{"id":"7d6f39df-a970-4b73-b0d5-6a3e68ef87d1","arxiv_id":"1908.04147","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For stable genera, Bousquet-Mélou-Schaeffer numbers factor as a product of explicit Pochhammer-type factors times a polynomial in the partition parts, and the paper proves this combinatorially.","lead":"This paper proves a structural quasi-polynomiality property for Bousquet-Mélou-Schaeffer numbers, which count tuples of permutations with prescribed cycle types. The proof is purely combinatorial, using the semi-infinite wedge formalism, and does not rely on the topological recursion that was conjectured to imply it.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The proof of Theorem 6.1 rests on quasi-rationality identities whose proof omits three parameter regimes after eq. (80) and eq. (84); if any omitted case hides an extra pole or degree excess, the polynomiality conclusion fails.","rationale":"The reader's conditional verdict identifies the correct weak point: the quasi-rationality of the A-operators in Propositions 5.11 and 5.13 is exactly what converts finite sums of wedge-space matrix elements into the factorized form of Theorem 6.1. The written proof handles only the most favorable parameter regime in detail and postpones the remaining cases with 'easier' analogues. Since Theorem 6.1 is the main combinatorial claim and is not independently machine-checked, this is a genuine structural gap in the exposition. I do not see a reason to move away from the conditional verdict: the omitted computations are likely routine and could be supplied, but until they are written out or verified symbolically, the proof is not fully rigorous. The topological-recursion derivation in Section 2 is motivational for Theorem 6.1 and is not what the combinatorial proof depends on, so I do not treat the analytic-continuation issues there as the primary blocker.","tokens_in":30547,"tokens_out":41457,"duration_ms":440924,"concrete_test":"Use a symbolic CAS to verify Proposition 5.11 for small m, p, q by computing both sides: for m = 2, 3, 0 ≤ p ≤ 3, 1 ≤ q ≤ 5, sum eq. (74) explicitly over all i_1 + ... + i_m = q + k and all σ, then polynomial-interpolate S_{p,l,q}(k) in k; repeat for the identity-part formula eq. (81) versus Proposition 5.13. If any run yields a non-polynomial S, an extra pole, or a degree exceeding 6p + q, the proof of Theorem 6.1 fails in an omitted regime.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 5.1.3 proves Proposition 5.11 only for the regime q ≥ σ and σ ≥ m; the other three combinations are dismissed as 'almost completely analogous, only easier' directly after eq. (80). Proposition 5.13 similarly defers the case σ < m after eq. (84). For q < σ, identity (80) cannot be applied because q + k − σ < k, and the individual summands in eq. (74) acquire extra denominators such as (k − r); the text gives no cancellation argument for these summands. In addition, Proposition 5.10 states that Q^p_{0,...,0}(k,l) is divisible by k, but for p = 0 this is false (Q^0_{0,...,0} = 1), so the claimed cancellation of the 1/k factor in the σ = 0 term needs a separate argument. Proposition 5.20 also defers 'all other possible variants' of the q, σ, m positioning to an analogous computation. If any of these omitted regimes produces an extra pole at a negative integer or a numerator of higher degree, the residue cancellation in eq. (147) and the conclusion that Poly_{g,n} is a polynomial would fail. This is the load-bearing step of Theorem 6.1, and no machine-checked verification or explicit derivation of the omitted cases is supplied.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proves a structural quasi-polynomiality theorem for Bousquet-MélouâSchaeffer numbers. After setting up the OrlovâScherbin tau-function description and the semi-infinite wedge formalism, the authors introduce A-operators and establish quasi-rationality statements for their matrix elements. The main Theorem 6.1 asserts that in the stable range the connected BMS number b^o_{g,mu} equals an explicit factorial prefactor times a polynomial Poly_{g,n}(mu_1,...,mu_n) times a product of factors (mu_i-j_i)/m over a finite range of j_i not divisible by m. The proof combines these quasi-rationality statements with a residue argument that eliminates poles at negative integers. Section 2 additionally derives the BouchardâEynard topological recursion for BMS numbers from the AlexandrovâChapuyâEynardâHarnad theorem.","tokens_in":30818,"tokens_out":15680,"duration_ms":150481,"significance":"If Theorem 6.1 is fully established, it settles the conjectured quasi-polynomiality predicted by topological recursion and provides an ELSV-type structure for BMS numbers. The proof is combinatorial, uses explicit fermionic Fock-space operators, and is independent of the still-incomplete CohFT/topological-recursion correspondence. This is a substantial contribution in the line of combinatorial proofs of quasi-polynomiality for Hurwitz-type numbers. The derivation of topological recursion for BMS numbers from the ACHE theorem is a useful addendum, although it is not needed for the main theorem.","major_comments":[{"comment":"The divisibility assertion used to cancel the 1/k factor in the sigma=0 term is false for p=0: for p=0 we have R_0(k,l,i_1,...,i_m)=1 and hence Q^0_{0,...,0}(k,l)=1, which is not divisible by k. This matters because p=0, q>0 contributions occur in stable correlators, for instance in the (g,n)=(0,3) case with shifts q_1=1, q_2=-1, q_3=0. As written, the proof of Proposition 5.11 does not establish the required denominator form for these sigma=0, p=0 terms. A direct cancellation using the factor (mk)_m in the numerator may repair the argument, but the repair is not present in the manuscript and must be supplied.","section":"Section 5.1.2â5.1.3, Propositions 5.9â5.11"},{"comment":"Several load-bearing parameter regimes are deferred rather than proved. Proposition 5.11 is proved only for q >= sigma and sigma >= m; the three other combinations of q, sigma, and m are dismissed as 'almost completely analogous, only easier.' Proposition 5.13 defers the case sigma < m, and Proposition 5.20 defers 'all other possible variants' of q, sigma, and m. These omissions are not merely cosmetic: for q < sigma, the factorial (q+k-sigma)! and the falling factorial (mk-sigma)_{k+q-sigma} in eq. (74) have a different shape, and the displayed cancellation of factors (k-r) in eq. (80) does not automatically apply. If any omitted case produced an extra pole at a positive integer, at a point j/m, or at a point j/(m-1), the residue cancellation in eq. (147) and the conclusion that Poly_{g,n} is a polynomial would fail. The authors should provide complete derivations for all these regimes.","section":"Section 5.1.3, after eq. (80); Section 5.1.3, after eq. (84); Section 5.2.3, after eq. (131)"}],"minor_comments":[{"comment":"The passage from the ACHE theorem outside the discriminant to the discriminant point epsilon=(1,...,1) is justified by an asserted analyticity of both sides in epsilon. Since the contour and the differentials are defined through residues that may collide at the discriminant, a more detailed analytic-continuation argument should be given, or the statement should be clearly marked as conditional. This does not affect the main combinatorial theorem, which is independent of Section 2.","section":"Section 2.1, Proposition 2.1"},{"comment":"The displayed formulas for the coefficients of the A-operators omit some parentheses around the rational prefactor, making the order of operations hard to parse. Adding explicit parentheses would improve readability.","section":"Section 5.1.3, eqs. (74) and (81)"},{"comment":"The statement that R_p(k,l,0,...,0) is divisible by k should be amended to exclude p=0 or replaced by the weaker statement used in the actual argument. As written, Proposition 5.9 is false for p=0, and this is not merely a typo because the proof of Proposition 5.11 invokes exactly this divisibility.","section":"Section 5.1.2, Proposition 5.9"},{"comment":"In the passage following eq. (152), the degree bound is explained only for the variable mu_1, and symmetry is then used to conclude polynomiality in all variables. Since the connected correlator is symmetric and the bound is independent of the other variables, the argument is sound, but a sentence spelling out that the rational function has no other poles as a function of mu_1 would make the reasoning more transparent.","section":"Section 6, proof of Theorem 6.1"}],"recommendation":"major_revision","confidential_remarks":"The paper is likely correct in substance, and Theorem 6.1 is an important statement. However, the proof of the central quasi-rationality statements is not self-contained for several parameter regimes that are load-bearing for the final residue argument. The false divisibility claim for p=0 in Propositions 5.9 and 5.10 is a concrete local error that indicates Section 5.1.3 needs a careful rewrite. I recommend major revision with a request to supply the omitted computations in full rather than a rejection, since the gaps appear fixable within the manuscript's scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear [Colleague],\n\nThe headline is that this paper proves a structural quasi-polynomiality statement for Bousquet-Mélou–Schaeffer numbers that was previously only conjectural, and it does so by a purely combinatorial route, avoiding the not-yet-existing Givental/topological-recursion correspondence. That is the main result to know. Theorem 6.1 is new and the paper is honest that the TR side is a small addendum to Alexandrov–Chapuy–Eynard–Harnad; the independent proof of the polynomiality is the actual contribution.\n\nThe technical machinery — A-operators in the semi-infinite wedge, difference operators, residue cancellations — is developed carefully and the overall strategy is coherent. I find the proof largely convincing. The paper is also refreshingly candid about the missing theory in Remarks 2.2–2.4.\n\nWhere I would push back is not the theorem but the completeness of the written proof. In Section 5.1.3, Proposition 5.11 is proved only for q ≥ σ and σ ≥ m; the other three regime combinations are dismissed with 'almost completely analogous, only easier' right after (80), and Proposition 5.13 similarly defers σ < m. Proposition 5.20 also punts on 'all other possible variants' of the q, σ, m ordering. These regimes are load-bearing: the cancellation of poles at negative integers in Section 6 depends on the exact rational form of the A-operator coefficients. The stress-test note worries that an omitted regime could hide an extra pole or a higher-degree numerator; I can't rule that out from the text, and neither can the reader of the arXiv version. I suspect the omitted cases really are easier, but a referee should ask for them to be written out. The same goes for the statement in Proposition 5.10 that Q^p_{0,...,0} is divisible by k — for p=0 it's false as stated, though the 1/k cancellation in that case can be rescued because (mk)_{q+k} already contains a factor of k. Still, the text's explanation is not accurate there.\n\nThe analytic continuation in Proposition 2.1 is asserted by analyticity in ε rather than demonstrated; that part is peripheral to the main theorem, which does not rely on it.\n\nAll in all: the central result is strong and the proof strategy is sound, but the written proof has a few corners that need to be filled before I'd call it fully rigorous. This deserves a serious referee — it's exactly the kind of paper that should go through peer review with requests for clarification, not be desk-rejected.\n\nBest,\n[You]","headline":"A new, genuinely combinatorial quasi-polynomiality theorem for BMS numbers, with a mostly solid proof that leaves a few parameter checks to the reader.","tokens_in":31342,"tokens_out":9887,"would_cite":true,"duration_ms":96468,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["05A15","05E10","14N10"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that Bousquet-Mélou–Schaeffer numbers, after dividing out explicit Pochhammer and linear factors, are polynomials in the partition parts for every stable genus and part count.","keywords":["Bousquet-Mélou–Schaeffer numbers","quasi-polynomiality","topological recursion","semi-infinite wedge","A-operators","Hurwitz numbers","spectral curve","enumerative combinatorics"],"falsifier":"For a small stable case such as $m=2$, $g=1$, $n=2$, compute the connected correlator $[\\hbar^{2g-2+n}]\\langle A(\\mu_1,\\hbar)A(\\mu_2,\\hbar)\\rangle^\\circ$ directly from the A-operator expansion of Section 5, and check the residues at $\\mu_1=-r$ for positive integers $r$: the theorem predicts zero, while any nonzero residue produces a pole at a negative integer and contradicts formula (134). Equivalently, evaluate the ratio in (12) for many integer pairs $(\\mu_1,\\mu_2)$ and verify that it is a polynomial of degree at most $6(2g-2+n)=12$; a persistent non-polynomial remainder would falsify the theorem.","tokens_in":30357,"feed_emoji":"🧮","tokens_out":12059,"duration_ms":109387,"temperature":0.7,"pith_summary":"The paper proves a structural quasi-polynomiality statement for the Bousquet-Mélou–Schaeffer (BMS) numbers, which count factorizations of a permutation of given cyclic type into m permutations with a fixed total number of cycles, or equivalently ramified coverings of the sphere. Theorem 6.1 states that for every stable pair (g, n) the connected BMS number $b^\\circ_{g,\\mu}$ equals the product of an explicit Pochhammer factor $\\prod_i (m\\mu_i - m)!/(\\mu_i!(m\\mu_i-\\mu_i-1)!)$, a finite product of linear factors $\\prod_{m\\le j\\le 4g-4+2n-1,\\ m\\nmid j}((\\mu_i - j)/m)$, and a polynomial $\\mathrm{Poly}_{g,n}(\\mu_1,\\dots,\\mu_n)$. This is exactly the shape that topological recursion, specialized to the spectral curve $x=(1+z)^m/z$, would force; the paper gives a purely combinatorial proof that does not rely on the still-missing cohomological-field-theory extension at higher-order critical points. The proof works in the semi-infinite wedge formalism, where BMS numbers appear as vacuum expectations of A-operators, and the main step is to show that all residues at negative integers cancel. As a corollary, the n-point generating functions of BMS numbers lie in finite-dimensional spaces spanned by the $\\xi$-functions and their derivatives on the spectral curve.","feed_headline":"BMS numbers turn into polynomials after explicit factors","feed_subtitle":"A purely combinatorial proof establishes the structural formula predicted by topological recursion.","key_machinery":"The load-bearing object is the family of A-operators acting on the charge-zero sector of the semi-infinite wedge space. With $\\check{A}(k,\\hbar) = \\hbar^{-k} e^{\\alpha_1}D(\\hbar)^m \\frac{\\alpha_{-k}}{k} D(\\hbar)^{-m}e^{-\\alpha_1}$ and its rescaling $A(k,\\hbar) = \\bigl(\\frac{(mk-m)!}{k!(mk-k-1)!}\\bigr)^{-1}\\check{A}(k,\\hbar)$, equation (133) turns BMS numbers into coefficient extractions of connected correlators $\\langle\\prod_i A(\\mu_i,\\hbar)\\rangle^\\circ$. The key structural input is the quasi-rationality of these operators' coefficients: the off-diagonal part $[\\hbar^{q+p}][E_{l-q,l}]A(k,\\hbar)$ has denominator $(k+1)\\cdots(k+q)\\prod_{m\\le j\\le 2p-1,\\ m\\nmid j}((k-j)/m)$ with a polynomial numerator of degree at most $6p+q$, and the identity part has a similar rational form whose numerator is divisible by $k^2(mk-k+1)$. These forms come from writing the finite-difference operator $\\Delta^t(P_k^m)$ in the basis of falling factorials $(k)_{s_1}\\cdots(k)_{s_m}$, with the auxiliary polynomials $R_p(k,l,i_1,\\dots,i_m)$ carrying the polynomiality and divisibility needed for cancellations. Proposition 5.20 then identifies the residue of $A(k,\\hbar)$ at $k=-r$ with the adjoint operator $\\check{A}^\\dagger(r,\\hbar)$, and the Heisenberg relation $[\\alpha_k,\\alpha_l]=k\\delta_{k+l,0}$ forces these residues to cancel in stable connected correlators. In a second register, the $\\xi$-functions $\\xi_i = z^i/((1+z)^{m-1}(-1+(m-1)z))$ on the spectral curve $x=(1+z)^m/z$ generate finite-dimensional spaces $\\Xi_d$ whose $X$-expansions realize the polynomial prefactors via Proposition 3.1, converting the theorem into the generating-function statement of Corollary 6.2.","core_discovery":"The central result is Theorem 6.1. For $(g,n)\\notin\\{(0,1),(0,2)\\}$, the connected BMS number equals\n$$b^\\circ_{g,\\mu} = \\prod_{i=1}^n \\frac{(m\\mu_i-m)!}{\\mu_i!(m\\mu_i-\\mu_i-1)!}\\cdot \\mathrm{Poly}_{g,n}(\\mu_1,\\dots,\\mu_n)\\cdot \\prod_{i=1}^n \\prod_{\\substack{m\\le j_i\\le 4g-4+2n-1\\\\ m\\nmid j_i}} \\frac{\\mu_i-j_i}{m},$$\nwith $\\mathrm{Poly}_{g,n}$ a polynomial. The first factor is the Bessel-type weight already visible in the genus-zero formula of Bousquet-Mélou and Schaeffer; the second product encodes the finite collection of 'forbidden' parts where the denominator would vanish, and its range depends only on $g$ and $n$. The theorem is proved by writing $b^\\circ_{g,\\mu}$ as a coefficient extraction $[\\hbar^{2g-2+n}]$ of the connected vacuum expectation of $n$ rescaled A-operators in the semi-infinite wedge space. The proof establishes explicit rational forms for the A-operator coefficients, with polynomial numerators of bounded degree and denominators of the type $(k+1)\\cdots(k+q)\\prod((k-j)/m)$, and then shows that residues at all negative integers $k=-r$ vanish in stable connected correlators via the adjoint $\\check{A}^\\dagger$-operators and the free-fermion commutation relations. Only the allowed poles at $j/m$ remain, and symmetry in the $\\mu_i$ promotes the rational function to a polynomial. The paper also derives (Proposition 2.1) the Bouchard–Eynard topological recursion for BMS numbers on the spectral curve $x=(1+z)^m/z$, completing the Alexandrov–Chapuy–Eynard–Harnad theorem at the discriminant by an analytic-continuation argument.","pith_inferences":["The A-operator expansion is explicit enough to compute the low-genus polynomials $\\mathrm{Poly}_{g,n}$ symbolically; comparing such a computation for $m=2$ against the topological-recursion expansion would directly test the conjectured equivalence of the two descriptions.","The same quasi-rationality plus residue-cancellation scheme should apply to any hypergeometric $\\tau$-function in the Orlov–Scherbin family, predicting analogous quasi-polynomiality formulas for the associated weighted Hurwitz numbers, with the forbidden set determined by the degree of the spectral curve.","The sharp cutoff $4g-4+2n-1$ in the linear factors is a numerical footprint of the conjectural cohomological field theory: if an ELSV-type formula is eventually found, the upper limit should coincide with the top degree of the relevant tautological intersection ring."],"forward_implications":["For every stable pair $(g,n)$, the connected BMS numbers are quasi-polynomial in the parts: after the explicit factors, only the finite product of linear terms $(\\mu_i-j)/m$ with $m\\le j\\le 4g-4+2n-1$ and $m\\nmid j$ remains, and everything else is a genuine polynomial $\\mathrm{Poly}_{g,n}$ of degree at most $6(2g-2+n)$ in each variable.","The $n$-point generating functions of BMS numbers become finite linear combinations of products of $\\xi$-functions and their $x$-derivatives on the spectral curve $x=(1+z)^m/z$, so the spectral curve emerges directly from the combinatorics (Corollary 6.2).","The BMS numbers satisfy Bouchard–Eynard topological recursion for the spectral curve with a higher-order critical point at the discriminant, completing the Alexandrov–Chapuy–Eynard–Harnad result by analytic continuation (Proposition 2.1).","If the conjectural generalization of the topological-recursion/Givental correspondence to higher-order critical points is supplied, Theorem 6.1 becomes the combinatorial verification of the ELSV-type formula for BMS numbers described in Remarks 2.2–2.4."],"supporting_citations":[{"why":"Supplies the original definition of BMS numbers and the genus-zero formula (3) whose overall shape Theorem 6.1 generalizes.","marker":"[7]"},{"why":"Its Theorem 1.1 gives topological recursion for weighted Hurwitz numbers away from the discriminant, which the paper extends to the BMS point by analytic continuation in Proposition 2.1.","marker":"[1]"},{"why":"Provides the Bouchard–Eynard formulation of topological recursion used throughout the paper's spectral-curve statements.","marker":"[4]"},{"why":"The direct combinatorial-predecessor paper whose A-operator and difference-operator method the present proof adapts to BMS numbers.","marker":"[30]"},{"why":"Together with [16], identifies the Givental/cohomological-field-theory formula with topological recursion; its higher-order-critical-point generalization is the conjectural motivation.","marker":"[11]"},{"why":"The topological-recursion/intersection-theory correspondence whose missing higher-order-critical-point version is discussed in Remarks 2.2–2.4.","marker":"[16]"},{"why":"Proves the $m=2$ case of the topological recursion for this type of enumeration, a precedent for Proposition 2.1.","marker":"[28]"}],"fun_headline_variants":["BMS numbers: explicit factors yield polynomiality","Polynomial formula for BMS numbers via topological recursion","BMS numbers become polynomials: combinatorial proof","Explicit polynomial structure for BMS numbers"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument leans on the unproven-in-detail claim (Propositions 5.11 and 5.13, with several parameter regimes deferred as 'almost completely analogous, only easier') that the coefficients of the A-operators have exactly these rational forms, with no extra poles or larger numerators in any omitted case; if any such exception existed, the cancellation that removes the negative-integer poles would fail and $\\mathrm{Poly}_{g,n}$ might not be a polynomial.","fun_headline_variants_meta":{"raw":{"variants":["BMS numbers: explicit factors yield polynomiality","Polynomial formula for BMS numbers via topological recursion","BMS numbers become polynomials: combinatorial proof","Explicit polynomial structure for BMS numbers"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000178,"raw_usage":{"total_tokens":1375,"prompt_tokens":1100,"completion_tokens":275,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":716,"completion_tokens_details":{"reasoning_tokens":217}},"tokens_in":716,"tokens_out":275,"duration_ms":3376,"temperature":1.0,"reasoning_tokens":217,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:49:58.883709+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For a small stable case such as $m=2$, $g=1$, $n=2$, compute the connected correlator $[\\hbar^{2g-2+n}]\\langle A(\\mu_1,\\hbar)A(\\mu_2,\\hbar)\\rangle^\\circ$ directly from the A-operator expansion of Section 5, and check the residues at $\\mu_1=-r$ for positive integers $r$: the theorem predicts zero, while any nonzero residue produces a pole at a negative integer and contradicts formula (134). Equivalently, evaluate the ratio in (12) for many integer pairs $(\\mu_1,\\mu_2)$ and verify that it is a polynomial of degree at most $6(2g-2+n)=12$; a persistent non-polynomial remainder would falsify the theorem.","supporting_citations":[{"cited_title":"Bousquet-M´ elou and G","cited_arxiv_id":null,"evidence_quote":"Supplies the original definition of BMS numbers and the genus-zero formula (3) whose overall shape Theorem 6.1 generalizes."},{"cited_title":"Weighted Hurwitz numbers and topological recursion","cited_arxiv_id":"1806.09738","evidence_quote":"Its Theorem 1.1 gives topological recursion for weighted Hurwitz numbers away from the discriminant, which the paper extends to the BMS point by analytic continuation in Proposition 2.1."},{"cited_title":"Bouchard and B","cited_arxiv_id":null,"evidence_quote":"Provides the Bouchard–Eynard formulation of topological recursion used throughout the paper's spectral-curve statements."},{"cited_title":"Kramer, D","cited_arxiv_id":null,"evidence_quote":"The direct combinatorial-predecessor paper whose A-operator and difference-operator method the present proof adapts to BMS numbers."},{"cited_title":"Dunin-Barkowski, N","cited_arxiv_id":null,"evidence_quote":"Together with [16], identifies the Givental/cohomological-field-theory formula with topological recursion; its higher-order-critical-point generalization is the conjectural motivation."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The topological-recursion/intersection-theory correspondence whose missing higher-order-critical-point version is discussed in Remarks 2.2–2.4."},{"cited_title":"Kazarian and P","cited_arxiv_id":null,"evidence_quote":"Proves the $m=2$ case of the topological recursion for this type of enumeration, a precedent for Proposition 2.1."}],"review_version":1}