{"id":"08432f96-378f-47a9-b875-d52d6b2e6c22","arxiv_id":"1908.08660","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper proves the inequality M_{2k,N}(n) > N_{2k,N}(n) for finite analogues of partition rank and crank moments for all n ≥ 1, N ≥ 1, and k ≥ 1.","lead":"This paper proves that a finite, size-bounded version of the partition crank moment always exceeds the corresponding rank moment, confirming a conjecture from the authors' earlier work. It extends Garvan's symmetrized-moment technique using a new Bailey-pair identity.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified; the flagged limit identity (3.7) is true and the double limit is well-defined.","rationale":"The theorem is proven by establishing a nonnegative-coefficient generating function for the difference of finite symmetrized crank and rank moments (Theorem 2.5) and then converting to ordinary moments via Garvan's Stirling-type numbers S^*(k,j). I traced the dependence chain: Theorem 2.5 ← Corollary 5.4 ← Proposition 5.1 ← Bailey's lemma + identity (3.7). The only unproved input is (3.7) and the associated division by (1-ρ1)(1-ρ2). I verified the identity by expanding ρ1=1/(1-u) and ρ2=1/(1-v): the product over j=1..k of (1-q^j)(1-abq^j)/((1-aq^j)(1-bq^j)) is ∏(1 - uv q^j/(1-q^j)^2 / ((1+su)(1+sv))), so (1-P)/(uv) tends to ∑ q^j/(1-q^j)^2. The induction step in Proposition 5.1 has the n1=0 term with an extra (q)_{n2} denominator, which makes S-L0 vanish to first order, so the double limit is not a 0/0 ambiguity. The remaining algebra in Proposition 5.1 (substituting α' and β' from Bailey's lemma and applying the induction hypothesis to (α',β')) is routine and consistent. The positivity argument then works: each coefficient of Theorem 2.5's right-hand side is nonnegative because (q^{n1+1})_{N-n1}^{-1} has nonnegative coefficients; S^*(k,j) > 0; and the j=1 term gives 2 spt(n,N) > 0. No fatal or material gap was found. The flagged step is a minor exposition issue.","tokens_in":14051,"tokens_out":31026,"duration_ms":251663,"concrete_test":"Verify Proposition 5.1 for N=2 and k=1,2 with β_n=1/(q)_n, keeping ρ1 and ρ2 symbolic: after clearing denominators and taking the double limit using an independent derivation of (3.7) (expanding u=1-ρ1 and v=1-ρ2 through order uv), check that the coefficient of every monomial q^m in the two sides matches.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim rests on Theorem 2.5, whose proof uses Proposition 5.1. The delicate step is the double limit in Prop 5.1, which invokes identity (3.7) without proof. I checked this step. Writing a=1/ρ1=1-u and b=1/ρ2=1-v, each factor (1-q^j)(1-ab q^j)/((1-aq^j)(1-bq^j)) equals 1 - uv q^j/(1-q^j)^2 + O(u^2v, uv^2); hence (1-P)/(uv) tends to ∑_{j=1}^k q^j/(1-q^j)^2, so (3.7) is correct. In the induction step of Prop 5.1, the n1=0 term carries an extra (q)_{n2} in the denominator, which makes the subtraction S-L0 vanish linearly in (1-ρ1)(1-ρ2); the limit is therefore well-defined and no pole remains. The paper would benefit from proving or citing (3.7), but the gap is expository, not substantive. No blocking objection.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves Conjecture 1.2 from the authors' earlier work [15]: for every fixed natural number N and every even k ≥ 2, the finite analogue of the crank moment exceeds the finite analogue of the rank moment: M_{k,N}(n) > N_{k,N}(n) for all n ≥ 1. To do this, the authors define finite analogues of symmetrized rank and crank moments η_{k,N}(n) and μ_{k,N}(n), derive their generating functions (Theorems 2.2 and 2.4), and then prove a nonnegative-coefficient generating function for the difference μ_{2k,N}(n) − η_{2k,N}(n) (Theorem 2.5). The proof of Theorem 2.5 is built on an induction using Bailey's lemma (Proposition 5.1) together with two classical Bailey pairs. The final step links ordinary finite moments to symmetrized moments through Garvan's numbers S*(k,j), yielding the stronger bound M_{2k,N}(n) − N_{2k,N}(n) ≥ 2 spt(n,N) > 0. The paper is a direct finite analogue of Garvan's proof of the classical rank-crank moment inequality.","tokens_in":14295,"tokens_out":7547,"duration_ms":74179,"significance":"If the proof is accepted, the paper settles a conjecture from [15] and provides a finite/restricted analogue of Garvan's theorem, extending a well-known circle of ideas in partition theory. The main technical contribution, Theorem 2.5, is a finite analogue of Garvan's difference generating function and appears to be new. The arguments are explicit q-series manipulations: Theorems 2.2 and 2.4 are derived from known partial-fraction identities, and Proposition 5.1 is a genuine application of Bailey's lemma. The positivity conclusion is clean and yields a stronger statement than the conjecture itself, namely the lower bound in terms of spt(n,N). The main weakness is that a key limiting identity, Eq. (3.7), is quoted without proof; I verified it independently and found it correct, but the manuscript should supply a derivation or reference. There are also several small typographical errors. Overall, the result is worthy of publication once the missing justification is added.","major_comments":[{"comment":"The limit identity (3.7) is stated without proof or reference and is used in a load-bearing way in both the base case and the induction step of Proposition 5.1, specifically when the proof divides by (1−ρ1)(1−ρ2) and then lets ρ1,ρ2 → 1. The identity is true: writing a = 1/ρ1 = 1−u and b = 1/ρ2 = 1−v, each factor (1−q^j)(1−ab q^j)/((1−a q^j)(1−b q^j)) expands as 1 − uv q^j/(1−q^j)^2 + O(u^2 v, u v^2), which gives (3.7). However, the manuscript should provide this derivation or cite a source, and it should comment on the double limit at the points where the denominators (q/ρ1)_n(q/ρ2)_n vanish. As written, the proof of Proposition 5.1 is incomplete at a critical step, even though the gap is readily fixable.","section":"Section 3, Eq. (3.7)"}],"minor_comments":[{"comment":"The displayed generating function in the proof of Conjecture 1.2 reads (μ_{2t,N}(n)q^n − η_{2t,N}(n))q^n, which appears to be a typo; it should be (μ_{2t,N}(n) − η_{2t,N}(n))q^n.","section":"Section 5, proof of Conjecture 1.2"},{"comment":"The sentence 'Divide both sides of (5.4) by (q)_N' refers to the unnumbered display in Corollary 5.4, not to equation (5.4) in Proposition 5.5. Please add a cross-reference or number the display in Corollary 5.4.","section":"Section 5, proof of Theorem 2.5"},{"comment":"In the sentence 'Substituting the above Bailey pair in Theorem 5.1', the reference should be to Proposition 5.1, not Theorem 5.1.","section":"Section 5, Corollary 5.2"},{"comment":"There are several typographical slips, including 'conjectur ed' and 'c rank' in the abstract, and 'Garvan himself' in the first sentence; these should be corrected in a final revision.","section":"Abstract and Introduction"},{"comment":"When the proof says 'letting ρ1 → 1, ρ2 → 1 and using (3.7)', it may help the reader to note explicitly that all sums are finite (since the indices are bounded by N), so the interchange of limit and summation is immediate.","section":"Section 5, proof of Proposition 5.1"}],"recommendation":"major_revision","confidential_remarks":"The manuscript relies on the authors' earlier paper [15] for definitions and for the conjecture, which is appropriate in this context. The only substantive issue is the unproved identity (3.7); I checked it and it is correct, so the revision should be straightforward. After the proof or citation is added and the minor typographical errors are fixed, the paper would be suitable for publication. No concerns about novelty or scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The conjecture was open, and this paper settles it. The main new ingredients are finite analogues of symmetrized rank and crank moments, explicit generating functions for them, and a Bailey-pair lemma (Proposition 5.1) that is used to produce a manifestly nonnegative generating function for the difference. That last step is the load-bearing part, and it holds up. I checked the flagged double limit in Proposition 5.1: identity (3.7) is correct, as a formal expansion in u = 1 - rho_1 and v = 1 - rho_2 shows, and the n_1 = 0 term vanishes linearly so no pole survives. The paper would be easier to referee if (3.7) were proved or cited rather than quoted, but this is an expository gap, not a substantive one.\n\nThe proof follows Garvan's template closely, as the authors say, but that is fair: the finite analogues are genuinely new, the generating function identities are new, and the conclusion is stronger than the original conjecture (the difference is at least 2 spt(n,N), which is positive). The reliance on the authors' earlier paper [15] for definitions and the conjecture is legitimate background, not circularity. The main theorem is proved from explicit q-series manipulations, and I did not find any place where the conjecture was assumed.\n\nSoft spots are minor. The text is very dense; a non-specialist will struggle to follow the index juggling in Section 5, and a few more signposts would help. There is a typo in the display in the proof of Conjecture 1.2 (an extra q^n appears in the summand), and a couple of other harmless slips. The concluding remarks suggest asymptotics and congruences as future work, which is appropriate rather than a deficiency.\n\nThis paper is for partition theorists and q-series specialists. It will not reorganize the field, but it proves a natural conjecture with explicit, checkable identities. It deserves a serious referee. My recommendation: send to a specialist, ask for a proof or citation for (3.7) and a fix of the typos, and accept after minor revision.","headline":"A careful, dense proof of a natural open conjecture in partition theory; the machinery is borrowed but the result is new and the argument checks out.","tokens_in":14800,"tokens_out":1861,"would_cite":true,"duration_ms":20980,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11P80","11P81","11P82","05A17"],"pacs":[],"model":"deepseek-v4-flash","headline":"For every size bound $N$ and every even $k$, the finite crank moment $M_{k,N}(n)$ strictly exceeds the finite rank moment $N_{k,N}(n)$ for all $n \\geq 1$.","keywords":["partitions","finite analogues","rank moments","crank moments","symmetrized moments","smallest parts function","moments inequality","q-series identities"],"falsifier":"Compute the finite moments directly from the definitions (1.9) and (1.10) for small values, say $N=2$, $k=2$, and $n=1,\\dots,10$, and compare with the claim; a single $n$ with $M_{2,N}(n) \\leq N_{2,N}(n)$ would refute the theorem. Alternatively, expand the right-hand side of (2.9) as a $q$-series to a chosen order and check that no coefficient is negative — the proof asserts none can be.","tokens_in":13870,"feed_emoji":"🧮","tokens_out":15949,"duration_ms":118070,"temperature":0.7,"pith_summary":"Working with finite analogues of partition rank and crank statistics — weighted counts over vector partitions whose parts are restricted by a bound $N$ — the paper proves that the even-order moments always satisfy $M_{k,N}(n) > N_{k,N}(n)$ for all $n \\geq 1$ and fixed $N$. This is the finite version of a classical inequality between crank and rank moments, and it was previously left as a conjecture in the paper that introduced these finite analogues. The proof derives an explicit generating function for the difference of the finite symmetrized moments whose coefficients are manifestly nonnegative, and then uses a positive change of basis to pass from symmetrized to ordinary moments. A byproduct is the stronger estimate $M_{2k,N}(n) - N_{2k,N}(n) \\geq 2\\,\\mathrm{spt}(n,N)$, where $\\mathrm{spt}(n,N)$ is the finite smallest-parts count.","feed_headline":"Finite crank moments beat rank moments for every size N","feed_subtitle":"A nonnegative-coefficient generating function for the gap settles the finite analogue of the rank-crank inequality.","key_machinery":"The load-bearing object is the generating function identity of Theorem 2.5 for $\\mu_{2k,N}(n) - \\eta_{2k,N}(n)$, the difference of the finite symmetrized crank and rank moments. It expresses that difference as a finite sum over non-increasing sequences $N \\geq n_k \\geq \\cdots \\geq n_1 \\geq 1$ of terms with manifestly nonnegative coefficients, so the difference is a nonnegative integer for every $n$. The second ingredient is the family of polynomials $g_j(m)=\\prod_{i=0}^{j-1}(m^2-i^2)$ together with the positive integers $S^*(k,j)$, which rewrite ordinary $2k$-th moments as positive linear combinations of symmetrized moments. These two parts turn coefficient nonnegativity into the strict moment inequality.","core_discovery":"The central discovery is the identity of Theorem 2.5: the generating function for the difference of the finite symmetrized crank and rank moments equals a single chain sum $$\\sum_{N \\geq n_k \\geq \\cdots \\geq n_1 \\geq 1} \\frac{$q^{{n_1+\\cdots+n_k}}$}{(1-$q^{{n_1}}$)^2 \\cdots (1-$q^{{n_k}}$)^2 ($q^{{n_1+1}}$;q)_{N-n_1}},$$ which expands as a power series with nonnegative coefficients. Hence each coefficient $\\mu_{2k,N}(n) - \\eta_{2k,N}(n)$ is nonnegative. Using the finite analogue of the classical relations between moments and symmetrized moments — expressed through the positive integers $S^*(k,j)$ — the paper converts this coefficient nonnegativity into the strict inequality $M_{2k,N}(n) > N_{2k,N}(n)$ for every even order, and indeed shows the gap is at least $2\\,\\mathrm{spt}(n,N)$.","pith_inferences":["The same positivity mechanism may yield a stronger lower bound for $M_{2k,N}(n)-N_{2k,N}(n)$ in terms of higher-order finite spt-functions, since the proof only uses the $j=1$ term of the symmetrized expansion.","A natural test is to compute the ratio $(M_{2k,N}(n)-N_{2k,N}(n))/(2\\,\\mathrm{spt}(n,N))$ for small $N,k,n$; the paper's proof forces it to be at least $1$, and numerical values would show how sharp that bound is.","If the double limit in Proposition 5.1 were justified as a formal power series identity in $q$ for all $N$ and $k$, the proof would hold without any convergence assumptions, and the same style of argument could be tried on other pairings of partition statistics.","The chain sum in Theorem 2.5 resembles a count of marked chains and may admit a direct combinatorial interpretation for $\\mathrm{spt}_k(n,N)$, mirroring the classical interpretation of higher-order spt-functions."],"forward_implications":["For each fixed $N$, the finite gap $M_{2k,N}(n) - N_{2k,N}(n)$ is at least $2\\,\\mathrm{spt}(n,N)$, so the difference grows at least as fast as the finite smallest-parts function in $n$.","The finite higher-order spt-function $\\mathrm{spt}_k(n,N) := \\mu_{2k,N}(n) - \\eta_{2k,N}(n)$ is nonnegative for all $k,n,N$, and equals $\\mathrm{spt}(n,N)$ when $k=1$.","Letting $N \\to \\infty$ in the chain-sum identity recovers the classical identity for the difference of symmetrized crank and rank moments, and hence the classical inequality between the $2k$-th crank and rank moments.","Because the coefficients $S^*(k,j)$ are positive, nonnegativity of the symmetrized differences implies nonnegativity of the ordinary moment differences for every even $k$, not just for the smallest order."],"supporting_citations":[{"why":"supplies the definitions of the finite rank and crank moments, the finite spt identity, and the conjecture that the paper proves.","marker":"[15]"},{"why":"provides the generating function for symmetrized crank moments and the positive basis numbers S^*(k,j), and proves the classical inequality being imitated.","marker":"[21]"},{"why":"introduces the symmetrized rank moments whose finite analogue is defined here and gives their generating function.","marker":"[3]"},{"why":"supplies the two explicit hypergeometric pairs used to evaluate the sums in Corollaries 5.2 and 5.4.","marker":"[1]"},{"why":"supplies the identity np(n)=M_2(n)/2 used to identify the k=1 case with the finite smallest-parts function.","marker":"[17]"},{"why":"defines the smallest parts function spt(n) whose finite analogue spt(n,N) appears in the lower bound.","marker":"[4]"}],"fun_headline_variants":["Crank moments dominate rank moments for all finite sizes","Generating function proof: finite crank moments exceed rank moments","Nonnegative coefficients prove crank moments dominate rank","Chain sum shows finite crank moments beat rank for all N","Inequality proved: finite crank moments always larger than rank"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof depends on a double limit in Proposition 5.1 where the expressions are divided by $(1-\\rho_1)(1-\\rho_2)$ and then $\\rho_1, \\rho_2$ are taken to $1$ inside factors that vanish at that limit; the paper invokes a limiting identity for this step without supplying a full formal justification, so if that limit is not valid the nonnegative-coefficient representation is not established.","fun_headline_variants_meta":{"raw":{"variants":["Crank moments dominate rank moments for all finite sizes","Generating function proof: finite crank moments exceed rank moments","Nonnegative coefficients prove crank moments dominate rank","Chain sum shows finite crank moments beat rank for all N","Inequality proved: finite crank moments always larger than rank"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000733,"raw_usage":{"total_tokens":3220,"prompt_tokens":828,"completion_tokens":2392,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":444,"completion_tokens_details":{"reasoning_tokens":2314}},"tokens_in":444,"tokens_out":2392,"duration_ms":14418,"temperature":1.0,"reasoning_tokens":2314,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:32:58.375895+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the finite moments directly from the definitions (1.9) and (1.10) for small values, say $N=2$, $k=2$, and $n=1,\\dots,10$, and compare with the claim; a single $n$ with $M_{2,N}(n) \\leq N_{2,N}(n)$ would refute the theorem. Alternatively, expand the right-hand side of (2.9) as a $q$-series to a chosen order and check that no coefficient is negative — the proof asserts none can be.","supporting_citations":[{"cited_title":"Untrodden pathways in the theory of the restricted partition function $p(n, N)$","cited_arxiv_id":"1812.01424","evidence_quote":"supplies the definitions of the finite rank and crank moments, the finite spt identity, and the conjecture that the paper proves."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"provides the generating function for symmetrized crank moments and the positive basis numbers S^*(k,j), and proves the classical inequality being imitated."},{"cited_title":"Andrews, Partitions, Durfee symbols, and the Atkin-Garvan moments o f ranks , Invent","cited_arxiv_id":null,"evidence_quote":"introduces the symmetrized rank moments whose finite analogue is defined here and gives their generating function."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"supplies the two explicit hypergeometric pairs used to evaluate the sums in Corollaries 5.2 and 5.4."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"supplies the identity np(n)=M_2(n)/2 used to identify the k=1 case with the finite smallest-parts function."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"defines the smallest parts function spt(n) whose finite analogue spt(n,N) appears in the lower bound."}],"review_version":1}