{"id":"d18beed2-0072-4ab5-8d2d-b7f0bd8b1583","arxiv_id":"1908.07720","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Theorem 1 proves Suzuki's local integral (2) equals L(π(n) × τ(n), ns − (n−1)/2) for all r<nm, and Section 5 gives conditional global integral representations.","lead":"Local integrals on metaplectic covers of GL(r) are shown to compute tensor-product L-functions in full generality, extending known cases. The paper also sketches global integral constructions, most of which depend on an open conjecture about representations on covering groups.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1's proof is not self-contained: the Section 4 identity connecting I to Suzuki's integral (2) is asserted by analogy with [G1] without verifying hypotheses for general tau(n), and the final evaluation (17) relies on two arXiv preprints.","rationale":"The manuscript is a plausible extension of prior work and no explicit mathematical error was found. However, the proof of Theorem 1 is incomplete in exactly the way the reader identifies: the Section 4 identity connecting I to Suzuki's integral is asserted by analogy with [G1] without verifying that the argument applies to the general induced representation Theta(tau(n)), and the final evaluation of the one-dimensional integrals is delegated to two arXiv preprints. I focused on the Section 4 identity as the single most load-bearing step because it is the only step that relates the auxiliary integral to the theorem's stated object. The proposed concrete test re-derives this identity in a case not covered by the quoted range of [G1], which would settle the concern. Since the reader's CONDITIONAL verdict already reflects this incompleteness, no change to that verdict is needed.","tokens_in":15034,"tokens_out":16518,"duration_ms":154816,"concrete_test":"Independently prove the Section 4 inner integral identity for the unproved case r=2, n=2, m=4 (so nm=8, nrm=16, m=2n, outside Suzuki's range): expand W^{(2)}_{tau(2),16} via equation (11) and compute the V_2-Fourier coefficient explicitly, verifying the right-hand side W^{(2)}_{tau(2),8}(diag(g,I_6)) |g|^{7/2}. If the identity changes by a power of |g| or the argument of W differs, the equality between I and integral (2) is wrong; if new hypotheses are needed, Theorem 1 is at best conditional.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim (Theorem 1) equates Suzuki's integral (2) with a local L-function. The proof uses an auxiliary integral I (eq. 4) computed in two ways. Section 3 computes I as the L-factor; Section 4 asserts that I equals integral (2). The load-bearing step is the identity claimed just after (18): the integral over V_r of W^{(n)}_{tau(n),nrm}(diag(vg, I_{r(nm-1)})) psi^{-1}(v) equals W^{(n)}_{tau(n),nm}(diag(g, I_{nm-r})) times |g|^{(nm-1)(r-1)/2}. This is the only bridge from I to the original Suzuki integral. The paper says it follows by 'a similar proof as the proof of Theorem 2 in [G1]' with 'the only different ... replacing n there with nm', but [G1]'s Theorem 2 was proved in the m=1/theta setting; it is not demonstrated that the proof survives for general induced representations Theta(tau(n)) with this constant. If this identity fails, the equality I = integral (2) is unsupported, even if Section 3 is correct. The same Section also omits the reduction (13)-(14). The later identification (17) of the one-dimensional integrals with L-factors depends on Theorem 46 in [K] and Theorem 8.1 in [C2], both arXiv preprints, whose precise statements and hypotheses are not reproduced.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims to prove Theorem 1: for a local non-archimedean field containing the n-th roots of unity, unramified representations π^(n) of GL_r^(n) and τ^(n) of GL_m^(n), and r < nm, Suzuki's local integral (2) equals the local unramified tensor-product L-function L(π^(n) × τ^(n), ns − (n−1)/2). The proof introduces a bi-K_r-invariant generating function W^(n)_{τ^(n),nrm} attached to an induced theta-type representation Θ(τ^(n)) and defines an auxiliary integral I in equation (4). Section 3 computes I through the unramified vector as the product of local L-factors, while Section 4 computes I through a Whittaker function and identifies it with Suzuki's integral (2). The paper also contains conditional global constructions in Section 5, including the Suzuki conjecture, qualitative results on the partial L-function, and a new global doubling integral for n = 1. For n = 1 the computation recovers the classical Jacquet--Piatetski-Shapiro--Shalika result. If correct, Theorem 1 proves Suzuki's local integral conjecture in full generality.","tokens_in":15415,"tokens_out":6203,"duration_ms":57479,"significance":"If the proof is completed, this is a substantial result: it extends the author's earlier generating-function computation from [G1] to the full tensor-product integral on metaplectic covering groups of GL_r, covering all r < nm rather than the special cases (r = 1, 2) treated by Suzuki. The introduction of the generating function W^(n)_{τ^(n),nrm} and the clean distinction between the two computations of I are valuable ideas that are likely to be useful for further local and global Rankin--Selberg constructions. The paper is also transparent in marking the global parts as conditional on the Suzuki conjecture. However, the central proof is not self-contained: several load-bearing identities are either omitted with a reference to [G1] or deferred to the arXiv preprints [K] and [C2], which are used as black boxes. Because these deferred steps are the actual mechanism of the proof, the published value of the paper depends on the author supplying the missing details and verifying the hypotheses of the external results.","major_comments":[{"comment":"The reduction of the inner integral over V_r and U^0_{nm,r} in (13) to the single integral (14), and then the transformation to (15), is stated to be 'exactly as in [G1]' and 'We omit it here.' This is a load-bearing step in the first computation of I: it removes the V_r integration and produces the integral over U^3_{nm,r} whose Fourier expansion leads to the values W_{τ^(n),nm}(diag(a_i, I_{nm−1})) in (16). Since Theorem 1 depends on this equality, the author should either provide the full root-exchange/unipotent-change-of-variables computation or isolate a precise theorem from [G1] and verify its hypotheses here. An analogy with [G1] is not sufficient for this step. The same omission also affects the claim just before (11) that the unramified integral defining W^(n)_{τ^(n),nrm} is nonzero, which is justified by reference to the computation in Section 3.","section":"Section 3, equations (13)--(15)"},{"comment":"The identity asserting that the integral over V_r of W^(n)_{τ^(n),nrm}(diag(vg, I_{r(nm−1)})) ψ^{-1}(v) dv equals W^(n)_{τ^(n),nm}(diag(g, I_{nm−r})) |g|^{(nm−1)(r−1)/2} is stated to follow by 'a similar proof as the proof of Theorem 2 in [G1]' with the only change being to replace n by nm. This identity is the sole bridge from I to Suzuki's original integral (2), and it is therefore load-bearing for Theorem 1. The cited theorem in [G1] was proved in the m = 1 situation of the construction; replacing n by nm also changes the inducing representation Θ(τ^(n)), whose τ^(n) is now a genuine representation of GL_m^(n) rather than a character, as well as the sizes of the unipotent radicals and the normalization factors. The paper does not demonstrate that the proof of [G1, Theorem 2] survives these changes. The author should provide a complete proof or a reduction to a stated theorem whose hypotheses are explicitly verified.","section":"Section 4, identity after equation (18)"},{"comment":"The final evaluation of the one-dimensional integrals in (17) as ∏_i L(χ_i^n × τ^(n), ns − (n−1)/2) is justified solely by 'Theorem 46 in [K]' and 'Theorem 8.1 in [C2]', both of which are arXiv preprints whose statements and hypotheses are not reproduced. These external results are used as black boxes in the central computation of Theorem 1. The reader cannot verify that the relevant hypotheses, such as the support properties of the unramified Whittaker function and the uniqueness statements for the induced representations in question, are satisfied in the present setting. The author should state the exact theorems used, reproduce their hypotheses, and check that they apply here, or give a self-contained proof in an appendix.","section":"Section 3, after equation (17)"}],"minor_comments":[{"comment":"In the sentence referring to the 'Rakin-Selberg integral', the name should be 'Rankin-Selberg'.","section":"Section 5.2, page 11"},{"comment":"The notation for the generating function shifts between W^(n)_{τ^(n),nrm,f}, W^(n)_{τ^(n),nrm}(h), and the bar version f^(n)_{nrm,W} without a single explanatory paragraph; please clarify the normalization and the dependence on the vector f.","section":"Section 2, around equation (11)"},{"comment":"The sentence 'it follows from the Theorem on page 753 in [S]' cites a result by page number rather than by theorem number; please cite a numbered theorem from [S].","section":"Section 5.1, page 9"},{"comment":"Several steps in the proof of Theorem 4, including the root-exchange step connecting equations (29) and (30) and the constant-term computation producing equation (35), are asserted without proof. Since this part is conditional and secondary to Theorem 1, this is a presentation issue, but the reader should be told which steps are intended to be routine and which require additional hypotheses.","section":"Section 5, proof of Theorem 4"}],"recommendation":"major_revision","confidential_remarks":"The core issue is that Theorem 1 is the paper's main new result, yet its proof is not self-contained at three specific junctures: the omitted V_r/U^0_{nm,r} reduction in Section 3, the deferred identity in Section 4, and the reliance on two arXiv preprints for the final L-factor evaluation. These are computational and structural gaps rather than an obvious contradiction, so a major revision with a detailed appendix or precise theorem statements could make the proof verifiable. The global sections are explicitly conjectural and do not block the local theorem. The novelty relative to [G1] is precisely the extension from m = 1 to general m, and it is exactly that extension that is deferred, so the author should be encouraged to provide the missing arguments in full."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know two things. First, the paper's main theorem is exactly the kind of result the metaplectic L-function community has been waiting for: it extends Suzuki's local integral identity to all r < nm, covering cases that were previously out of reach. Second, the proof as written is not self-contained, and the load-bearing steps are the least detailed ones. Nothing I saw is wrong, but a referee will need to push hard on Section 4 and the use of the two preprints.\n\nWhat is genuinely new: the full-range local identity in Theorem 1, and the global doubling integral in Section 5.3 for the linear case (n = 1). The method is the author's own generating-function technique from [G1], applied to the tensor-product setting. That is a real extension, not a trivial repackaging. Section 3's computation of integral I as the tensor L-factor is sketched but mostly follows the pattern of [G1], and the algebra up to equation (17) looks coherent. The global sections are explicitly conjectural, which is honest and appropriate.\n\nWhere the soft spots are, in proportion: the reduction from (13) to (15) is simply omitted, with a pointer to [G1]. That is annoying but minor if the analogy really is exact. The bigger issue is Section 4: the bridge from the auxiliary integral I to Suzuki's integral (2) is asserted by 'a similar proof as Theorem 2 in [G1]' with only the parameter change n to nm. The stress-test note is right that [G1]'s Theorem 2 was proved in a different setting, and it is not demonstrated that the proof survives for the general induced representation Theta(tau(n)) with the Whittaker-Speh-Shalika functional. If that identity fails, Theorem 1 collapses, even if Section 3 is correct. Also, the final identification of the one-dimensional integrals with L-factors depends on Theorem 46 of [K] and Theorem 8.1 of [C2], both arXiv preprints. The author does not reproduce their hypotheses or statements. That is not fatal by itself, but it makes the paper hard to check and risky to rely on.\n\nWho this is for: specialists in metaplectic forms and Rankin-Selberg theory. A non-specialist will not be able to fill the gaps. The paper deserves a serious referee, because the result is important and likely correct, but it should not be accepted as is. The referee should ask the author to write out the Section 4 identity with enough detail to verify the hypotheses, and to state precisely which results from [K] and [C2] are being used, including their hypotheses.","headline":"A plausible and valuable extension of Suzuki's local identity, but the central proof leans on omitted computations and two preprints; worth refereeing if the gaps are fillable.","tokens_in":15937,"tokens_out":889,"would_cite":false,"duration_ms":95581,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11F70","11F55","11F66"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that a local integral on metaplectic covers of $GL_r$ equals the tensor-product $L$-function with shifted argument whenever $r<nm$.","keywords":["metaplectic covering groups","tensor product L-functions","local unramified integrals","Rankin-Selberg convolutions","Whittaker functions","generating functions","covering groups","unipotent integration"],"falsifier":"For fixed unramified data, say $r=2,m=2,n=3$, one could compute the left-hand side of Theorem 1 directly by integrating the relevant Whittaker functions over $V_2\\setminus GL_2$, and compare the result with the product formula from equation (7) after the shift $s\\mapsto ns-(n-1)/2$; a mismatch would settle the claim false.","tokens_in":14815,"feed_emoji":"🧮","tokens_out":18100,"duration_ms":549715,"temperature":0.7,"pith_summary":"This paper claims that a local integral originally introduced in the literature on metaplectic covering groups of $GL_r$ computes the standard tensor-product $L$-function in full generality, provided the two group sizes satisfy $r<nm$. Previously the identity was known only for small $r$ ($r=1,2$) with restrictions on $m$; the author says the general proof is a straightforward extension of his earlier generating-function method. The strategy is to write the integral as part of a larger auxiliary integral, evaluate that auxiliary integral twice—once with a spherical function and once by unfolding against a Whittaker function—and equate the results. If the theorem is correct, it supplies the missing local identity that makes global convolution constructions on covering groups represent the expected partial $L$-functions, with the $L$-function argument shifted by the cover degree.","feed_headline":"Metaplectic local integral equals tensor-product L-function","feed_subtitle":"For every r<nm on metaplectic covers, this integral recovers the tensor-product L-function with the parameter shifted by the cover degree.","key_machinery":"The engine of the proof is the generating function $W^{(n)}_{\\tau^{(n)},nrm}(h)$, defined on $GL^{(n)}_{nrm}$ as a Whittaker-type integral of the unramified vector over a unipotent subgroup, with character $\\psi_{U_{nm,r}}$. This function is not itself a Whittaker function, but it has the left equivariance and bi-invariance properties that make the auxiliary integral $I$ (equation (4)) well defined and computable. In Section 3 the computation of $I$ uses the spherical function of $\\pi^{(n)}$ and a chain of unipotent integrations to reduce the integral to a product of one-dimensional integrals; the final identification of each one-dimensional integral with a local $L$-factor is quoted from known evaluations for covering groups. In Section 4 the same $I$ is unfolded against an arbitrary Whittaker function of $\\pi^{(n)}$, and the result is exactly the original integral (2).","core_discovery":"Theorem 1 states that for $r<nm$ the local integral over $V_r\\setminus GL_r$ of $W_{\\pi^{(n)}}(g)\\,W_{\\tau^{(n)},nm}(\\mathrm{diag}(g,I_{nm-r}))\\,|g|^{s-(nm-r)/2}\\,dg$ equals the unramified tensor-product $L$-function $L(\\pi^{(n)}\\times\\tau^{(n)}, ns-(n-1)/2)$. Here $W_{\\tau^{(n)},nm}$ is the unique Whittaker function—that is, the function with the standard equivariance under a maximal unipotent subgroup—of an induced representation built from $n$ copies of $\\tau^{(n)}$, and the shift $ns-(n-1)/2$ records the covering degree. The core discovery is that this particular integral—with the multi-copy induced representation in place of a single $\\tau^{(n)}$—makes the classical convolution computation survive on covers, and that the computation can be carried out for every $r<nm$, not just the previously treated $r=1,2$ cases.","pith_inferences":["I infer that Theorem 1 should extend to $r\\ge nm$ as well: the paper notes that the spherical-side computation works for all $mn>1$, so the only missing piece is the corresponding unfolding identity in Section 4, which is not proved for that range.","I infer that the generating-function trick is a general recipe for covering groups: replace $\\tau^{(n)}$ by $n$ copies of itself in an induced representation, use the unique Whittaker function of that induced object, and convolve with the representation on $GL_r$; the same pattern may apply to other convolution integrals.","A concrete testable extension is to verify the omitted unipotent-integration identities in a small case such as $r=2,m=2,n=3$ by direct computation; that would either certify the deferred steps or expose a gap in the chain.","I infer that the main remaining obstacle for the global theorems is not the local identity, which is now established, but the construction of the automorphic representations $\\epsilon^{(n)}(\\tau)$; all global applications in the paper are conditional on that existence."],"forward_implications":["For every pair of ranks with $r<nm$, the local unramified integral (2) represents $L(\\pi^{(n)}\\times\\tau^{(n)}, ns-(n-1)/2)$, removing the earlier restrictions to $r=1,2$ and the accompanying bounds on $m$.","At $n=1$, the identity reduces to the classical convolution integral of [J-PS-S], so the construction is an extension of the linear theory.","Combined with the global unfolding of Section 5.2, and assuming the stated existence conjecture for the representations $\\epsilon^{(n)}(\\tau)$, the partial tensor-product $L$-function is holomorphic away from the possible simple poles at $s=0,1$ in the equal-rank case.","The new global doubling integral on the linear group $GL_r\\times GL_m$ unfolds to the same local integral (4), giving a doubling-method representation of the partial tensor-product $L$-function $L^S(\\pi\\times\\tau,s)$.","The spherical-side computation of the auxiliary integral is valid for all $mn>1$, independently of the condition $r<nm$; only the unfolding step that recovers the original integral uses this inequality."],"supporting_citations":[{"why":"Supplies the linear convolution integral (1) that this construction extends, and the n=1 identity.","marker":"[J-PS-S]"},{"why":"Introduced the local integral (2) and the covering-group conjecture, and proved the special cases r=1,2.","marker":"[S]"},{"why":"Introduced the generating-function method and contains the unipotent-integration steps that Sections 3 and 4 defer to.","marker":"[G1]"},{"why":"Constructs the theta representations on covers that have the unique Whittaker function used in defining the auxiliary functions.","marker":"[K-P]"},{"why":"Studied the r=1 case and supplies the spherical function on covers used in the first computation of the auxiliary integral.","marker":"[B-F]"},{"why":"Provides the theorem that the space of the generating functions is one-dimensional with the stated invariance.","marker":"[C1]"},{"why":"Defines the function $W^{(n)}_{\\tau^{(n)},nrm}$ used as the generating function.","marker":"[C-F-G-K]"},{"why":"Contains the evaluation of unramified Whittaker integrals that identifies the remaining one-dimensional integral with a local L-factor.","marker":"[K]"},{"why":"Provides the theorem on unramified Whittaker functions that underlies the same final identification.","marker":"[C2]"}],"fun_headline_variants":["Tensor L-function on metaplectic covers for all r<nm","Local integral equals tensor L-function on covers","All r<nm: metaplectic integral yields tensor L-function","Cover-degree shift: integral equals tensor L-function","Metaplectic integral recovers tensor L-function"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The central equality depends on several long unipotent-integration computations that are not reproduced here but are said to be identical to computations in earlier work, and it also depends on two quoted evaluations of one-dimensional Whittaker integrals; if any of these deferred steps is wrong, the claimed equality does not follow from the argument presented.","fun_headline_variants_meta":{"raw":{"variants":["Tensor L-function on metaplectic covers for all r<nm","Local integral equals tensor L-function on covers","All r<nm: metaplectic integral yields tensor L-function","Cover-degree shift: integral equals tensor L-function","Metaplectic integral recovers tensor L-function"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000891,"raw_usage":{"total_tokens":3835,"prompt_tokens":929,"completion_tokens":2906,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":545,"completion_tokens_details":{"reasoning_tokens":2838}},"tokens_in":545,"tokens_out":2906,"duration_ms":105170,"temperature":1.0,"reasoning_tokens":2838,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:58:10.057334+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For fixed unramified data, say $r=2,m=2,n=3$, one could compute the left-hand side of Theorem 1 directly by integrating the relevant Whittaker functions over $V_2\\setminus GL_2$, and compare the result with the product formula from equation (7) after the shift $s\\mapsto ns-(n-1)/2$; a mismatch would settle the claim false.","supporting_citations":[],"review_version":1}