{"id":"b5b41a88-1218-4b71-862c-cfbfd0a4c843","arxiv_id":"2509.10880","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"For depth-zero, simple, middle, and biquadratic supercuspidal representations of GL(4,F), being a functorial transfer from SO(5,F) is equivalent to having trivial central character, with an extra twist condition in the simple case.","lead":"This paper studies which supercuspidal representations of GL(4) over a p-adic field come from the smaller group SO(5). It proves this happens exactly when the central character is trivial, for four families, by explicitly computing integrals attached to L-functions.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The arbitrary-twist claim fails for quartic characters: a symplectic parameter twisted by a non-quadratic quartic character is not self-dual, so Theorem 1.2(3) cannot imply transfer.","rationale":"The reader's objection to Lemma 2.9 is valid and already fatal for the p=2 step: the Whittaker models of π and π⊗χ are not equal as subspaces, and the claimed equality is what underpins the p=2 reduction in §4.3. However, the single most load-bearing problem is broader. Theorem 1.2 asserts that for depth-zero, middle, and biquadratic supercuspidals, trivial central character is sufficient for π⊗χ to be a transfer from SO(5,F), for arbitrary quasi-characters χ=η∘det. This is incompatible with the standard compatibility of the LLC with twisting by characters: if φ is symplectic, then φ⊗η is self-dual only when η^2=1. For a quartic η, the twisted parameter is not even self-dual, so it cannot lie in Sp(4,C), and the exterior-square L-function has no pole at s=0. This gives a concrete counterexample to the stated theorem using the paper's own untwisted depth-zero case, independent of the detailed computations in §4. The reader's Lemma 2.9 concern is a symptom of the same over-optimistic erasure of the twist, but the quartic-twist argument shows the theorem cannot be repaired merely by correcting Lemma 2.9; the arbitrary-twist formulation is false. I therefore recommend REJECT, in agreement with the reader's verdict, while noting that the strongest reason is the structural obstruction to quartic twists, not only the false lemma.","tokens_in":27758,"tokens_out":23213,"duration_ms":222526,"concrete_test":"Take F=Q_p with p≡1 mod 4, fix a tamely ramified quartic character η, and let π be a depth-zero supercuspidal of GL(4,F) with trivial central character (for instance a level-zero lift of a cuspidal Deligne–Lusztig representation with trivial central character). Using the local Langlands correspondence, verify that φ_{π⊗(η∘det)} = φ_π⊗η and compute ∧^2(φ_π⊗η) = η^2⊗∧^2φ_π. Since η^2 is a nontrivial quadratic character, this representation contains no trivial constituent; hence L(s,π⊗(η∘det),∧^2) has no pole at s=0 and π⊗(η∘det) has no Shalika model. This directly contradicts the implication (3)⇒(1) in Theorem 1.2 for depth-zero, middle, and biquadratic twists.","verdict_should_be":"REJECT","load_bearing_attack":"Even setting aside Lemma 2.9, Theorem 1.2 as stated is inconsistent with the local Langlands correspondence and with its own untwisted depth-zero case. Let π be a depth-zero supercuspidal of GL(4,F) with trivial central character; by the depth-zero case cited from [23], π is a transfer from SO(5,F), so its L-parameter φ is a four-dimensional symplectic representation of W_F. Choose F with p≡1 mod 4 and a tamely ramified quartic character η of F^× (so η^4=1 but η^2≠1), and put χ=η∘det. Then π⊗χ has trivial central character, so Theorem 1.2(3) asserts that π⊗χ is a transfer. However, the L-parameter of π⊗χ is φ⊗η, and (φ⊗η)^∨ ≅ φ^∨⊗η^{-1} ≅ φ⊗η^{-1}; since φ is irreducible and η^2≠1, φ⊗η is not self-dual and hence cannot have image in Sp(4,C). Equivalently, ∧^2(φ⊗η) = η^2⊗∧^2φ contains no trivial subrepresentation, so by Theorems 2.7–2.8 the local exterior-square L-function has no pole at s=0 and π⊗χ has no Shalika model. Thus condition (3) does not imply condition (1) for arbitrary twists. The failure is not confined to the p=2 argument: it is visible already for p odd. Lemma 2.9 is also false as stated—W(π⊗χ,ψ_F)=χ·W(π,ψ_F), not W(π,ψ_F)—but the quartic-twist obstruction is the more fundamental problem.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the local Langlands functoriality transfer from SO(5,F) to GL(4,F) for supercuspidal representations of GL(4,F) twisted by characters χ=η∘det. It constructs a new family of depth-one \"biquadratic\" supercuspidals, supplies explicit Paškūnas–Stevens Whittaker functions for the middle and biquadratic families, and performs explicit Λ0-period computations. The main theorem, Theorem 1.2, claims that for depth-zero, middle, and biquadratic supercuspidals, being a transfer is equivalent to possessing a nonzero Shalika model and to having trivial central character, with an extra parameter condition in the simple supercuspidal case. The proof of the untwisted middle and biquadratic cases is carried out in Sections 4.1 and 4.2, and Section 4.3 attempts to pass to arbitrary twists using Lemma 2.9 and a reduction for residue characteristic 2.","tokens_in":28131,"tokens_out":9911,"duration_ms":92164,"significance":"The detailed period computations in Sections 4.1 and 4.2 are substantial and appear internally coherent; the construction of biquadratic supercuspidals and their explicit Whittaker functions is a potentially useful contribution. However, the central claim of the paper, the arbitrary-twist characterization in Theorem 1.2, is false as stated. The error is not a minor gap: it stems from twisting by quartic characters, which changes a symplectic L-parameter into a non-self-dual one. Because the main theorem is false, the paper cannot be accepted in its current form, although the untwisted core and the new family might be salvageable in a revised manuscript.","major_comments":[{"comment":"Theorem 1.2 is false for arbitrary twists, already for residue characteristic p odd. Let F have residue characteristic p≡1 mod 4 and let η be a tamely ramified quartic character of F^×, so η^4=1 but η^2≠1. Let π be a depth-zero supercuspidal representation of GL(4,F) with trivial central character. By the untwisted depth-zero case cited in the paper, π is a transfer from SO(5,F), so its L-parameter φ is a four-dimensional irreducible symplectic representation of W_F. Put χ=η∘det. Then π⊗χ has trivial central character, because χ is trivial on the center F^× of GL(4,F). Theorem 1.2(3) therefore asserts that π⊗χ is a transfer and has a Shalika model. But the L-parameter of π⊗χ is φ⊗η, and (φ⊗η)^∨ ≅ φ^∨⊗η^{-1} ≅ φ⊗η^{-1}. Since φ is irreducible and η≠η^{-1}, φ⊗η is not self-dual, so it cannot factor through Sp(4,C). Equivalently, ∧^2(φ⊗η) = η^2⊗∧^2φ has no trivial constituent because η^2 is nontrivial, so by Theorems 2.7 and 2.8 the exterior-square L-function has no pole at s=0 and π⊗χ has no Shalika model. Thus condition (3) does not imply conditions (1) and (2) for arbitrary twists, and the failure is not confined to the p=2 argument in Section 4.3.","section":"Theorem 1.2 and Section 4.3"},{"comment":"Lemma 2.9 is false as stated. The identity of Whittaker models W(π,ψ_F)=W(π⊗χ,ψ_F) is not valid. For a Whittaker function W∈W(π,ψ_F), the corresponding function in W(π⊗χ,ψ_F) is g↦χ(g)W(g), so the two models are related by multiplication by χ, not identical. The displayed proof only establishes an isomorphism of Hom-spaces, i.e. that a nonzero Whittaker functional exists for π if and only if one exists for π⊗χ. Even when χ=η∘det is trivial on N(4,F), the second displayed isomorphism is valid, but it does not identify the spaces of functions themselves. This matters directly: Section 4.3 uses the false model equality to assert that the pole at s=0 of L(s,π′⊗((Ψη)∘det),∧^2) is equivalent to the pole for L(s,π′,∧^2). That inference is unsupported.","section":"Lemma 2.9"},{"comment":"The p odd case of the reduction is also incorrect. The sentence \"η is tamely ramified. This implies that π⊗χ is of the same type as π\" establishes, at most, that π⊗χ belongs to the same family of supercuspidals; it does not establish that the untwisted Shalika/pole criterion applies. Twisting by η changes the L-parameter by η, and the untwisted Propositions 4.2 and 4.4 say nothing about the twisted L-function. In fact, as shown in the first major comment, when η has order four the twisted representation has trivial central character but is not a transfer. The reduction to the untwisted case is therefore invalid even when no residue-characteristic-2 phenomenon is involved.","section":"Section 4.3, p odd reduction"}],"minor_comments":[{"comment":"There are incorrect cross-references in the proof of Proposition 4.4: \"Using equations (4.26) and (4.17)\" appears before equation (4.26) has been introduced, and \"Equation (4.26) implies\" in the T2 calculation should presumably refer to equation (4.27). These make an already intricate computation harder to follow.","section":"Section 4.2, T1 and T2"},{"comment":"The proof of Proposition 3.5 says only that it \"follows similarly\" from [18, Proposition 3.2]. Since biquadratic supercuspidal representations are a new family introduced in this paper, the bijection with triples (f̄_M, χ_M, ζ) and the verification that the constructed types are maximal should be spelled out more fully.","section":"Section 3.3, Proposition 3.5"},{"comment":"The equality ζ′=ζ·η(−vϖ_F)^{-1} is asserted after choosing a tamely ramified Ψ with (Ψη)^4≡1. The text says \"Since Ψ=η^{-1} on μ'_F and ϖ_F\", but this was not part of the earlier choice of Ψ and requires justification; as written, the displayed formula is not derived.","section":"Section 4.3, equation (4.31)"}],"recommendation":"reject","confidential_remarks":"The manuscript relies substantially on the author's own preprint [18] for the middle supercuspidal construction and Whittaker function; this is not circular, but the Editor may wish to confirm that the overlap is disclosed and that [18] is publicly available. The untwisted period computations and the biquadratic construction in Sections 3.3, 4.1, and 4.2 could form the basis of a useful revised paper if the claims are restricted to, for example, quadratic twists, but the arbitrary-twist Theorem 1.2 as stated is false and cannot be repaired by a local argument within the current scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's the deal. The paper has a genuinely useful untwisted core: biquadratic supercuspidals are constructed, explicit Whittaker functions are given, and the Λ0 computations in Propositions 4.2 and 4.4 look coherent and extend the Ye–Zelingher/Luo–Stevens line. If you only care about untwisted middle and biquadratic families, there is valuable material here.\n\nThe main theorem, however, is false for arbitrary twists. The reader's objection lands. Lemma 2.9 is wrong: the Whittaker model of π⊗χ is {χ(g)W(g) : W∈W(π)}, not equal to W(π). The proof shows the Hom spaces are isomorphic, which only gives existence of a Whittaker functional, not equality of functions. The p=2 reduction leans on this false lemma.\n\nMore fundamentally, there is a counterexample visible already for p odd. Take a depth-zero π with trivial central character; by the paper's own cited depth-zero result, π is a transfer, so its parameter φ is symplectic. Choose a tame quartic character η with η²≠1 (possible when p≡1 mod 4), and let χ=η∘det. Then π⊗χ has trivial central character, so Theorem 1.2 predicts it is a transfer and has a Shalika model. But the parameter of π⊗χ is φ⊗η, which is not self-dual because φ is irreducible and η²≠1, so it cannot have image in Sp(4,C); ∧²(φ⊗η) contains no trivial subrepresentation, and by the paper's own Theorems 2.7–2.8 there is no pole at s=0 and no Shalika model. So condition (3) does not imply (1) or (2). This is not a technical gap; the statement is wrong.\n\nThe paper is not a waste. The biquadratic construction is new, and the untwisted period computations deserve careful checking rather than dismissal. But the claimed equivalence for arbitrary twists needs to be restricted — the argument goes through for quadratic twists, for instance, but not for quartic ones.\n\nBottom line: the paper should not be accepted as is. A serious referee can still learn from the untwisted half and the new construction, and the author can likely repair the theorem by stating it for the right class of twists. It deserves referee time, but the outcome should be major revision or reject-and-resubmit, not acceptance.","headline":"The untwisted middle and biquadratic computations are a real contribution, but the arbitrary-twist theorem is false — a quartic twist of a depth-zero transfer is not a transfer.","tokens_in":28586,"tokens_out":7665,"would_cite":false,"duration_ms":67661,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["22E50","11F70"],"pacs":[],"model":"deepseek-v4-flash","headline":"For depth-zero, middle, and biquadratic supercuspidals of $\\mathrm{GL}(4,F)$, being an $\\mathrm{SO}(5)$-transfer is equivalent to having a nonzero Shalika model and trivial central character; simple supercuspidals need the extra condition…","keywords":["local Langlands functoriality","local Shalika model","supercuspidal representation","exterior square L-function","maximal simple types","SO(5) to GL(4)","simple supercuspidal","biquadratic supercuspidal"],"falsifier":"Pick a depth-zero supercuspidal $\\pi$ with trivial central character and a tamely ramified character $\\eta$ with $\\eta^4=1$ but $\\eta$ nontrivial on units, and compute $L(s,\\pi\\otimes(\\eta\\circ\\det),\\wedge^2)$ directly. If the pole at $s=0$ disappears, the claimed equivalence for twists fails; if the pole survives, the residue-characteristic-2 argument still needs a correct identification of the Whittaker model, since the models are related by multiplication by $\\eta\\circ\\det$, not equality.","tokens_in":27569,"feed_emoji":"🧮","tokens_out":16485,"duration_ms":477005,"temperature":0.7,"pith_summary":"This paper asks which irreducible supercuspidal representations of $\\mathrm{GL}(4,F)$ — the building-block representations over a non-archimedean local field of characteristic zero — arise as local Langlands functoriality transfers from the five-dimensional special orthogonal group $\\mathrm{SO}(5,F)$. For depth-zero, middle, and biquadratic supercuspidals, it proves that every twist $\\pi\\otimes\\chi$ by $\\chi=\\eta\\circ\\det$ is such a transfer exactly when the twisted representation has trivial central character and a nonzero local Shalika model; for simple supercuspidals the same equivalence holds with the extra parameter condition $\\zeta=\\pm\\eta(-v\\varpi_F)$. The interest is that the transfer is normally detected only indirectly, through a pole of the exterior square $L$-function or the existence of a Shalika functional, whereas triviality of the central character is an elementary invariant associated with the defining maximal simple type. A correct proof would give the first explicit type-theoretic description of this transfer and a template for the $\\mathrm{GL}(2N)$ question.","feed_headline":"Trivial central character detects SO(5)-to-GL(4) transfer","feed_subtitle":"Depth-zero, middle, and biquadratic supercuspidals need only trivial central character; simple ones add parameter check.","key_machinery":"The central object is the maximal simple type $(J,\\Lambda)$: every supercuspidal $\\pi$ is compactly induced from a compact-open-modulo-center subgroup $J$ carrying an irreducible representation $\\Lambda$, built from a simple stratum $[A,1,0,\\beta]$ (with $\\beta=0$ as the depth-zero case). To each such type the paper attaches an explicit Whittaker function $W_\\pi$, whose support is known to lie in a disjoint union of double cosets $N(4,F)\\beta^k J$ or $N(4,F)\\varpi_F^k J$. The computation runs through the twisted Shalika period $\\Lambda_{s_0}(W)$, a double integral over $\\mathrm{GL}(2,F)$ and $\\mathrm{Mat}(2\\times2,F)$; by the product formula for $L(s,\\pi,\\wedge^2)$, a pole at $s=0$ occurs exactly when $\\Lambda_0$ is nonzero on some Whittaker function while the central character is trivial. The paper evaluates $\\Lambda_0$ on the symmetric Whittaker function $\\pi(\\sigma_4)W_\\pi$, after using support lemmas to restrict the integration to three or fewer double cosets, and obtains nonzero volume terms exactly under the stated conditions.","core_discovery":"On the paper's own terms, Theorem 1.2 is the central discovery. For a depth-zero, middle, or biquadratic supercuspidal representation $\\pi$ of $\\mathrm{GL}(4,F)$ and $\\chi=\\eta\\circ\\det$, the following are equivalent: $\\pi\\otimes\\chi$ is a local Langlands functoriality transfer from $\\mathrm{SO}(5,F)$; $\\pi\\otimes\\chi$ has a nonzero local Shalika model; and $\\pi\\otimes\\chi$ has trivial central character. For a simple supercuspidal $\\pi=\\pi(v,\\phi,\\zeta)$, the same equivalences hold with the extra necessary and sufficient condition $\\zeta=\\pm\\eta(-v\\varpi_F)$. The proof uses the known theorem [14] identifying these three properties with the pole at $s=0$ of the exterior square $L$-function $L(s,\\pi\\otimes\\chi,\\wedge^2)$, then relies on the product formula for that $L$-function in terms of the twisted Shalika periods $\\Lambda_{s_0}$. For middle and biquadratic supercuspidals the paper computes $\\Lambda_0$ on an explicit Whittaker function, shows it is nonzero whenever the central character is trivial, and concludes; the depth-zero and simple untwisted cases are taken from [23] and [24], and the twist argument extends them to all $\\chi=\\eta\\circ\\det$.","pith_inferences":["A corrected residue-characteristic-2 argument would likely need to treat the Whittaker model of $\\pi\\otimes\\chi$ as obtained from that of $\\pi$ by multiplication by $\\chi$, rather than as the same set of functions; the paper's Lemma 2.9 asserts the stronger and false equality of model spaces.","The pattern in the minimal polynomials suggests a testable generalization: minimax supercuspidals whose minimal polynomial has only even powers may all satisfy the trivial-central-character criterion, while those with odd power terms may behave like the simple case.","The extra condition $\\zeta=\\pm\\eta(-v\\varpi_F)$ for simple supercuspidals may correspond, under the explicit local Langlands correspondence, to the Langlands parameter factoring through the embedded subgroup $\\mathrm{Sp}(4,\\mathbb{C})\\subset\\mathrm{GL}(4,\\mathbb{C})$; checking this would give a parameter-level proof of the simple case.","A direct test of the paper's method would be to compute $\\Lambda_0$ on a different Whittaker function for a depth-one minimax supercuspidal whose minimal polynomial has odd power terms, since the paper's chosen function gives zero and it is open whether another choice could still produce a nonzero Shalika period."],"forward_implications":["For depth-zero, middle, and biquadratic supercuspidals, membership in the $\\mathrm{SO}(5)$-to-$\\mathrm{GL}(4)$ transfer can be checked by triviality of the central character alone; neither a Shalika functional nor an $L$-function computation is needed.","For simple supercuspidals, the transfer condition becomes a completely explicit statement about the parameters $(v,\\phi,\\zeta)$: a twist by $\\eta\\circ\\det$ is a transfer precisely when $\\zeta=\\pm\\eta(-v\\varpi_F)$.","The result converts the local Langlands transfer for these families into a statement about maximal simple types, giving the first explicit type-theoretic characterization of the transfer from $\\mathrm{SO}(5)$ to $\\mathrm{GL}(4)$.","Combined with the equivalence in [14], the theorem pins down exactly when the exterior square $L$-function $L(s,\\pi\\otimes\\chi,\\wedge^2)$ has a pole at $s=0$ for arbitrary twists of these families."],"supporting_citations":[{"why":"It supplies the theorem identifying an SO(5)-transfer with possession of a nonzero local Shalika model and with a pole of the exterior square L-function at s=0.","marker":"[14]"},{"why":"It gives the product formula for L(s,τ,∧²) in terms of nonzero values of the periods Λ_{s0}, which is how the pole at s=0 is detected.","marker":"[16]"},{"why":"It provides the maximal simple type classification and simple stratum theory on which all the supercuspidal constructions rest.","marker":"[5]"},{"why":"It supplies the explicit Whittaker functions attached to maximal simple types, the functions the paper evaluates in its Λ0 computations.","marker":"[21]"},{"why":"It gives the construction of middle supercuspidal representations and their explicit Whittaker functions, which the paper builds on.","marker":"[18]"},{"why":"It provides the untwisted depth-zero exterior-square base case that the twist argument for depth-zero supercuspidals extends.","marker":"[23]"},{"why":"It provides the untwisted simple-supercuspidal exterior-square base case that the twist argument for simple supercuspidals extends.","marker":"[24]"},{"why":"It parametrizes simple supercuspidal representations by triples (v,ϕ,ζ), giving the paper the parameter that enters the extra condition ζ=±η(-vϖ_F).","marker":"[11]"}],"fun_headline_variants":["SO(5) transfer for supercuspidals: trivial central character is key","Supercuspidal SO(5) transfer: central character decides for four families","Shalika model and L-function pole pin SO(5) supercuspidal transfer","Trivial central character suffices for SO(5) transfer in four supercuspidal families","Depth-zero and middle supercuspidals: SO(5) transfer iff trivial central character"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that twisting by a character of $\\mathrm{GL}(4,F)$ trivial on the center preserves the pole at $s=0$ of the exterior square $L$-function; the paper derives this from a lemma asserting the Whittaker models of $\\pi$ and $\\pi\\otimes\\chi$ are equal, and that lemma is false as stated, so the residue-characteristic-2 reduction is unsupported.","fun_headline_variants_meta":{"raw":{"variants":["SO(5) transfer for supercuspidals: trivial central character is key","Supercuspidal SO(5) transfer: central character decides for four families","Shalika model and L-function pole pin SO(5) supercuspidal transfer","Trivial central character suffices for SO(5) transfer in four supercuspidal families","Depth-zero and middle supercuspidals: SO(5) transfer iff trivial central character"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001624,"raw_usage":{"total_tokens":6486,"prompt_tokens":998,"completion_tokens":5488,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":614,"completion_tokens_details":{"reasoning_tokens":5375}},"tokens_in":614,"tokens_out":5488,"duration_ms":37108,"temperature":1.0,"reasoning_tokens":5375,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T15:53:01.534144+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Pick a depth-zero supercuspidal $\\pi$ with trivial central character and a tamely ramified character $\\eta$ with $\\eta^4=1$ but $\\eta$ nontrivial on units, and compute $L(s,\\pi\\otimes(\\eta\\circ\\det),\\wedge^2)$ directly. If the pole at $s=0$ disappears, the claimed equivalence for twists fails; if the pole survives, the residue-characteristic-2 argument still needs a correct identification of the Whittaker model, since the models are related by multiplication by $\\eta\\circ\\det$, not equality.","supporting_citations":[{"cited_title":"Jiang, C","cited_arxiv_id":null,"evidence_quote":"It supplies the theorem identifying an SO(5)-transfer with possession of a nonzero local Shalika model and with a pole of the exterior square L-function at s=0."},{"cited_title":"Derivatives and Exceptional Poles of the Local Exterior Square $L$-Function for $GL_m$","cited_arxiv_id":"1804.04613","evidence_quote":"It gives the product formula for L(s,τ,∧²) in terms of nonzero values of the periods Λ_{s0}, which is how the pole at s=0 is detected."},{"cited_title":"Pašk¯ unas and S","cited_arxiv_id":null,"evidence_quote":"It supplies the explicit Whittaker functions attached to maximal simple types, the functions the paper evaluates in its Λ0 computations."},{"cited_title":"Ye and E","cited_arxiv_id":null,"evidence_quote":"It provides the untwisted depth-zero exterior-square base case that the twist argument for depth-zero supercuspidals extends."},{"cited_title":"Ye and E","cited_arxiv_id":null,"evidence_quote":"It provides the untwisted simple-supercuspidal exterior-square base case that the twist argument for simple supercuspidals extends."},{"cited_title":"Imai and T","cited_arxiv_id":null,"evidence_quote":"It parametrizes simple supercuspidal representations by triples (v,ϕ,ζ), giving the paper the parameter that enters the extra condition ζ=±η(-vϖ_F)."}],"review_version":2}