Pith. sign in

REVIEW 3 major objections 4 minor 18 references

Tensor Product $L$-Functions On Metaplectic Covering Groups of $GL_r$

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper proves that a local integral on metaplectic covers of $GL_r$ equals the tensor-product $L$-function with shifted argument whenever $r<nm$.

desk verdict 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. read the letter →

arxiv 1908.07720 v1 pith:N7TB7UNS submitted 2019-08-21 math.RT math.NT

David Ginzburg This is my paper · ORCID
classification math.RTmath.NT MSC 11F7011F5511F66
keywords metaplecticcoveringgroupstensorproductL-functionslocalunramifiedintegralsRankin-SelbergconvolutionsWhittakerfunctionsgeneratingunipotentintegration
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

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

What carries the argument

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).

What would settle it

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.

Watch

Extended reading notes

Core claim

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.

Load-bearing premise

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.

Editorial extensions

If this is right

  • 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.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • 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.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

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.

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 (3)
  1. [Section 3, equations (13)--(15)] 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.
  2. [Section 4, identity after equation (18)] 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.
  3. [Section 3, after equation (17)] 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.
minor comments (4)
  1. [Section 5.2, page 11] In the sentence referring to the 'Rakin-Selberg integral', the name should be 'Rankin-Selberg'.
  2. [Section 2, around equation (11)] 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.
  3. [Section 5.1, page 9] 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].
  4. [Section 5, proof of Theorem 4] 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.

Circularity Check

0 steps flagged · score 2.0 of 10

No constructional circularity: Theorem 1's L-function identity is computed from independent definitions; reliance on [G1], [K], and [C2] is a normal chain of cited lemmas, not an assumption of the target result.

full rationale

The claimed equality in Theorem 1 is not built into the definitions. The local L-function is defined in equation (7) as a product over Satake parameters, while Suzuki's integral (2) is an integral of Whittaker functions normalized at the identity; neither object is defined in terms of the other, and there is no fitted parameter later renamed as a prediction. The proof computes the auxiliary integral I in two ways. The Section 3 computation reduces (13) to (15) by asserting the proof is 'exactly as the proof in [G1]', and ultimately invokes 'Theorem 46 in [K] and Theorem 8.1 in [C2]' to identify one-dimensional integrals with L-factors. The Section 4 bridge from I to Suzuki's integral is asserted by 'a similar proof as the proof of Theorem 2 in [G1]' with 'the only different in the proof is replacing n there with nm.' These are deferred arguments and possible correctness risks: the hypotheses for the generalized bridge and the exact statements of [K, Thm 46] and [C2, Thm 8.1] are not reproduced in this paper. But none of these citations is being used to assert the target equality itself; each is a separate computation of Whittaker-function integrals, and the cited results do not appear to assume Theorem 1 as input. There is also no uniqueness theorem imported solely from the author's prior work to forbid alternatives, and the generating function W^(n)_{tau(n),nrm} is constructed from the representation Theta(tau(n)), not from the L-function it is used to compute. Thus the derivation chain is not circular, though it is not self-contained; the main weakness is omitted verification of deferred identities, which is a completeness issue rather than a circularity issue.

Assumptions & free parameters 0 free parameters · 5 assumptions · 2 invented entities

The central local claim (Theorem 1) rests on deferred and external computational results, while the global claims additionally assume Conjecture 1. No numerical free parameters appear; the shift in the L-function argument in Theorem 1 is a consequence of the computation, not a fitted constant.

assumptions (5)
  • domain assumption The framework of unramified principal series on metaplectic covers GL(n)_r and the L-function definition L(π(n) × τ(n), s) as in equation (7) are correct.
    The local representations and L-function are taken as given from [G1], [S], and [K-P]; the paper does not re-derive the covering group setup.
  • domain assumption The local Theta representation and the uniqueness of its Whittaker functional exist as described in [K-P].
    Used to construct the generating function W(n) of τ(n), nrm and to ensure uniqueness properties; cited from [K-P].
  • domain assumption Theorem 46 of [K] and Theorem 8.1 of [C2] correctly evaluate the unramified Whittaker function and yield the product of one-dimensional L-factors.
    This is the crucial step after equation (17) that identifies each one-dimensional integral with L(χ_i^n × τ(n), ...). Both [K] and [C2] are arXiv preprints not proven in this paper.
  • ad hoc to paper The omitted computational equalities hold when the arguments in [G1] are repeated with n replaced by nm.
    The paper states the proof of (13) to (15) is 'exactly as in [G1]' and omits it (after eq. (14)), and Section 4 defers to 'a similar proof as the proof of Theorem 2 in [G1]'. These are load-bearing and not shown.
  • domain assumption Suzuki Conjecture (Conjecture 1) holds, and the residue representations ǫ(n)(τ) and ǫ(n)(τ(n)) exist as claimed.
    All of Section 5 depends on this; the author explicitly says 'This is not known in general' and 'most of the discussion in that Section is still conjectural'.
invented entities (2)
  • Generating function W(n)_{τ(n), nrm}(h) on GL(n)_{nrm}
    purpose: Auxiliary function used to relate Suzuki's integral (2) to the spherical-function integral (4); extends [G1] equation (7).
    Defined locally via the Whittaker-Speh-Shalika functional in (11). Its existence and properties are established within the paper and cited works; there is no independent observable evidence outside the mathematical construction.
  • Suzuki representation ǫ(n)(τ) (and residue representation ǫ(n)(τ(n)))
    purpose: Automorphic representation on GL(n)_{nm}(A) conjectured to exist for every cuspidal τ; controls the global integrals and Shimura lift statements.
    Existence is exactly the Suzuki Conjecture, which the paper states is not known in general. Some cases exist as residues of Eisenstein series, but the general construction is conjectural.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Tensor Product $L$-Functions On Metaplectic Covering Groups of $GL_r$." pith.science (2026). https://pith.science/paper/N7TB7UNS

@misc{pith2026190807720,
  author       = {Pith},
  title        = {Pith review of: Tensor Product $L$-Functions On Metaplectic Covering Groups of $GL_r$},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/N7TB7UNS}},
  note         = {Machine review of arXiv:1908.07720}
}
abstract

In this note we compute some local unramified integrals defined on metaplectic covering groups of $GL$. These local integrals which were introduced by Suzuki, represent the standard tensor product $L$ function $L(\pi^{(n)}\times \tau^{(n)},s)$ and extend the well known local integrals which represent $L(\pi\times \tau,s)$. The computation is done using a certain "generating function" which extends a similar function introduced by the author in a previous paper. In the last section we discuss the Conjectures of Suzuki and introduce a global integral which unfolds to the above local integrals. This last part is mainly conjectural and relies heavily on the existence of Suzuki representations defined on covering groups. In the last subsection we introduce a new global doubling integral which represents the partial tensor product $L$ function $L^S(\pi\times \tau,s)$.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

18 extracted references · 17 canonical work pages

  1. [1]

    Automorphic forms, automorphic representations, and arithmetic (Fort Worth, TX, 1996), 1--17, Proc

    Bump, Daniel; Friedberg, Solomon Metaplectic generating functions and Shimura integrals. Automorphic forms, automorphic representations, and arithmetic (Fort Worth, TX, 1996), 1--17, Proc. Sympos. Pure Math., 66, Part 2, Amer. Math. Soc., Providence, RI, 1999

    work page 1996

  2. [2]

    Bump, Daniel; Friedberg, Solomon; Ginzburg, David Small representations for odd orthogonal groups. Int. Math. Res. Not. 2003, no. 25, 1363--1393

    work page 2003

  3. [3]

    Duke Math

    Bump, Daniel; Hoffstein, Jeffrey Some Euler products associated with cubic metaplectic forms on GL(3) . Duke Math. J. 53 (1986), no. 4, 1047--1072

    work page 1986

  4. [4]

    Duke Math

    Bump, Daniel; Hoffstein, Jeffrey On Shimura's correspondence. Duke Math. J. 55 (1987), no. 3, 661--691

    work page 1987

  5. [5]

    Cai, Yuanqing; Fourier coefficients for theta representations on covers of general linear groups. Trans. Amer. Math. Soc. 371 (2019), no. 11, 7585–7626

    work page 2019

  6. [6]

    Unramified Whittaker functions for certain Brylinski-Deligne covering groups

    Cai, Yuanqing; Unramified Whittaker functions for certain Brylinski-Deligne covering groups. In arXiv:1904.12594

    work page Pith review arXiv 1904

  7. [7]

    Doubling Constructions for Covering Groups and Tensor Product L-Functions

    Cai, Y.; Friedberg, S.; Ginzburg, D.; Kaplan E, Doubling Constructions for Covering Groups and Tensor Product L-Functions. In arXiv:1601.08240

    work page Pith review arXiv

  8. [8]

    Automorphic forms on covering groups of GL(2)

    Flicker, Yuval Z. Automorphic forms on covering groups of GL(2). Invent. Math. 57 (1980), no. 2, 119–182

    work page 1980

Show all 18 references
  1. [9]

    Friedberg, Solomon; Ginzburg, David; On the Genericity of Eisenstein Series and Their Residues for Covers of GLm. Int. Math. Res. Not. IMRN 2017, no. 4, 1000--1012

    2017

  2. [10]

    Ginzburg, David Generating functions on covering groups. Compos. Math. 154 (2018), no. 4, 671–684

    2018

  3. [11]

    Ginzburg, David On certain global constructions of automorphic forms related to a small representation of F4. J. Number Theory 200 (2019), 1–95

    2019

  4. [12]

    World Scientific Publishing Co

    Ginzburg, David; Rallis, Stephen; Soudry, David The descent map from automorphic representations of GL(n) to classical groups. World Scientific Publishing Co. Pte. Ltd., Hackensack, NJ, 2011. x+339 pp. ISBN: 978-981-4304-98-6; 981-4304-98-0

    2011

  5. [13]

    In arXiv:1808.01572

    Ginzburg, David; Soudry, David Two Identities relating Eisenstein series on classical groups. In arXiv:1808.01572

    arXiv

  6. [14]

    I.; Shalika, J

    Jacquet, H.; Piatetskii-Shapiro, I. I.; Shalika, J. A. Rankin-Selberg convolutions. Amer. J. Math. 105 (1983), no. 2, 367--464

    1983

  7. [15]

    On Fourier coefficients of automorphic forms of GL(n)

    Jiang, D.; Liu B. On Fourier coefficients of automorphic forms of GL(n). Intern. Math. Res. Notices, Vol. 2013, No. 17, pp. 4029--4071

    2013

  8. [16]

    In arXiv:1902.00880

    Kaplan E, Doubling Constructions and Tensor Product L-Functions: coverings of the symplectic group . In arXiv:1902.00880

    1902 arXiv

  9. [17]

    A.; Patterson, S

    Kazhdan, D. A.; Patterson, S. J. Metaplectic forms. Inst. Hautes Études Sci. Publ. Math. No. 59 (1984), 35--142

    1984

  10. [18]

    Suzuki, Toshiaki Distinguished representations of metaplectic groups. Amer. J. Math. 120 (1998), no. 4, 723–755

    1998

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.