix has rank at most 3 and is not skew-symmetrizable. The example instead uses b_{3,2}=3, and the claimed exchange polynomials f_1=x_4^2+1 and f_2=(x_3+1)^3 require exactly that entry. Consequently the appeals to Corollary 4.10, Theorem 3.9, and Theorem 4.2 do not apply to the seed as formally defined. The lower-left block should be corrected (presumably b_{k+1+r,k+2-r}=n_r for 1≤r≤k and b_{N,1}=m).","section":"Section 4.2, Theorem 4.14 (definition of B)"},{"comment":"The homomorphism phi is written with m+1 free coordinates u_1,...,u_{m+1} and m difference coordinates (u_1-u_2, ..., u_{m+1}-u_1), whose image has rank m, but G has free rank m-1. The stated kernel forces u_1=...=u_{m+1}; however, the subgroup generated by a_1,...,a_N only forces u_1=...=u_m and leaves n_1 beta_1 in coordinate m+1, so the two groups are not equal when n_1>1. Example 4.15 works only because it uses two coordinates u_1,u_2 for the free part Z, not three as in the theorem. The proof therefore does not establish C(A) ≅ G for the construction as written.","section":"Section 4.2, Theorem 4.14 (definition of phi)"},{"comment":"The proof sets d_i=n_i for 2≤i≤k+1 and writes the exchange polynomials as (x_{N-i+1}+1)^{n_i} for the same range, but n_i is only defined for 1≤i≤k; these should be n_{i-1}. The same misindexing affects the a_i vectors in the class-group computation. This is repairable but must be corrected for the theorem to be internally consistent.","section":"Section 4.2, Theorem 4.14 (indices of n_i)"},{"comment":"The local assumption 'R is an algebraically closed field' does not specify the characteristic. The proof needs f_1=x_N^m+1 to factor into m distinct linear factors and (x+1)^{n_i} to be squarefree; this fails for algebraically closed fields of characteristic p dividing m or n_i. If the standing characteristic-zero assumption from Section 2 is still in force, this should be stated explicitly; as written, Theorem 4.14 is false for some positive-characteristic algebraically closed fields.","section":"Section 4.2, opening of Section 4.2 and Theorem 4.14"}],"minor_comments":[{"comment":"The text says 'Then by Corollary 4.2 we have'; the statement used is Theorem 4.2, not Corollary 4.2.","section":"Example 4.11"},{"comment":"The proof states L_{x_i}=U[x_i^{-1}], which is not literally correct with the paper's notation for the mutated Laurent ring; the desired conclusion can be obtained directly from U⊆L_{x_i} and x_i∈r_jL_{x_i}, so please rephrase the localization argument.","section":"Corollary 4.10, proof"},{"comment":"The prime ideal p for x_1 is written as (x_2+1)L_{x_2}∩A, but to contain x_1 it should be obtained from the mutated Laurent ring L_{x_1} (as in Example 4.11). This appears to be a typo in the localization index.","section":"Example 4.13(2)"},{"comment":"The block labels under the displayed matrix ('k z}|{ k + 1 z}|{') are unclear and should be replaced with a precise description of the nonzero entries; this is especially important because the current matrix is inconsistent with the example.","section":"Theorem 4.14, displayed matrix"},{"comment":"There is a typo: 'exchange polymomials' should be 'exchange polynomials'.","section":"Example 4.15"}],"recommendation":"major_revision","confidential_remarks":"The main theorem is a nice potential result, but Theorem 4.14 as written has several indexing inconsistencies that are load-bearing; the accompanying Example 4.15 shows that the intended construction is repairable. There is no circularity concern: the arguments rely on published external results. The characteristic hypothesis should be clarified to avoid a false statement over positive-characteristic algebraically closed fields."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's the thing you should know: the headline result—every finitely generated abelian group is the class group of a generalized cluster algebra—is almost certainly correct, and it's a good result. But Theorem 4.14 as printed does not prove it. The B-matrix in the proof puts the n_i in the wrong column, so the exchange polynomials computed from the displayed matrix do not match the ones used later. The homomorphism φ also has one too many free coordinates: the printed map lands in a free part of rank m, while G has free rank m−1, so the kernel is not the subgroup generated by the a_i. Example 4.15 shows what the construction should be, and the stress-test note is right that the example and the theorem are inconsistent. These look like repairable typos, not a broken strategy.\n\nWhat's genuinely new: torsion class groups for generalized cluster algebras, a sharp contrast with classical cluster algebras (whose Krull class groups are free abelian), and a clean way to realize arbitrary finitely generated abelian groups using acyclic coprime seeds and the GELS19 class-group formula. The FF-domain and strong-atom results are routine extensions but are handled consistently. The LP algebra section is fine and doesn't affect the main point.\n\nTwo issues need fixing. First, the indexing in Theorem 4.14: the B-matrix entries and the definition of φ have to match the example. Concretely, the n_i should sit in the column of the cluster variable whose exchange polynomial they are meant to produce, and φ should use m coordinates for the free part, not m+1. Second, the construction assumes an algebraically closed field but needs characteristic not dividing m for x^m+1 to split into m distinct irreducible factors. Since the theorem only asserts existence, specifying an algebraically closed field of characteristic zero (e.g., C) removes the problem. Both are minor to fix but load-bearing as written.\n\nThe paper leans on external theorems—GELS19 and BCDX20—but that's normal, and those cited results are published. Self-citation to Pom25 is also legitimate there.\n\nWho this is for: people working on cluster algebras, factorization theory, or Krull domains. The realization theorem is worth knowing once the proof is corrected. It deserves a serious referee; the referee should demand the indexing corrections and a clean characteristic statement, but the core idea is sound.","headline":"The realization theorem is very likely true, but Theorem 4.14's proof as written has load-bearing indexing errors and a missing characteristic hypothesis; both are easily repairable.","tokens_in":22814,"tokens_out":8463,"would_cite":true,"duration_ms":72441,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["13F60","13F05","13F15"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that every finitely generated abelian group is isomorphic to the class group of some acyclic and coprime generalized cluster algebra, and gives a criterion for when such algebras are unique factorization domains.","keywords":["generalized cluster algebras","class groups","Krull domains","finite factorization domains","strong atoms","torsion in class groups","Laurent phenomenon algebras","cluster algebras"],"falsifier":"Run the paper's construction for $G=\\mathbb{Z}/p\\mathbb{Z}$ over an algebraically closed field of characteristic $p$: the exchange polynomial $(x_2+1)^p=x_2^p+1$ has only one irreducible factor, so Corollary 4.10 gives a single height-1 prime ideal over $x_1$ instead of $p$ distinct ones, and the presentation of Theorem 4.2 yields a free abelian class group rather than $\\mathbb{Z}/p\\mathbb{Z}$.","tokens_in":21786,"feed_emoji":"🧮","tokens_out":12309,"duration_ms":105972,"temperature":0.7,"pith_summary":"Every commutative domain has a class group that measures how far it is from having unique factorization, and for Krull domains this is a well-defined abelian group. This paper proves that generalized cluster algebras, a variant of cluster algebras with multinomial exchange relations, can realize every finitely generated abelian group as their class group. That is a sharp break from classical cluster algebras, whose class groups are always free abelian and therefore cannot have torsion or finite class groups. The paper also shows that generalized cluster algebras are finite-factorization domains, that their cluster variables are strong atoms, and that a generalized cluster algebra that is a Krull domain is a UFD exactly when its exchange polynomials are irreducible. A final section proves analogous factorization and factoriality results for Laurent phenomenon algebras, whose Krull class groups, like those of ordinary cluster algebras, are always free abelian.","feed_headline":"Every finitely generated abelian group is a class group","feed_subtitle":"Unlike classical cluster algebras, generalized ones can have torsion and finite class groups.","key_machinery":"The load-bearing mechanism is the class-group presentation theorem (Theorems 4.1 and 4.2), which for a Krull domain $A$ that becomes a factorial Laurent polynomial ring after inverting cluster variables $x_1,\\dots,x_n$ says that $C(A)\\cong \\mathbb{Z}^r/\\langle a_1,\\dots,a_n\\rangle$, where $a_i=(a_{i1},\\dots,a_{ir})$ is the exponent vector in the factorization $x_iA=\\mathfrak{p}_1^{a_{i1}}\\cdots \\mathfrak{p}_r^{a_{ir}}$ into height-1 prime ideals. For an acyclic and coprime generalized seed, the starfish lemma and the equality $A=U=S_{\\mathbf{x}}$ (Theorem 3.9) place the algebra exactly in this situation. To realize a prescribed group $G\\cong \\mathbb{Z}^{m-1}\\oplus \\mathbb{Z}/n_1\\mathbb{Z}\\oplus\\cdots\\oplus \\mathbb{Z}/n_k\\mathbb{Z}$, the proof assembles a rank $N=2k+2$ exchange matrix whose nonzero entries are $m$, the $n_i$, and $-1$, and whose exchange polynomials include $x_N^m+1$ and $(x_{N-i+1}+1)^{n_i}$; the repeated binomial factors create torsion relations, while the distinct irreducible factors of $x_N^m+1$ create the free part. This is the mechanism that lets every finitely generated abelian group appear as a class group.","core_discovery":"The central claim, Theorem 4.14, states that for any finitely generated abelian group $G$ there exists an acyclic and coprime generalized cluster algebra $A$ over an algebraically closed field $R$ such that $A$ is a Krull domain, its class group $C(A)$ is isomorphic to $G$, and every class of $C(A)$ contains exactly $|R|$ prime divisors. The proof is constructive: from the invariant factors of $G$ it builds an explicit exchange matrix $B$ and a set of strings so that the exchange polynomials take the form $f_1=x_N^m+1$ and $f_i=(x_{N-i+1}+1)^{n_i}$ (with the remaining $f_i$ binomials), and it then presents $C(A)$ as $\\mathbb{Z}^{m+2k+1}$ modulo the lattice generated by the divisor-exponent vectors of the cluster variables. The same computation yields the paper's general structural result: if a generalized cluster algebra or generalized upper cluster algebra is a Krull domain, then $C(A) \\cong \\mathbb{Z}^r/\\langle a_1,\\dots,a_n\\rangle$, where the $a_i$ record the exponents of the height-1 prime ideals containing the cluster variables, and each class of $C(A)$ contains exactly $|R|$ prime divisors. Torsion arises exactly from exchange polynomials with repeated irreducible factors, such as $(x+1)^n$, which cannot occur for classical cluster algebras. The paper further proves that generalized cluster algebras are FF-domains and that their cluster variables are strong atoms, and that Krull Laurent phenomenon algebras have free abelian class groups.","pith_inferences":["A natural testable conjecture is that torsion in the class group of a generalized cluster algebra is determined entirely by the multiplicities of irreducible factors of exchange polynomials, so the realization theorem could be framed as a statement about which multisets of irreducible polynomials occur.","The construction is stated over algebraically closed fields; over other fields the same seed realizes a different class group, so one can extend the theorem by replacing $x_N^m+1$ with any degree-$m$ polynomial with $m$ distinct irreducible factors, provided the characteristic does not obstruct that splitting.","Because Claborn's theorem realizes every abelian group as the class group of a Dedekind domain, this result shows that being a generalized cluster algebra imposes no additional restriction on finitely generated class groups, tying cluster theory into the classical realization program.","One could attempt to realize the same groups with smaller rank by exploiting that a single exchange polynomial with several distinct irreducible factors contributes several generators to the free part, potentially lowering the ambient dimension $N$ of the seed."],"forward_implications":["Every finite abelian group, and every mix of a free part with finite torsion, occurs as the class group of a Krull generalized cluster algebra, so factorization in these algebras can be governed by a finite class group.","Classical cluster algebras have only free abelian class groups, so the generalized setting is strictly richer: torsion and finite class groups are genuinely new phenomena driven by repeated factors in exchange polynomials.","For a Krull generalized cluster algebra with $A=U=S_{\\mathbf{x}}$, the algebra is a unique factorization domain if and only if all exchange polynomials are irreducible, giving a direct factoriality criterion.","Each class of the class group contains exactly $|R|$ height-1 prime divisors, so when the ground field is algebraically closed (hence infinite) prime divisors are distributed evenly among all classes.","Laurent phenomenon algebras that are Krull domains have free abelian class groups of rank $r-n$, so the torsion phenomenon does not occur in the LP framework."],"supporting_citations":[{"why":"Supplies the class-group presentation theorem (Theorems 3.1 and 3.2) that turns divisor-exponent vectors of cluster variables into a presentation of $C(A)$, and shows classical cluster algebras have free class groups.","marker":"[GELS19]"},{"why":"Proves that an acyclic and coprime generalized seed satisfies $A=U=S_{\\mathbf{x}}$ (Theorem 3.10), the equality that lets the class-group computation apply to the constructed algebras.","marker":"[BCDX20]"},{"why":"Introduces generalized cluster algebras and proves the Laurent phenomenon and finite-type classification, the foundational framework for the construction.","marker":"[CS13]"},{"why":"Provides the upper-cluster-algebra class-group computation and the strong-atom argument for cluster variables that are reused in the generalized setting.","marker":"[Pom25]"},{"why":"Introduces Laurent phenomenon algebras and their Laurent phenomenon, the subject of the final section's structural results.","marker":"[LP16]"}],"fun_headline_variants":["All finitely generated abelian groups arise as class groups","Generalized cluster algebras realize any abelian group as class group","Class groups of generalized cluster algebras cover all abelian groups","Any finitely generated abelian group can be a class group","Generalized cluster algebras: every abelian group as a class group"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The construction assumes that the exchange polynomial $x_N^m+1$ splits into exactly $m$ distinct irreducible factors over the ground field, which requires the characteristic of the field not to divide $m$; the theorem as stated only assumes an algebraically closed field, and when the characteristic divides $m$ the factor count and the resulting class group change.","fun_headline_variants_meta":{"raw":{"variants":["All finitely generated abelian groups arise as class groups","Generalized cluster algebras realize any abelian group as class group","Class groups of generalized cluster algebras cover all abelian groups","Any finitely generated abelian group can be a class group","Generalized cluster algebras: every abelian group as a class group"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001527,"raw_usage":{"total_tokens":6117,"prompt_tokens":953,"completion_tokens":5164,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":569,"completion_tokens_details":{"reasoning_tokens":5079}},"tokens_in":569,"tokens_out":5164,"duration_ms":35061,"temperature":1.0,"reasoning_tokens":5079,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T14:43:24.347231+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the paper's construction for $G=\\mathbb{Z}/p\\mathbb{Z}$ over an algebraically closed field of characteristic $p$: the exchange polynomial $(x_2+1)^p=x_2^p+1$ has only one irreducible factor, so Corollary 4.10 gives a single height-1 prime ideal over $x_1$ instead of $p$ distinct ones, and the presentation of Theorem 4.2 yields a free abelian class group rather than $\\mathbb{Z}/p\\mathbb{Z}$.","supporting_citations":[],"review_version":1}