{"id":"33d1beca-0796-4318-ac1c-881e0362b80b","arxiv_id":"1908.04966","paper_version":1,"verdict":"REJECT","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"Using Eisenstein integer module theory, the paper classifies distributive Mendelsohn triple systems with order coprime to 3, but its enumeration formula for primes congruent to 1 mod 3 is incorrect.","lead":"This math paper classifies distributive Mendelsohn triple systems whose order is coprime to 3, using the arithmetic of Eisenstein integers. The structural classification is correct, but the paper's formula for counting isomorphism classes is wrong.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 3.8(a) sums independent factor choices instead of multiplying them, undercounting isomorphism classes for p ≡ 1 mod 3.","rationale":"The reader's verdict of REJECT is correct and is driven by the enumeration error, which we confirm and refine. The error in Theorem 3.8(a) is structural: independent choices for different part sizes are summed rather than multiplied, so the formula undercounts for every p ≡ 1 mod 3 and every n with a partition having at least two distinct part sizes. This invalidates the paper's stated enumeration claim. The classification Theorem 3.7 appears sound; the GAP-dependent part of Lemma 3.4 concerns only the ramified case S=Z/3^n, not the DNR classification, so it is not the load-bearing weakness for the central claim. The concrete check for n=4 decisively settles the issue: the correct count is 20, not 19, and the discrepancy comes from the partition (2,1,1).","tokens_in":22005,"tokens_out":11273,"duration_ms":105250,"concrete_test":"Compute d(7^4) by direct enumeration: list all Z[zeta]-modules of order 7^4 (decompositions into Z[zeta]/(pi^r) and Z[zeta]/(bar-pi^r) pieces with total exponent 4), form the corresponding quasigroups Lin(M,R), and count isomorphism classes via Theorem 2.12 (conjugacy of R in Aut(M)). The corrected product formula gives 20; the paper's formula gives 19. Alternatively, for the single partition (2,1,1), explicitly construct the six modules consisting of one Z[zeta]/(pi^2)-or-(bar-pi^2) piece and three Z[zeta]/(pi)-or-(bar-pi) pieces, and verify they are pairwise non-isomorphic.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 3.8(a) claims d(p^n) = sum over partitions (X,mu) of n of (sum_{r in X} mu(r)+1). For a fixed partition, the choices of root a vs a^{-1} on the mu(r) copies of Z/p^r are independent across distinct part sizes r: these correspond to distinct primary components of the Z[zeta]-module with different annihilators, and quasigroup isomorphism coincides with module isomorphism by Theorem 2.12. The correct count for that partition is the product over r in X of (mu(r)+1), not the sum. Concretely, for n=4 and partition (2,1,1), there are (1+1)=2 choices for the Z/p^2 factor and (2+1)=3 choices for the eigenvalue multiplicity on the three Z/p factors, giving 6 classes, whereas the paper's sum gives (1+1)+(2+1)=5. Summing over all partitions of 4 gives 20 classes, not 19. Thus the enumeration claim in the abstract and in Theorem 3.8(a) is false as stated; replacing the inner sum by a product would repair the enumeration without affecting Theorem 3.7.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies distributive Mendelsohn triple systems (MTS), i.e., Mendelsohn quasigroups that are self-distributive, and focuses on the case of order coprime with 3, called distributive non-ramified (DNR). The main structural theorem, Theorem 3.7, states that every DNR MTS is a direct product of linear MTS of the form Lin(Z/p^t, a) for primes p ≡ 1 mod 3 and Lin(Z/q^u[ζ]) for primes q ≡ 2 mod 3. This is obtained by representing entropic Mendelsohn quasigroups as finite modules over the Eisenstein integers Z[ζ] via a functor in Theorem 2.15, then applying the structure theory of modules over a PID. The paper also proposes an enumeration of isomorphism classes in Theorem 3.8, gives partial results for orders divisible by 3, proves that an entropic MTS is non-ramified iff it is pure iff it is self-orthogonal, and discusses self-converse systems. The core classification framework is elegant and appears sound, but the paper contains several false or unsupported load-bearing claims, especially in the enumeration formula for p ≡ 1 mod 3 and in the characterization of self-converse systems.","tokens_in":22290,"tokens_out":13923,"duration_ms":126534,"significance":"If the results were correct as stated, the paper would be a significant contribution: it gives a structural classification of all distributive Mendelsohn triple systems of order coprime with 3 using the Eisenstein integers, extends earlier work by Donovan, Griggs, McCourt, Opršal, and Stanovský, and provides a uniform module-theoretic framework that is parameter-free and derived from standard theorems (Fischer-Galkin-Smith, Kepka-Nemec, Bruck-Murdoch-Toyoda). The classification theorem is conceptually clean and likely correct. However, the enumeration claim in Theorem 3.8(a) is false, the self-converse characterization in Theorem 5.9 is false, and a key lemma used in the ramified case relies on an unshipped computer calculation. These errors affect both the abstract's enumeration promise and several secondary theorems, so the paper cannot be accepted in its present form. The strengths of the framework justify asking for a major revision rather than immediate rejection.","major_comments":[{"comment":"The enumeration formula (3.5) is incorrect: for a fixed partition (X, μ) of n, the choices of root a vs. a^{-1} on the μ(r) copies of Z/p^r are independent across distinct part sizes r, because these copies correspond to distinct primary components of the Z[ζ]-module with different annihilators, and Theorem 2.12 identifies quasigroup isomorphism with module isomorphism. The count for a fixed partition is therefore ∏_{r∈X}(μ(r)+1), not Σ_{r∈X}(μ(r)+1). For example, the partition (2,1,1) of n=4 gives (1+1)(2+1)=6 classes, not 5, and summing over all partitions of 4 gives 20, not 19. The proof's sentence 'This count applies to each element of X' incorrectly converts a product of independent choices into a sum. Replacing the inner sum by a product repairs the enumeration and does not affect Theorem 3.7.","section":"Theorem 3.8(a), Eq. (3.5)"},{"comment":"The converse of Lemma 5.8 is false: an isomorphism between direct products need not restrict to isomorphisms of the individual factors, and for linear MTS the relevant automorphism may permute factors. Consequently Theorem 5.9 is false. For p ≡ 1 mod 3, the DNR MTS Lin(Z/p, a) × Lin(Z/p, a^{-1}) is self-converse, since the swap automorphism of (Z/p)^2 conjugates the diagonal action diag(a, a^{-1}) to its inverse, yet its order has a prime p ≡ 1 mod 3. The correct self-converse criterion is a symmetry condition on the multiset of exponents attached to π and π̅ for each p ≡ 1 mod 3, not the condition that all primes dividing the order be ≡ 2 mod 3. This also invalidates Conjecture 5.10 and the self-converse portion of the theorem stated in the introduction.","section":"Lemma 5.8 and Theorem 5.9"},{"comment":"The proof of Lemma 3.4 for S = Z/9 depends on a GAP verification that is not shipped, and the argument for n ≥ 3 reduces to that verification. Since Lemma 3.4 is used in Proposition 5.4 and in the proof of the equivalence Theorem 5.13 for all entropic MTS, the computation (code and output, or a mathematical proof) must be included before the claimed equivalence is established. This is a missing support for a stated theorem, not merely a presentation issue.","section":"Lemma 3.4 (case S = Z/3^n)"},{"comment":"The introductory theorem characterizing DNR MTS states that an entropic Mendelsohn triple system has order coprime with 3 if and only if it is pure if and only if it is 'self-converse (orthogonal to its converse).' This conflates two distinct notions: self-converse means isomorphic to the converse (Definition 5.7), while 'orthogonal to its converse' is self-orthogonality (Definition 5.11). The paper's own Theorem 5.9 and Theorem 5.13 give different characterizations for these two properties, so the introductory statement is internally inconsistent and must be corrected.","section":"Introduction, second displayed theorem"}],"minor_comments":[{"comment":"As written, Lemma 3.4 quantifies over a prime p ≡ 2 mod 3 and then allows S = Z/3^n; since 3 is not congruent to 2 mod 3, the lemma should be split into separate cases for Z, for Z/p^n with p ≡ 2 mod 3, and for Z/3^n.","section":"Lemma 3.4 statement"},{"comment":"In Example 2.13, the text 'Lin(Z/7, 3), Lin(Z/7, 3)' should presumably read 'Lin(Z/7, 5)' in the second occurrence.","section":"Example 2.13"},{"comment":"There is a typo in Proposition 4.2 ('By Theorem, 2.23.(c)') and a missing comma after 'Theorem'.","section":"Section 4, Proposition 4.2"},{"comment":"The acronym DNR is introduced as 'distributive, non-ramified'; the standard English term is 'unramified'. Consider using 'unramified' throughout for consistency with number-theoretic usage.","section":"Global terminology"},{"comment":"The proof of Proposition 3.5 invokes Nakayama's lemma and a result from [28] after reducing modulo p; the reduction step is only sketched. A sentence explaining why the lifted minimal generating set is a basis (beyond the cited theorem) would help the reader.","section":"Proposition 3.5 proof"}],"recommendation":"major_revision","confidential_remarks":"The core classification theorem is valuable and likely correct, and the module-theoretic framework is a genuine strength. However, the paper as submitted contains multiple false statements in load-bearing places: the enumeration formula in Theorem 3.8(a), the self-converse theorem and its lemma, and the introductory characterization theorem. These are correctable in principle, which is why I recommend major revision rather than rejection. The GAP verification for Lemma 3.4 must be supplied or the affected results must be weakened; otherwise the purity/self-orthogonality equivalence for all entropic MTS is not fully established. The author should also be asked to reconcile the introduction's terminology with the actual definitions in Section 5."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nTwo things you should know. First, the structural theorem—every distributive Mendelsohn triple system of order coprime to 3 is a direct product of Lin(Z/p^t, a) factors for p ≡ 1 mod 3 and Lin(Z/q^u[ζ]) factors for q ≡ 2 mod 3—is a genuine extension of Donovan et al. and looks right. Second, the companion enumeration in Theorem 3.8(a) is wrong: for a fixed partition (X, μ) of n, the choices of root a vs a^{-1} on the various factors are independent across distinct part sizes, so the number of isomorphism classes is the product over r ∈ X of (μ(r)+1), not the sum. The paper's formula undercounts d(p^n) already at n=4. The structural classification survives, but the advertised enumeration is false as stated.\n\nWhat the paper does well: the Eisenstein-integer module framework is the right language, and it produces a clean classification for all prime powers rather than just p and p^2. The Section 5 results—for entropic MTS, pure is equivalent to self-orthogonal and to non-ramified—are correct and nicely proved. The exposition is mostly careful about module isomorphism versus quasigroup isomorphism.\n\nSoft spots: the enumeration bug is the main one; it is localized and fixable by replacing the inner sum with a product, with no damage to Theorem 3.7. Second, Lemma 3.4 relies on a GAP computation over Z/9 that is not shipped. That lemma is load-bearing for the inert-prime classification, so the paper currently rests on an unverifiable finite check. That may be easy to close, but as written it is a gap. The ramified case is explicitly partial and the conjectures are labelled as such, which is fine.\n\nWho this is for: people working on distributive quasigroups or Mendelsohn triple systems. I would not cite the enumeration as stated, but I would engage with the classification after the matrix lemma is checked and the counting fixed.\n\nRecommendation: send it to peer review. The core structural result is solid, the flaw is real but repairable, and the paper deserves referee time.","headline":"Real structural classification with a fixable but real enumeration bug; worth peer review despite the error.","tokens_in":22763,"tokens_out":2861,"would_cite":false,"duration_ms":28597,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["20N05","05B07"],"pacs":[],"model":"deepseek-v4-flash","headline":"Every distributive Mendelsohn triple system of order coprime with 3 decomposes via the Eisenstein integers.","keywords":["Mendelsohn triple system","distributive quasigroup","Eisenstein integers","linear quasigroup","entropic quasigroup","enumeration","self-orthogonality"],"falsifier":"Perform the missing exhaustive search over $M_2(\\mathbb{Z}/9)$: a single matrix $A$ with $A^2-A+I=0$ and $\\det(A)\\neq1$ would refute Lemma 3.4 and remove the uniqueness of the inert factor $\\operatorname{Lin}(\\mathbb{Z}/q^n[\\zeta])$ from Theorem 3.7.","tokens_in":21820,"feed_emoji":"🧩","tokens_out":11438,"duration_ms":104875,"temperature":0.7,"pith_summary":"This paper classifies all distributive Mendelsohn triple systems (MTS) whose number of points is coprime with $3$. It shows that every such system is a direct product of small cyclic linear systems, each built from a prime and a choice of a root of $X^2-X+1$ modulo a prime power, and it turns this description into explicit formulas for the number of isomorphism classes. The proof works by translating a Mendelsohn quasigroup into a module over the Eisenstein integers, where the structure theory of modules over a Euclidean domain supplies the decomposition. The paper also proves that for linear MTS the properties of being of order coprime with $3$, pure, and self-orthogonal are equivalent.","feed_headline":"All non-3-divisible Mendelsohn triple systems classified","feed_subtitle":"Every such system splits into cyclic factors tied to primes, giving exact counts of isomorphism classes.","key_machinery":"The load-bearing object is the Eisenstein integer ring $\\mathbb{Z}[\\zeta]=\\mathbb{Z}[X]/(X^2-X+1)$ with $\\zeta=e^{\\pi i/3}$. A linear Mendelsohn quasigroup $\\operatorname{Lin}(M,R)$ on an abelian group $M$ is exactly an Eisenstein module: multiplication is $xy=xR+y(1-R)$, and the semisymmetric law forces $R^2-R+1=0$, so $R$ acts as $\\zeta$. Because $\\mathbb{Z}[\\zeta]$ is a Euclidean domain, every finite module splits into cyclic primary components $\\mathbb{Z}[\\zeta]/(\\pi^n)$, and the prime classification separates split primes ($p\\equiv1\\pmod3$), inert primes ($p\\equiv2\\pmod3$), and the ramified prime $(1+\\zeta)$. The key technical lemma shows that over $\\mathbb{Z}/p^n$ with $p\\equiv2\\pmod3$, every $2\\times2$ matrix annihilated by $X^2-X+1$ has determinant and trace $1$, so it is similar to the companion matrix; this pins down the unique inert factor.","core_discovery":"The central claim is Theorem 3.7: if $Q$ is a distributive Mendelsohn quasigroup of order $n=\\prod_i p_i^{r_i}\\prod_j q_j^{s_j}$, with $p_i\\equiv1\\pmod3$ and $q_j\\equiv2\\pmod3$, then $Q$ is isomorphic to a direct product of factors $\\operatorname{Lin}(\\mathbb{Z}/p^{t},a)$, where $a$ is a root of $X^2-X+1$ modulo $p^{t}$, and factors $\\operatorname{Lin}(\\mathbb{Z}/q^{u}[\\zeta])$, the unique class on the module $(\\mathbb{Z}/q^u)^2$ with multiplication by the companion matrix of that polynomial. Each way of splitting the exponent $r_i$ into a partition of $t$-values gives a different isomorphism class, so the number $d(p^n)$ of classes is a sum over integer partitions of $n$ for $p\\equiv1\\pmod3$, while $p\\equiv2\\pmod3$ gives $d(p^{2n})=PE(n)$, the number of partitions of $2n$ into even parts. Theorem 5.13 adds that for any entropic (abelian-group-linear) MTS, non-ramified, pure, and self-orthogonal are equivalent.","pith_inferences":["Beyond the paper: the theorem gives an isomorphism certificate: two DNR MTS of the same order are isomorphic exactly when their split-prime partition data and root choices agree, so isomorphism testing reduces to comparing partition records.","Beyond the paper: the same partition-counting pattern across split, inert, and ramified primes suggests a uniform conjecture for all linear MTS: the number of isomorphism classes of a fixed order should always be a partition count of the exponents, with split primes contributing root multiplicities.","Beyond the paper: the principal isotopy to a left Eisenstein quasigroup raises the possibility of reading the triples of an MTS as coordinates in a $3$-web over an Eisenstein module, a geometric reading the manuscript does not pursue."],"forward_implications":["Every DNR MTS is pure: no two distinct points commute under the quasigroup multiplication.","For $p\\equiv1\\pmod3$, $d(p^n)=\\sum_{(\\mathcal X,\\mu)\\vdash n}\\sum_{r\\in\\mathcal X}(\\mu(r)+1)$; for $p\\equiv2\\pmod3$, $d(p^n)=PE(n)$, so in particular $d(p^{2k+1})=0$.","A DNR MTS is self-converse exactly when all primes dividing its order are $2\\pmod3$.","Entropic MTS of order divisible by $3$ are never pure; the paper conjectures the isomorphism classes of order $3^n$ are counted by the partition number $P(n)$.","Every distributive Mendelsohn quasigroup is principally isotopic to a left Eisenstein quasigroup, linking these systems to $3$-web coordinatization."],"supporting_citations":[{"why":"Supplies the prior classification for orders $p$ and $p^2$ and the root-count lemma for $X^2-X+1$ modulo prime powers that the extension builds on.","marker":"[13]"},{"why":"Supplies the prime-power decomposition theorem that reduces distributive quasigroups to the entropic (abelian-group-linear) case.","marker":"[14, 15, 32]"},{"why":"Gives the conjugacy criterion for isomorphism of linear piques, used to identify the standard representatives in Propositions 3.3 and 3.5.","marker":"[22]"},{"why":"Provides the criterion for similarity of a matrix to the companion of its characteristic polynomial over a commutative ring, the step that forces the inert factor into standard form.","marker":"[28]"},{"why":"Supplies Nakayama's lemma and basis lifting over local rings, used to find the cyclic vector in the inert and even-ramified cases.","marker":"[23]"},{"why":"Establishes the Euclidean-domain structure of $\\mathbb{Z}[\\zeta]$ and the classification of its primes, on which the split/inert/ramified trichotomy rests.","marker":"[19]"},{"why":"Provide the descriptions of quotient rings of the Eisenstein integers by primary ideals used in Theorem 2.23.","marker":"[5, 25]"}],"fun_headline_variants":["Eisenstein integers solve Mendelsohn triple systems","3-coprime Mendelsohn triples classified exactly","Non-3-divisible Mendelsohn triples fully classified","Exact classification of Mendelsohn triples via Eisenstein integers","Coprime-order Mendelsohn triples completely classified"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The classification of the inert-prime and even-ramified factors rests on Lemma 3.4, whose verification for matrices over $\\mathbb{Z}/9$ is a computer calculation not included in the manuscript; if that check fails, the companion-matrix form for those factors is not established.","fun_headline_variants_meta":{"raw":{"variants":["Eisenstein integers solve Mendelsohn triple systems","3-coprime Mendelsohn triples classified exactly","Non-3-divisible Mendelsohn triples fully classified","Exact classification of Mendelsohn triples via Eisenstein integers","Coprime-order Mendelsohn triples completely classified"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001541,"raw_usage":{"total_tokens":6219,"prompt_tokens":1054,"completion_tokens":5165,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":670,"completion_tokens_details":{"reasoning_tokens":5088}},"tokens_in":670,"tokens_out":5165,"duration_ms":36820,"temperature":1.0,"reasoning_tokens":5088,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:30:10.262319+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Perform the missing exhaustive search over $M_2(\\mathbb{Z}/9)$: a single matrix $A$ with $A^2-A+I=0$ and $\\det(A)\\neq1$ would refute Lemma 3.4 and remove the uniqueness of the inert factor $\\operatorname{Lin}(\\mathbb{Z}/q^n[\\zeta])$ from Theorem 3.7.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the prior classification for orders $p$ and $p^2$ and the root-count lemma for $X^2-X+1$ modulo prime powers that the extension builds on."},{"cited_title":"Kepka and P","cited_arxiv_id":null,"evidence_quote":"Gives the conjugacy criterion for isomorphism of linear piques, used to identify the standard representatives in Propositions 3.3 and 3.5."},{"cited_title":"Prokip, On similarity of matrices over commutative rings , Linear Algebra Appl., 399 (2005), 225-233","cited_arxiv_id":null,"evidence_quote":"Provides the criterion for similarity of a matrix to the companion of its characteristic polynomial over a commutative ring, the step that forces the inert factor into standard form."},{"cited_title":"Matsumura, Commutative Ring Theory , Cambridge Univ","cited_arxiv_id":null,"evidence_quote":"Supplies Nakayama's lemma and basis lifting over local rings, used to find the cyclic vector in the inert and even-ramified cases."},{"cited_title":"Ireland and M","cited_arxiv_id":null,"evidence_quote":"Establishes the Euclidean-domain structure of $\\mathbb{Z}[\\zeta]$ and the classification of its primes, on which the split/inert/ramified trichotomy rests."}],"review_version":1}