{"id":"9de2a859-59bb-4043-846c-18648d8b767e","arxiv_id":"2608.04204","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The N-independent constants of the ABJM and ADHM Bethe potentials and of the ABJM twisted index are expressed as finite combinations of the constant map function A.","lead":"This paper finds exact formulas for the constant terms that were missing in large-N expansions of the ABJM and ADHM supersymmetric partition functions. The formulas are built from a known function A and are verified to very high numerical precision, completing a piece needed for precise AdS/CFT comparisons.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The all-order exactness of the closed-form constants rests on a guessed coefficient formula; numerical agreement at ~10^-19 cannot rule out exponentially small remainders, so the central claim is not yet proven exact.","rationale":"The paper's strongest evidence is genuinely strong: 17-20 digit agreement with published values at k=1,2,4, mirror symmetry consistency, and cross-checks through new predicted coefficients and the functional identity (75). None of this, however, establishes the exactness of the all-order formulas. The pattern-recognition step (Section III.B) explicitly deduces the general coefficient formula from a few fitted rationals, and the backward resummation (Appendix A) constructs an integral representation matching that guessed series by construction. The pointwise differences at 10^-19--10^-20 demonstrate agreement with the numerical data at the sampled values, but an exact identity requires ruling out any additional term that is smaller than the numerical precision at those points, such as an exponentially small-in-k remainder. The paper is transparent about leaving a first-principles derivation to future work, which is precisely why the reader's CONDITIONAL verdict is appropriate. The reader's weakest assumption identifies the same load-bearing issue: the numerical extraction and pattern recognition may identify a very accurate approximation rather than the exact analytic function. A second numerical campaign at λ=60 and at new integer k would settle whether the extraction is contaminated by finite-N or fixed-λ systematics; an independent analytic derivation would settle the all-order exactness. Since neither test has been performed, the concern remains live but does not warrant rejection given the strong cross-checks already in hand.","tokens_in":26440,"tokens_out":11585,"duration_ms":103364,"concrete_test":"Run the same Bethe-Ansatz constant extraction at a second 't Hooft coupling, λ=60 (e.g., N=181 to 1201, WorkingPrecision 300), where nonperturbative corrections are suppressed by ~10^-30, and at new integer levels k=3,5,6,7 (N=60k) not present in the λ=30 data set. If the extracted constants deviate from (29), (38), (53) by more than the estimated nonperturbative error, the closed forms are numerically falsified. To probe the all-order claim directly, independently derive the general coefficient formula, e.g. (40), from the exact large-k asymptotics of the BAE or from a Fermi-gas/topological-string computation; if any coefficient order fails to match, the guessed formula is only an approximation.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim — e.g. Eq. (29) and its A-function form (32) — is that the N-independent constants are exact finite linear combinations of the constant map function A. The formulas are obtained in three steps: (i) subtract known N-dependent terms from finite-N BAE data at λ=30 and fit the residual over the basis {k^2, log k, 1, k^{-2n}}; (ii) infer the all-order coefficient formula from the first few fitted rationals, with AI-assisted pattern recognition (Section III.B); (iii) run the large-k expansion backward to construct the integral representation (Appendix A) and then the A-function combination (Appendix B). Step (ii) is a conjecture: nothing rules out a different all-order formula that matches the low-order fitted rationals. The backward resummation guarantees the integral reproduces the guessed series by construction, so it supplies no independent evidence for the all-order claim. The pointwise checks at the level 10^-19–10^-20 test the combined formula at the sampled k=N/30, but cannot exclude a remainder that is exponentially small in k, e.g. O(e^{-c k}) with c large, or a term lying outside the fitting basis; both would be invisible at the quoted precision yet destroy the exactness of (29)/(38)/(53). The paper explicitly leaves a first-principles derivation to future work, confirming that the exact forms are not established.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper determines closed-form expressions for the N-independent constant terms in the all-order 1/N expansions of the ABJM Bethe potential, the ABJM topologically twisted index at the superconformal point, and the ADHM Bethe potential. The central results, e.g., Eqs. (29), (32), (38), (42), and (53), express these constants as finite linear combinations of the constant map function A(k). The method is numerical: high-precision Bethe-Ansatz solutions at fixed 't Hooft coupling, subtraction of known N-dependent terms, a LinearModelFit of the residual, pattern recognition of the tail coefficients (with AI assistance), and backward resummation into integral representations. The paper also uses factorization relations to determine the first two leading orders of the large-squashing expansion of the Airy constant, and it conjectures a new functional identity for A(k). Independent checks include 20-digit matches to previously published values, mirror-symmetry consistency at N_f=1, and predicted higher-order coefficients.","tokens_in":26776,"tokens_out":6667,"duration_ms":58263,"significance":"If the closed forms are correct, they fill a long-standing gap in the exact large-N description of M2-brane partition functions: the N-independent constants were previously available only numerically. The results are concrete, falsifiable, and supply exact special values and predicted coefficients that can be tested further. The paper is commendably transparent about its numerical provenance, and the use of independent anchors (published values, mirror symmetry, exact A-values at integer arguments) strengthens confidence. However, as the paper itself states, no first-principles derivation is provided, so the exactness of the closed forms is not established.","major_comments":[{"comment":"The all-order exactness of the closed forms (29), (32), (38), (42), (53), and (56) is not established. The general-order Bernoulli-number formulas are inferred from a finite set of fitted rationals, and the backward resummation in Appendix A reproduces the guessed series by construction, so it provides no independent evidence for the all-order claim. Numerical agreement at the 10^-17 to 10^-20 level cannot exclude an exponentially small remainder such as O(e^{-c k}) with large c, nor a term that vanishes beyond the quoted precision but is analytically nonzero; both would alter the claimed exact expressions. The paper explicitly leaves a first-principles derivation for future work (Section III.B.1 and the analogous paragraph in Section IV.B). I therefore request that the abstract and the text moderate the claim from 'determine in closed form' to 'conjecture with strong numerical evidence', or that a derivation be supplied.","section":"III.B, IV.B, Appendix A"},{"comment":"The constant f0 = -8ζ'(-1) - (5/2) log 2 - (2/3) log 4π is presented as part of the closed-form result (38), but it is not derived. The text explains that the fitted value is 'identified' within the basis {ζ'(-1), log 2, log 4π}, motivated by analogy with the -log 2/6 identification in (29). The claimed independent confirmation via Eq. (43) uses the A-function representation (42), whose derivation already incorporates the identification of f0; while the match with published values is a nontrivial consistency check, it does not prove that the fitted constant equals this combination. A closed-form claim for f0 should follow from the integral representation rather than from a numerical fit and a guess of the transcendental basis.","section":"III.B.2, Eq. (39)"}],"minor_comments":[{"comment":"The equality signs connecting the integral representation, the large-k asymptotic expansion, and the small-k expansion are misleading: the second and third expressions are asymptotic series valid in different regimes and are not equal as convergent series. Using '~' or '=' with an explicit qualifier would be clearer.","section":"Eq. (12)"},{"comment":"The pattern-recognition step is described as being performed with the help of an AI assistant, but no details of the prompts, the candidate families considered, or the selection criteria are provided. Since this step is the basis for the general-order coefficient formulas, some additional documentation would improve reproducibility.","section":"Section III.B.1"},{"comment":"The functional identity Φ(x) = -Φ(4/x) is a new mathematical statement, but it is only verified numerically (to 40 digits) and is not proved. The text should label it as a conjecture rather than a 'functional identity', unless a proof is included.","section":"Section V, Eq. (75)"},{"comment":"The phrase 'verify them down to the level of non-perturbative corrections' is slightly overstated: the residuals are at the level of 10^-19 to 10^-20, while the leading non-perturbative correction is estimated as ~10^-21 at the chosen 't Hooft coupling; the verification is close to but not strictly at that level.","section":"Abstract"}],"recommendation":"major_revision","confidential_remarks":"The paper is an example of 'experimental mathematics' in hep-th: the main formulas are supported by strong numerical evidence and independent checks, but they are not derived. Whether this meets the standard of the journal is a policy question. The authors are transparent about the conjectural status in the body, but the abstract overstates the result. The overlap with the independent works [56] and [80] should be monitored by the editor. I would advise that the revision clearly label the central results as conjectures and weaken the abstract accordingly."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The new stuff is real: Eq. (29)/(32) for the ABJM Bethe potential constant, Eq. (34) for the generic-flavor k^0 term, Eq. (53)/(55) for the ADHM Bethe potential constant, and the functional identity (75), plus the TTI closed forms (38)/(42). These fill an actual gap, since the previous literature had only numerical values. The cross-checks are strong: 17–20 digit agreement with BAE data, exact matches to previously published numbers in [13] and [30], and consistency with 3d mirror symmetry at N_f=1. The paper is also unusually transparent: it says in so many words that the closed forms are reconstructed from numerics, that the pattern-recognition step was AI-assisted, and that a first-principles derivation is left for future work. That candor earns credit.\n\nThe soft spot is the one the stress-test note identifies, and it is a genuine one. The exactness claim rests on inferring an all-order coefficient formula from the first few fitted rationals. Nothing rules out a different all-order formula that matches those low-order terms, or an exponentially small remainder invisible at 10^-19. Running the large-k expansion backward to get the integral representation is consistency, not independent evidence. So the formulas are best described as high-confidence conjectures, not established theorems. The paper does not overclaim on this point, but readers should not mistake the numerics for a proof.\n\nTwo smaller issues: the numerical code and data are not provided, which makes independent verification harder, and the overlap with the unpublished works [56] and [80] cannot be assessed. Neither is disqualifying; both are normal for a paper in this area, and the author flags them.\n\nFor a serious referee: yes. The results are important to the localization/holography community, the numerical evidence is unusually precise, and the central claim is plausibly correct even if unproven. A referee can ask for an independent derivation or at minimum release of the data, and the paper would be stronger for it. I would take it to a reading group and would cite it if I worked on these partition functions, though with a caveat that the constants are conjectural.","headline":"Highly plausible closed-form constants in ABJM/ADHM 1/N expansions, reconstructed from high-precision numerics rather than derived; the paper is honest about that, and it deserves peer review.","tokens_in":27246,"tokens_out":2054,"would_cite":true,"duration_ms":19983,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"M2-brane partition constants take exact closed form","keywords":["M2-branes","ABJM theory","ADHM theory","topologically twisted index","Bethe potential","constant map function","1/N expansion","Airy function"],"falsifier":"Compute the ABJM twisted index constant at an untested integer level, for example k=6, from Bethe-Ansatz data at fixed 't Hooft coupling with N up to about 500; the closed form predicts specific exact rational tail coefficients including f6 and f7. If subtracting the closed form leaves a stable power-law residual above the estimated non-perturbative scale of roughly $10^{-21}$, the identification is only numerically accurate, not exact.","tokens_in":26218,"feed_emoji":"🧮","tokens_out":7323,"duration_ms":66653,"temperature":0.7,"pith_summary":"This paper claims that the previously numerical, N-independent constant terms in the all-order 1/N expansions of the ABJM Bethe potential and topologically twisted index, and of the ADHM Bethe potential, are finite linear combinations of one special function, A(k), the constant map function already known from the round three-sphere partition function. The constants are reconstructed from high-precision Bethe-Ansatz numerics: after subtracting known N-dependent terms, the residual tails are fitted, shown to match a Bernoulli-number pattern, and resummed back into integral representations and A-combinations. If the identification is exact, the missing constant sector of these holographic partition functions is no longer numerical, and through factorization relations the squashed-sphere Airy constant is fixed in closed form through the first two leading orders of large squashing. The reader should care because this removes the last undetermined piece in a web of all-order localization results for M2-brane theories and their gravity duals.","feed_headline":"M2-brane partition constants take exact closed form","feed_subtitle":"The elusive N-independent pieces of ABJM and ADHM indices reduce to the constant map function A.","key_machinery":"The central object is the constant map function $A(k)$, defined by the integral in Eq. (12), which resums the all-genus constant-map contributions of the dual topological string. Its large-$k$ expansion has coefficients built from products of Bernoulli numbers, $|B_{2n}B_{2n+2}|$, and this signature is what lets the author recognize the same structure in the numerical tails of the Bethe potential and the index. Running that expansion backward resums the guessed general-order series into integral representations, and elementary integral identities convert those integrals into finite linear combinations of A-functions. The factorization relations of [32] are the second load-bearing mechanism, carrying the Bethe and index constants into the squashed-sphere Airy constant.","core_discovery":"At the superconformal point, the ABJM Bethe potential constant is given by $\\hat g_0(k,\\Delta_{\\rm sc}) = A(k)-A(k/2)+\\tfrac{k}{2}A(4/k)-\\tfrac{k}{4}A(8/k)-\\frac{\\zeta(3)}{8\\pi^2 k^2}$, and the ABJM topologically twisted index constant is another finite A-combination, linearly related to $\\hat g_0$ with an explicit $\\log k$ term. The ADHM Bethe potential constant satisfies the mirror-symmetry-preserving relation $\\hat g_0^{\\rm ADHM}(N_f)=\\tfrac12\\hat g_0(N_f)+\\tfrac14\\hat g_0(2N_f)+\\cdots$, which at $N_f=1$ equals the ABJM value at $k=1$; its leading coefficient requires the non-elementary value $A(1/2)$. The paper reaches these formulas by fitting residual data at fixed 't Hooft coupling, recognizing exact rational tail coefficients, resumming the large-$k$ series into integrals, and identifying the pure constant within the transcendental basis $\\{\\zeta'(-1),\\log 2,\\log 4\\pi\\}$. Checks at $k=1,2,4$ reproduce known 20-digit values, and the remaining residuals sit at the expected non-perturbative scale.","pith_inferences":["The same A-combination pattern is a natural template for the other M2-brane SCFTs listed in the paper, and repeating the high-precision fit for those theories is a direct, concrete test of whether the structure is universal.","If the closed forms are exact, gravity-side derivations that currently reproduce only the Airy data now have a precise target: they must generate A-combinations with the direct and inverted arguments $\\{K,K/2,2K;2/K,4/K,8/K\\}$, giving a sharper constraint on quantum M-theory localization.","The functional identity $\\Phi(x)=-\\Phi(4/x)$ appears to be a self-contained property of $A$; a direct proof from the integral representation would close the gap the paper leaves open and may explain why those particular arguments appear.","The partial result at generic flavor chemical potentials suggests, but does not prove, that the full flavor-dependent constant is again a finite A-combination with arguments shifted by the $\\Delta_a$; computing the next order in $\\Delta$ would settle the conjecture."],"forward_implications":["The closed forms for the ABJM Bethe potential and twisted index constants replace previous numerical fits and give exact values at integer levels such as $k=1,2,4$.","Through the Cardy-like relation (7), the new constants transfer directly to the first two orders of the Cardy expansion of the superconformal index.","Through factorization, the Airy constant of the squashed three-sphere is fixed in closed form through the two leading orders in large squashing, reproducing the planar $k^2$ coefficient and the universal $\\log k$ coefficient $-1/6$.","The ADHM and ABJM constants obey the 3d mirror symmetry constraint $\\hat g_0^{\\rm ADHM}(1)=\\hat g_0(1)$, and the ADHM result correctly involves $A(1/2)$ rather than only integer arguments.","The web of exact anchors yields the functional identity $A(x)+A(2-x)=-(A(4/x)+A(2-4/x))$, verified numerically to 40 digits and derived on an infinite discrete set of squashing values."],"supporting_citations":[{"why":"Supplies the Fermi-gas derivation of the constant map function A used throughout.","marker":"[19]"},{"why":"Provides exact values of A at integer arguments, used to verify the closed forms.","marker":"[46]"},{"why":"Provides the all-order 1/N expansion of the ABJM twisted index, the numerical Bethe-Ansatz method, and the prior numerical constant f0.","marker":"[13]"},{"why":"Provides the all-order Bethe potential expansion, the Cardy-like relation to the superconformal index, and the prior numerical value at k=1.","marker":"[30]"},{"why":"Provides the all-order expansions of the ADHM Bethe potential and twisted index and the numerical baseline for ADHM.","marker":"[14]"},{"why":"Provides the factorization relations that convert the Bethe and index constants into the squashed-sphere Airy constant.","marker":"[32]"},{"why":"Supplies the Airy-form conjecture, special-case values, and the planar and large-k checks used to test the results.","marker":"[15]"}],"fun_headline_variants":["Closed forms for M2-brane index constants","Exact constants in ABJM and ADHM indices","M2-brane constants pinned via mirror symmetry","Partition function constants solved exactly","M2-brane constants reduced to constant map A"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument stands on the premise that the finite fitting basis and the guessed general-order Bernoulli pattern recover the exact analytic function rather than an extremely accurate asymptotic approximation; the paper explicitly leaves a first-principles derivation for future work.","fun_headline_variants_meta":{"raw":{"variants":["Closed forms for M2-brane index constants","Exact constants in ABJM and ADHM indices","M2-brane constants pinned via mirror symmetry","Partition function constants solved exactly","M2-brane constants reduced to constant map A"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000257,"raw_usage":{"total_tokens":1618,"prompt_tokens":1023,"completion_tokens":595,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":639,"completion_tokens_details":{"reasoning_tokens":525}},"tokens_in":639,"tokens_out":595,"duration_ms":5367,"temperature":1.0,"reasoning_tokens":525,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T14:40:50.937853+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the ABJM twisted index constant at an untested integer level, for example k=6, from Bethe-Ansatz data at fixed 't Hooft coupling with N up to about 500; the closed form predicts specific exact rational tail coefficients including f6 and f7. If subtracting the closed form leaves a stable power-law residual above the estimated non-perturbative scale of roughly $10^{-21}$, the identification is only numerically accurate, not exact.","supporting_citations":[],"review_version":2}