{"id":"3ab2758d-eae9-4038-a4eb-50840b6a4916","arxiv_id":"1908.07331","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A quotient-ring reduction theorem for Smith forms of companion-matrix polynomials yields closed-form homology for all Brieskorn manifolds M(r,s,n) with r,s coprime.","lead":"This paper develops a general reduction for the Smith normal form of matrices that are polynomials in a companion matrix, and uses it to compute the integer homology of all Brieskorn 3-manifolds M(r,s,n) with r,s coprime. The upshot is a closed-form description of the first homology group, previously known only for r=2 or by algorithm.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem C's proof omits the entire x>1 case: the asserted Smith forms of f1(C_h1) and f2(C_h2) are not derived, so the main formula is not established for the general setting.","rationale":"The reader's conditional verdict already flagged the omitted Case 2 as a notable gap, but chose the topological bridge in Section 2.2 as the weakest assumption. My independent read also lands on the omitted Case 2, because it is an explicit, self-admitted missing proof inside the paper's own central algebraic argument. The topological bridge rests on established external theorems ([25, Theorem 3.1], [5, Proposition 7]) and is more likely to be correct; by contrast, the assertion about f1(C_h1) and f2(C_h2) is the precise step that produces the nonzero invariant factors in Theorem C for x>1. The paper gives no complete derivation for that step, so the theorem's general form is not verified. However, my spot checks of the intended formula and the analogous Case 1 argument suggest the omitted details can likely be filled, so I do not move the verdict: it stays conditional pending completion of Case 2 or a computational verification. Agreement is partial because the reader's stated weakest assumption is the topological bridge, not the internal gap.","tokens_in":16081,"tokens_out":41998,"duration_ms":373279,"concrete_test":"Compute in Sage for several triples with x>1, e.g. (r,s,n)=(21,10,15), (14,15,6), (15,14,21): let f(t)=((t^{rs}-1)(t-1))/((t^s-1)(t^r-1)), g(t)=t^n-1, form M=f(companion_matrix(g)) (or its degree-reduced circulant matrix), and compute M.smith_form(). Compare the sorted invariant factors with Theorem C's prediction: ones repeated n-(x-1)(y-1)-(y-x)-(x-1) times, r/x repeated y-x times, rs/(xy) repeated x-1 times, and 0 repeated (x-1)(y-1) times. A mismatch in any entry or multiplicity would falsify the omitted Case 2 claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 7, proof of Theorem C, Case 2: after reducing g to h=h1h2 and f to f1f2, the paper asserts without proof that the Smith form of f1(C_h1) has non-unit invariant factors (s/y) repeated x−1 times and that of f2(C_h2) has (r/x) repeated y−1 times, saying the argument is 'completely analogous' and 'we omit the details'. This is the entire content of Theorem C when x>1: the zero factors are handled by Lemma 5.3, but the nonzero factors (r/x)^{y−x} and (rs/(xy))^{x−1} arise only from multiplying these two asserted diagonal Smith forms. If either asserted invariant factor has an additional factor, or the multiplicities differ, Theorem C and hence Corollary D fail. The cited analogy is not a proof: the Case 1 argument used β=1,...,α_i and k|y, whereas Case 2 requires β to start at γ_i+1 (the exponent in x) and also requires a symmetric restriction on the s-side; the interaction of these restrictions with Lemma 7.2 is exactly what is left unverified. (Also, the theorem's wording 'non-unit invariant factors' is not literal when r/x or rs/(xy) equals 1, e.g., r=2,s=3,n=6, though this does not affect Corollary D.)","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a general theory of Smith forms of matrices of the form f(C_g), where C_g is the companion matrix of a monic polynomial g over an elementary divisor domain. Theorem A reduces f(C_g) to F(C_G) ⊕ 0_m when f and g share a monic common divisor z, with F=f/z and G=g/z; Corollary B identifies the last nonzero determinantal divisor as the resultant Res(F,G). After establishing auxiliary results on cyclotomic matrices (Theorem 7.4), the paper applies this machinery to the case where f(t) is the Alexander polynomial of the torus knot K(r,s) and g(t)=t^n−1. Theorem C claims a complete description of the non-unit invariant factors of the resulting circulant matrix, and Corollary D converts this into the first homology of the Brieskorn manifold M(r,s,n). The proof of Theorem A is elementary and detailed, and Theorem 7.4 is proved fully. The proof of Theorem C is complete for the case x:=gcd(r,n)=1, but the case x>1 contains a substantial omitted argument.","tokens_in":16373,"tokens_out":22206,"duration_ms":206788,"significance":"If the main results are correct, Theorem A is a useful and general reduction in the Smith-form theory of companion-matrix polynomials, and Theorem C together with Corollary D would settle the homology of all 3-dimensional Brieskorn manifolds M(r,s,n) with r,s coprime, generalizing the previously known r=2 case. The matrix-theoretic core is largely self-contained and rests on standard resultant and cyclotomic facts, and the proofs of Theorem A and Theorem 7.4 are written in sufficient detail to be checked. The paper's headline application, however, depends on an unproved step in Case 2 of the proof of Theorem C, so the central claim is not yet fully established as written.","major_comments":[{"comment":"The assertions that the Smith form of f1(C_{h1}) has non-unit invariant factors s/y repeated x−1 times, and that the Smith form of f2(C_{h2}) has non-unit invariant factors r/x repeated y−1 times, are not proved. The manuscript states 'as apart from this subtlety the argument is completely analogous, we omit the details.' These two assertions are load-bearing: together with the (x−1)(y−1) zero factors they are exactly the nonzero non-unit invariant factors claimed in Theorem C, and Corollary D depends on them. The analogy to Case 1 is not immediate, because in Case 2 the exponent β runs from γ_i+1 to α_i rather than from 1 to α_i, and because the reduction modulo h1 involves quotients of cyclotomic polynomials whose numerator and denominator both vanish at roots of h1; establishing the divisibility of the first determinantal divisor requires a separate argument, for example via evaluation of the quotient using derivatives or a suitable variant of Lemma 7.2. The authors need to supply the complete proof of these two Smith-form statements before Theorem C can be accepted.","section":"Section 7, proof of Theorem C, Case 2 (page 17)"}],"minor_comments":[{"comment":"The phrase 'non-unit invariant factors' is not literally correct in cases where r/x or rs/(xy) equals 1, for example r=2, s=3, n=6; in such cases some of the listed entries are units (1), and the Smith form also contains additional unit invariant factors whose number is not stated. The intended meaning is clear, but the wording should be adjusted, e.g., 'the invariant factors are: ..., together with the appropriate number of 1s.'","section":"Theorem C statement"},{"comment":"The notation γ_i is used for the exponent of the prime p_i in the factorization of x, but γ_i already denotes the i-th determinantal divisor in Section 2.1. This reuse is confusing; a different symbol, such as τ_i or β_i, would be preferable.","section":"Section 7, proof of Theorem C, Case 2 (notation)"},{"comment":"The sentence 'the sizes of the identity matrices is clear from the context' should be made explicit, since the identity blocks in I⊕f1(C_{h1}) and I⊕f2(C_{h2}) are essential for the final multiplication step; explicitly stating their dimensions would help the reader verify the product of the two Smith forms.","section":"Section 7, proof of Theorem C, Case 2 (dimensions)"},{"comment":"The abstract states 'Brieskorn manifolds M(r,s,n) where r,s are coprime' without specifying n≥2, although n≥2 is used throughout and in Corollary D; this should be stated for accuracy.","section":"Abstract and Introduction"}],"recommendation":"major_revision","confidential_remarks":"The missing proof in Case 2 of Theorem C is the only substantive obstacle to acceptance. I expect the omitted argument is fillable using the techniques of Case 1 together with a careful handling of cyclotomic quotient congruences, but the current manuscript does not establish the main application as written. The external topological results cited in Section 2.2 are standard and appropriate, and I see no circularity in the dependence on them."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThis paper deserves a serious look, but not a rubber stamp. The headline result is a genuinely useful reduction theorem: for an elementary divisor domain R, the Smith form of f(C_g) is the Smith form of F(C_G) plus a zero block, where F,G are quotients by a common divisor. That is clean, new, and broadly applicable; the proof via companion rings is elementary and convincing. The application to Brieskorn manifolds is the payoff: Theorem C gives a closed formula for the Smith form of the circulant matrix coming from the torus knot Alexander polynomial, and Corollary D turns it into H1 of M(r,s,n) for coprime r,s. This generalizes the known r=2 case and makes Randell's algorithm explicit. The external topological bridge (cyclic presentations from branched covers) is standard.\n\nThe soft spot is precisely where the stress-test lands. In the proof of Theorem C, the case x>1 is disposed of with 'we omit the details'. That is not a trivial case: it is the part of the theorem that goes beyond the r=2 setting, and the asserted invariant factors for f1(C_{h1}) and f2(C_{h2}) are exactly what produces the nonzero factors in the final answer. The analogy with Case 1 is plausible—the coprimality conditions that made the earlier argument work still hold after shifting the exponent range—but it is not immediate, and the interaction with the symmetric restriction on the other side is left unverified. As written, the theorem is not fully proven; a referee should require the details to be written out. If they check out, Theorem C stands and the paper is a solid contribution. There is also a minor wording issue: 'non-unit invariant factors' is not literal when r/x or rs/(xy) equals 1, though this doesn't affect the homology application.\n\nMy recommendation: send it to peer review. The main ideas are good, the reduction theorem is independently valuable, and the gap looks repairable. But the paper should not be accepted until the omitted case is fully proved, either in the text or in an appendix.","headline":"Clean Smith-form reduction plus a plausible Brieskorn formula; the main theorem's proof skips a case that must be written out before acceptance.","tokens_in":16880,"tokens_out":8605,"would_cite":true,"duration_ms":83167,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11C20","11C99","15A15","15A21","15B33","15B36","20J05","57M50"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves a quotient-polynomial reduction for Smith forms of companion-matrix polynomials and uses it to compute the first homology of every Brieskorn manifold $M(r,s,n)$ with $r,s$ coprime.","keywords":["Smith form","elementary divisor domain","circulant matrix","companion matrix","cyclically presented group","Brieskorn manifold","first homology","Alexander polynomial"],"falsifier":"Work out the integer Smith normal form of the circulant $f(C_g)$ for $r=5$, $s=3$, $n=6$ (so $g(t)=t^6-1$ and $f$ is the Alexander polynomial of $K(5,3)$); Theorem C predicts invariant factors $1,1,1,1,5,5$, so a direct computation producing any other multiset would refute the paper's formula.","tokens_in":15906,"feed_emoji":"🧶","tokens_out":13576,"duration_ms":117657,"temperature":0.7,"pith_summary":"The paper establishes a general reduction for Smith forms of matrices $f(C_g)$ formed by evaluating a polynomial at the companion matrix of another monic polynomial $g$, over any elementary divisor domain. If $f=Fz$ and $g=Gz$ share a monic factor $z$, then $f(C_g)$ is equivalent to the smaller matrix $F(C_G)$ together with a block of $\\deg z$ zeros; when $z$ is the $\\gcd$, the last non-zero determinantal divisor is the resultant $\\mathrm{Res}(F,G)$. This turns a matrix-computation problem into a polynomial-divisibility problem. The paper applies the reduction to the Alexander polynomial of the torus knot $K(r,s)$ with $g(t)=t^n-1$, giving the full Smith form of the resulting circulant matrix, and converts that into an explicit formula for the first homology of every 3-dimensional Brieskorn manifold $M(r,s,n)$ with $r,s$ coprime. The result matters because it unifies circulant, skew-circulant, and Toeplitz matrices under one mechanism and settles a whole family of knot-cover homologies rather than case by case.","feed_headline":"Smith forms reduce to a quotient, yielding Brieskorn homologies","feed_subtitle":"A quotient-polynomial reduction turns Alexander polynomials of torus knots into explicit homology groups.","key_machinery":"The companion ring $R_g=\\{f(C_g):f\\in R[t]\\}$ is isomorphic to the quotient ring $Q_g=R[t]/\\langle g\\rangle$ through the map sending $[t]$ to $C_g$; this lets polynomial division in $Q_g$ predict matrix equivalence over $R$. The proof of the reduction uses a unimodular Toeplitz similarity $U_z$ that block-triangularizes $C_g$ with respect to the factorization $g=Gz$, and Lemma 5.3 shows the leftover block contributes only zero invariant factors. The resultant formula then follows from the eigenpair description $\\theta\\mapsto f(\\theta)$ on the roots of $g$.","core_discovery":"The central claim is that Smith forms of matrices in companion rings are controlled by the quotient ring $R[t]/\\langle g\\rangle$, with a sharp reduction theorem: when $g=Gz$ and $f=Fz$ for a monic $z$, $f(C_g)\\sim F(C_G)\\oplus 0_{(\\deg z)\\times(\\deg z)}$, and the last non-zero determinantal divisor equals $\\mathrm{Res}(F,G)$ when $z$ is the $\\gcd$. Specializing to $R=\\mathbb{Z}$, $g(t)=t^n-1$, and $f$ the Alexander polynomial of the torus knot $K(r,s)$, Theorem C states that for coprime $r,s$, with $x=(r,n)\\le y=(s,n)$, the Smith form of the circulant $f(C_g)$ has non-unit invariant factors $r/x$ repeated $y-x$ times, $rs/(xy)$ repeated $x-1$ times, and $0$ repeated $(x-1)(y-1)$ times. Corollary D then identifies $H_1(M(r,s,n))$ with the corresponding direct sum of cyclic groups, with the roles of $r$ and $s$ swapped when $y<x$.","pith_inferences":["Beyond the paper, the same quotient reduction should compute Smith forms over any elementary divisor domain, not just $\\mathbb{Z}$; for example, over $\\mathbb{F}[x]$ it would give the invariant-factor structure of polynomial circulant analogues.","A testable extension is to apply Theorem A to Alexander polynomials of torus links rather than knots; the reduction and cyclotomic technology should still constrain the homology of cyclic branched covers of links.","The resultant formula suggests a practical check: for random coprime $f,g\\in\\mathbb{Z}[t]$, the absolute value of $\\mathrm{Res}(F,G)$ should equal the product of the nonzero invariant factors of $f(C_g)$, a computation that can be run without any of the paper's topological language."],"forward_implications":["Whenever $f$ and $g$ share a monic divisor $z$, the Smith form of $f(C_g)$ is the Smith form of the smaller matrix $F(C_G)$ plus a zero block of size $\\deg z$; in particular the nonzero invariant factors can be read from a matrix of smaller dimension.","If $z$ is the full $\\gcd$, the last non-zero determinantal divisor of $f(C_g)$ is the resultant $\\mathrm{Res}(F,G)$, which over $\\mathbb{Z}$ gives the order of the largest torsion subquotient when the matrix is singular.","For the Alexander polynomial of $K(r,s)$ and $g(t)=t^n-1$, the non-unit invariant factors are $r/x$ repeated $y-x$ times and $rs/(xy)$ repeated $x-1$ times, with $(x-1)(y-1)$ zero invariant factors, where $x=(r,n)\\le y=(s,n)$.","Consequently $H_1(M(r,s,n))$ is the displayed direct sum of cyclic groups for every coprime pair $(r,s)$, reproducing the known $r=2$ family and giving trivial homology exactly when $r,s,n$ are pairwise coprime."],"supporting_citations":[{"why":"Supplies the cyclic presentation of the fundamental group of an $n$-fold cyclic branched cover of a $(1,1)$-knot.","marker":"[25, Theorem 3.1]"},{"why":"Shows the representer polynomial of that presentation is the Alexander polynomial projected modulo $t^n-1$.","marker":"[5, Proposition 7]"},{"why":"Earlier computation of the $r=2$ family that Corollary D generalizes and against which the result can be checked.","marker":"[6]"},{"why":"Gives resultant formulas for pairs of cyclotomic polynomials used to compute the invariant factors in Theorems 7.4 and C.","marker":"[1]"},{"why":"Provides the resultant $\\mathrm{Res}(\\Phi_m,t^n-1)$ used repeatedly in the proof.","marker":"[9, Theorem 3]"},{"why":"States the Smith theorem for elementary divisor domains, the background framework for all invariant-factor statements.","marker":"[12, Theorem 1.14.1]"},{"why":"Gives cyclotomic identities such as $\\Phi_{np^k}(t)=\\Phi_n(t^{p^k})/\\Phi_n(t^{p^{k-1}})$ used in the proof of Theorem 7.4.","marker":"[26, p. 160]"}],"fun_headline_variants":["Circulant Smith forms compute Brieskorn homology groups","From Alexander polynomials to Brieskorn homology via Smith forms","Quotient ring reduction yields Smith form and Brieskorn H1","Explicit homology of Brieskorn manifolds via circulant Smith forms","Smith form of circulant matrices reveals Brieskorn homologies"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise of the topological application is that, for coprime $r,s$, the $n$-fold cyclic branched cover of $S^3$ over $K(r,s)$ has a cyclic presentation whose representer polynomial is the projection of the torus-knot Alexander polynomial into $\\mathbb{Z}[t]/\\langle t^n-1\\rangle$; the homology conclusion collapses if that external bridge fails.","fun_headline_variants_meta":{"raw":{"variants":["Circulant Smith forms compute Brieskorn homology groups","From Alexander polynomials to Brieskorn homology via Smith forms","Quotient ring reduction yields Smith form and Brieskorn H1","Explicit homology of Brieskorn manifolds via circulant Smith forms","Smith form of circulant matrices reveals Brieskorn homologies"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000889,"raw_usage":{"total_tokens":3913,"prompt_tokens":1097,"completion_tokens":2816,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":713,"completion_tokens_details":{"reasoning_tokens":2723}},"tokens_in":713,"tokens_out":2816,"duration_ms":19688,"temperature":1.0,"reasoning_tokens":2723,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:22:17.477959+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Work out the integer Smith normal form of the circulant $f(C_g)$ for $r=5$, $s=3$, $n=6$ (so $g(t)=t^6-1$ and $f$ is the Alexander polynomial of $K(5,3)$); Theorem C predicts invariant factors $1,1,1,1,5,5$, so a direct computation producing any other multiset would refute the paper's formula.","supporting_citations":[{"cited_title":"Cavicchioli","cited_arxiv_id":null,"evidence_quote":"Earlier computation of the $r=2$ family that Corollary D generalizes and against which the result can be checked."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives resultant formulas for pairs of cyclotomic polynomials used to compute the invariant factors in Theorems 7.4 and C."}],"review_version":1}