{"id":"53a6e32b-8b97-4362-a69f-10feb9567cec","arxiv_id":"1908.03392","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For every level-zero Bernstein component of GL_n(F), the typical representations are exactly the irreducible subrepresentations of the induced Bushnell-Kutzko type ind_{P_I(1)}^{GL_n(O_F)}(τ_I).","lead":"This paper finds a complete description of certain small building-block representations that can appear inside representations of the p-adic matrix group GL_n(F), for all level-zero components. The result helps mathematicians identify which family a representation belongs to and supports work connecting p-adic groups to number theory.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proposition 4.3 asserts an unproved containment of the Clifford piece in parabolic induction; this is the decisive step for the nr>1 case.","rationale":"The paper is a serious and mostly coherent contribution: the reduction to level-zero types is natural, the Zelevinsky/Casselman machinery in the nr=1 case is plausible, and the external Paskunas unicity theorem cited is a standard result. The reader's weakest-assumption choice is defensible but not the most load-bearing place. The real soft spot is internal: in the nr>1 half of the main induction, Proposition 4.3's conclusion that every irreducible subrepresentation of the Clifford piece (22) embeds into ind_{P_I∩K}^{K_{n+1}} τ'_I is stated without proof and is not a routine consequence of the surrounding lemmas. The representation (22) is induced from a subgroup Z(η_k) carrying a nontrivial unipotent character, while the target is induced from the standard parabolic intersection on which that same unipotent subgroup acts trivially; bridging the two requires a Mackey decomposition and a nontrivial intertwiner. The manuscript does not provide this computation. The finite-field Gelfand-Graev analogy suggests the containment can fail, so the burden is on the author to supply the missing argument. Because the central classification theorem may still be true, I would not reject the paper outright; I would make acceptance conditional on a complete proof of the containment in Proposition 4.3, or on a replacement argument establishing atypicality of the Clifford pieces in the nr>1 case.","tokens_in":109,"tokens_out":34838,"duration_ms":806648,"concrete_test":"Test the finite-field analogue of the asserted containment: for q=2 or q=3, take G=GL_3(q), P=P_{(1,2)}(q), U its unipotent radical, and choose a nontrivial additive character ψ of U. Let Z be the stabilizer of ψ in P, let U_ψ be the corresponding Clifford character/extension, and let τ' be a non-cuspidal representation of the Levi GL_1(q)×GL_2(q), for instance a principal series. Use GAP or Magma to decide whether ind_Z^G(U_ψ ⊗ τ') is isomorphic to a subrepresentation of ind_P^G τ'. If this fails, the proof's embedding step is false in the finite model and the p-adic step lacks the necessary Mackey argument; if it holds, extract the explicit Mackey terms from the computation to see whether the same mechanism generalizes.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"In Proposition 4.3 (nr>1 case), the proof reduces atypicality to the assertion that the representation (22), namely ind_{P_I(1,m)}^{K_{n+1}} ind_{Z(η_k)}^{P_I(1,m)}(U_{η_k} ⊗ res_{Z∩M_I} τ'_I) ≅ ind_{Z(η_k)}^{K_{n+1}}(U_{η_k} ⊗ res_{Z∩M_I} τ'_I), 'occurs as a subrepresentation of' ind_{P_I∩K_{n+1}}^{K_{n+1}} τ'_I. This containment is the only bridge from the Clifford-theoretic decomposition to parabolic induction, and it is not a formal consequence of transitivity or Frobenius reciprocity. The subgroup Z(η_k) contains K_I(m)∩\\bar U, on which the Clifford character U_{η_k} is nontrivial, whereas the target induces a representation of P_I∩K that is trivial on the lower unipotent part; moreover Z(η_k) is not contained in P_I∩K. An embedding therefore requires an explicit Mackey double-coset computation over Z(η_k)\\K_{n+1}/(P_I∩K_{n+1}), and the paper supplies none. In the finite-group analogue, inducing a nontrivial unipotent character from a parabolic stabilizer is the Gelfand-Graev construction, which is not contained in a single parabolically induced representation; the asserted containment is thus a strong and non-obvious claim. If it fails, the atypicality of irreducible subrepresentations of ind_{P_I(1)}^{K_{n+1}}(U^0_{(1,m)}(τ_I)) is not established, so Theorem 3.2 and Corollary 3.3 are unsupported for all level-zero components with r>1 and n_r>1.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper classifies typical representations for level-zero Bernstein components of GL_n(F). Theorem 1.1 and Corollary 3.3 assert that for a level-zero component s=[M_I,σ_I], the K_n-irreducible subrepresentations of ind_{P_I(1)}^{K_n}(τ_I) are precisely the typical representations for s, and that every typical representation occurs there with the same multiplicity as in the parabolic induction i_{P_I}^{GL_n(F)}(σ_I). The proof combines Bushnell-Kutzko types, Paškūnas' unicity theorem for supercuspidal components, the Iwasawa decomposition to restrict parabolic induction to K_n, a filtration by the compact subgroups P_I(m), and an induction on n; the cases n_r=1 (Borel) and n_r>1 are treated separately, with Clifford theory and finite-field lemmas in the latter.","tokens_in":20278,"tokens_out":11435,"duration_ms":124807,"significance":"If the main theorem is correct, the paper completes the level-zero classification of typical representations, extends the Henniart and Paškūnas results for cuspidal components to all level-zero Bernstein components, and provides a multiplicity formula. The strategy is well chosen and the paper is honest about relying on external inputs: Bushnell-Kutzko types, Paškūnas' unicity, Casselman's restriction theorem, and Zelevinsky's derivative formalism are cited explicitly, and I see no fitted parameters or circularity. However, two steps in the n_r>1 branch of the proof are not justified as written; both are load-bearing for Theorem 3.2 and Corollary 3.3, so the main claim is not yet established in the present text.","major_comments":[{"comment":"The proof of Proposition 4.3 asserts that the representation in Eq. (22), namely ind_{Z(η_k)}^{K_{n+1}}(U_{η_k} ⊗ res_{Z∩M_I} τ'_I), 'occurs as a subrepresentation of' ind_{P_I∩K_{n+1}}^{K_{n+1}}(τ'_I). This containment is the only bridge from the Clifford-theoretic decomposition to parabolic induction, and it is not a formal consequence of transitivity of induction or Frobenius reciprocity. The subgroup Z(η_k) contains K_I(m)∩\\bar U, on which the Clifford character U_{η_k} is nontrivial, while the target representation is trivial on the lower unipotent part, and Z(η_k) is not contained in P_I∩K_{n+1}. An embedding therefore requires an explicit Mackey double-coset computation over Z(η_k)\\K_{n+1}/(P_I∩K_{n+1}), or an equivalent argument, and none is supplied. Since Proposition 4.3 is the decisive step for all level-zero components with r>1 and n_r>1, Theorem 3.2 and Corollary 3.3 are unsupported for that case until this containment is proved.","section":"4, Proposition 4.3, Eq. (22)"},{"comment":"In the proof of Lemma 3.8, the passage from Eq. (9) to the equation involving M_{ll} drops the off-diagonal terms M_{lj}U_j^{tr} for j<l. Since M is block upper triangular, the l-th block of Eq. (9) is Σ_{j≤l} M_{lj}U_j^{tr} = U_l^{tr}B, and these lower-block terms need not vanish even when l is chosen maximal with U_l≠0. Consequently the claim that M_{ll} preserves ker T1 (or that B preserves ker T2) is not established as written. This lemma is used in Proposition 4.3 to produce the unipotent subgroup U with H∩U={id}, so the gap affects the same load-bearing step of the main theorem.","section":"3.1, Lemma 3.8"}],"minor_comments":[{"comment":"In the definition of U_I(R), 'unipotent unipotent matrices' should read 'unipotent matrices'.","section":"2.1"},{"comment":"In the proof of Lemma 3.9, the symbol γ in ind_H^G(γ) is undefined; it should be the representation ξ, and the intended Mackey-decomposition argument should be spelled out.","section":"3, Lemma 3.9"},{"comment":"The notation for the last character wavers between χ_n and χ_{n+1}; for example the summand U_m(χ_{I_n})⊠χ_n should presumably be U_m(χ_{I_n})⊠χ_{n+1}.","section":"4, n_r=1 case"},{"comment":"Eq. (22) is referred to in the text as 'The representation 22'; the equation number should be typeset consistently.","section":"4, Proposition 4.3"}],"recommendation":"major_revision","confidential_remarks":"The decisive issue is the unproved containment in Proposition 4.3 and the related gap in Lemma 3.8. If the author can supply a correct Mackey double-coset proof or an alternative argument, the paper may well be acceptable. In its current form, however, the n_r>1 case of the main theorem is not established, so I cannot recommend acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague:\n\nThe paper claims a complete classification of typical representations for all level-zero Bernstein components of GL_n(F). If correct, that is a real advance on Paskunas (cuspidal case) and Henniart (n=2). The overall strategy is sound: Bushnell–Kutzko types, a complement lemma to isolate candidates, then an induction splitting nr=1 and nr>1. The nr=1 case is worked out in genuine detail using Zelevinsky derivatives and Casselman's GL2 restriction theorem; I was impressed by that section. The multiplicity statement in Corollary 3.3 is a useful extra.\n\nThe problem is the nr>1 case. Proposition 4.3 reduces atypicality to the assertion that the Clifford piece ind_{Z(η_k)}^{K_{n+1}}(U_{η_k} ⊗ res τ'_I) embeds as a subrepresentation of ind_{P_I∩K}^{K_{n+1}} τ'_I. That containment is simply asserted. It is not a formal consequence of transitivity or Frobenius reciprocity. The subgroup Z(η_k) is not contained in P_I∩K, and U_{η_k} is nontrivial on the lower unipotent subgroup, where the parabolically induced representation is trivial. You need a Mackey double-coset computation or an intertwining argument, and the paper supplies none. In the finite-group analogue, inducing a nontrivial unipotent character from a parabolic stabilizer gives the Gelfand–Graev representation, which does not sit inside a single parabolically induced representation. So this is not a minor missing detail; without it, Theorem 3.2 and Corollary 3.3 are unsupported for all components with r>1 and n_r>1.\n\nThe rest of the paper is honest. The major inputs—Bushnell–Kutzko types, Paskunas unicity, Casselman’s theorem, Zelevinsky derivatives—are all cited, there are no fitted parameters, and the structural limits are visible inside the proof rather than hidden. The dependence on Paskunas unicity is external, not circular. The writing is dense but mostly clear.\n\nWho is this for? Specialists in p-adic representation theory who work on types and the Breuil–Mézard conjecture. The paper deserves a serious referee: the question is important and most of the machinery is probably salvageable. But the referee should demand a proof of the containment in Proposition 4.3 before accepting. As it stands, I would not cite the main classification in my own work.\n\nRecommendation: send to peer review with an explicit request to repair or justify Proposition 4.3.","headline":"A plausible and well-organized extension of the level-zero typical-representation classification, but the nr>1 case relies on an unproved containment in Proposition 4.3 that the referee must force the author to justify.","tokens_in":20838,"tokens_out":2091,"would_cite":false,"duration_ms":20855,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["22E50","11S37","22E35"],"pacs":[],"model":"deepseek-v4-flash","headline":"For a level-zero Bernstein component of $\\mathrm{GL}_n(F)$, the typical representations are exactly the irreducible subrepresentations of an explicit induced type.","keywords":["typical representation","Bernstein component","level zero","Bushnell-Kutzko type","supercuspidal representation","cuspidal support","parabolic induction","GL_n"],"falsifier":"Take a level-zero component $s=[M_I,\\sigma_I]$ for a small $n$, list all irreducible subrepresentations of $\\operatorname{ind}_{P_I(1)}^{K_n}(\\tau_I)$, and check whether any other irreducible $K_n$-type is typical for $s$ by testing whether it occurs in $\\operatorname{res}_{K_n}i_{P_I}^{\\mathrm{GL}_n(F)}(\\sigma_I)$; finding such a $\\Gamma$ outside the listed set, or finding one whose multiplicity in the two sides of Corollary 3.3 differs, would refute the classification.","tokens_in":19708,"feed_emoji":"🎯","tokens_out":11880,"duration_ms":125720,"temperature":0.7,"pith_summary":"This paper classifies the typical representations for level-zero Bernstein components of a non-Archimedean local field $F$. A typical representation is an irreducible representation of the maximal compact subgroup $\\mathrm{GL}_n(O_F)$ whose appearance inside an irreducible smooth representation $\\pi$ forces $\\pi$ to have a prescribed cuspidal support. The main theorem states that for a level-zero component $s=[M_I,\\sigma_I]$, every typical representation occurs inside the induced representation $\\operatorname{ind}_{P_I(1)}^{\\mathrm{GL}_n(O_F)}(\\tau_I)$ built from the level-one Bushnell--Kutzko type; conversely, every irreducible subrepresentation of that induced representation is typical. The paper also proves that the multiplicity of each typical representation in this induced representation equals its multiplicity in the corresponding parabolic induction, and the result holds independently of the characteristic of the base field.","feed_headline":"Level-zero typical representations sit inside one induced type","feed_subtitle":"The paper pinpoints every K_n-representation detecting a level-zero Bernstein component, with exact multiplicities.","key_machinery":"The carrying object is the induced representation $\\operatorname{ind}_{P_I(1)}^{K_n}(\\tau_I)$ and its filtration by $\\operatorname{ind}_{P_I(m)}^{K_n}(\\tau_I)$, where $P_I(m)$ is the preimage of $P_I(O_F/P_F^m)$ in $K_n$. Iwahori decomposition lets $P_I(m)\\cap U_I$ and $P_I(m)\\cap \\overline{U}_I$ act trivially, so Lemma 2.5 expresses the full induced representation as a union of finite-dimensional pieces. The proof decomposes each successive quotient by Clifford theory into the identity character plus non-trivial characters $\\eta$, whose stabilizers $Z(\\eta)$ are controlled by a matrix-stabilizer lemma over the residue field; Lemma 3.9 then converts a trivial intersection with a unipotent group into the existence of a non-cuspidal representation containing the given piece. In the rank-one case $n_r=1$, Zelevinsky's derivative functors and Casselman's restriction decomposition for $\\mathrm{GL}_2$ are used. Together these force all subrepresentations of the complement to be atypical, which is the engine behind the classification.","core_discovery":"The paper's central claim, Theorem 1.1 and Corollary 3.3, is that for a level-zero inertial class $s=[M_I,\\sigma_I]$ with ordered partition $I=(n_1,\\dots,n_r)$ of $n$, the $K_n=\\mathrm{GL}_n(O_F)$-irreducible subrepresentations of $\\operatorname{ind}_{P_I(1)}^{K_n}(\\tau_I)$ are precisely the typical representations for $s$. Here $\\tau_I=\\tau_1\\boxtimes\\cdots\\boxtimes\\tau_r$, where each $\\tau_i$ is the unique typical representation attached to the supercuspidal component $[G_{n_i},\\sigma_i]$, realized as the inflation of a cuspidal representation of $\\mathrm{GL}_{n_i}(k_F)$. The proof shows that the complement $U_m(\\tau_I)$ in $\\operatorname{ind}_{P_I(m)}^{K_n}(\\tau_I)$ contains only atypical subrepresentations, so no typical representation can hide outside the level-one induced representation. In addition, for every typical $\\Gamma$, the paper proves $\\dim_{\\mathbb{C}}\\operatorname{Hom}_{K_n}(\\Gamma,\\operatorname{ind}_{P_I(1)}^{K_n}(\\tau_I))=\\dim_{\\mathbb{C}}\\operatorname{Hom}_{K_n}(\\Gamma,i_{P_I}^{\\mathrm{GL}_n(F)}(\\sigma_I))$.","pith_inferences":["Editor's inference: the explicit finite list of typical representations should make the level-zero inertial-support detection problem effectively computable for small $n$, for example by tabulating the irreducible subrepresentations of $\\operatorname{ind}_{P_I(1)}^{K_n}(\\tau_I)$ for each residue cardinality.","Editor's inference: if the complement argument could be reworked without invoking the cuspidal unicity theorem, the classification would likely have a natural analogue for positive-level components where typical representations are not known to be unique.","Editor's inference: the equality of multiplicities suggests that typical representations may index a natural basis of the space of $K_n$-fixed vectors in parabolic induction, possibly visible through a Hecke-module structure attached to the type."],"forward_implications":["For every level-zero Bernstein component $s$, the set of typical representations is finite: it is exactly the finite set of irreducible subrepresentations of $\\operatorname{ind}_{P_I(1)}^{K_n}(\\tau_I)$.","A $K_n$-representation is typical for $s$ exactly when it appears in this single explicitly given induced representation, so inertial-support detection for level-zero components reduces to listing that finite set.","The multiplicity identity means the numerical data carried by typical representations over $K_n$ coincide with embeddings of $\\sigma_I$ in parabolic induction, allowing computations on the compact group to be translated into parabolic-induction computations.","Because the result is independent of the characteristic of $F$, it covers both characteristic-zero local fields and equal-characteristic local fields such as Laurent series fields over finite fields."],"supporting_citations":[{"why":"Builds the general theory of types for $\\mathrm{GL}_n(F)$, guaranteeing the existence of a Bushnell--Kutzko type $(J_s,\\lambda_s)$ for every Bernstein component.","marker":"[BK93]"},{"why":"Supplies the explicit level-zero type $(P_I(1),\\tau_I)$ used throughout the paper, via Section 8.3.1.","marker":"[BK99]"},{"why":"Provides the unicity of typical representations for supercuspidal components, which is the inductive base and the input to Lemma 2.4; also the source of Lemma 3.9.","marker":"[Pas05]"},{"why":"Classifies the irreducible subrepresentations of the induced type, framing the set that the paper proves is exactly the typical representations.","marker":"[SZ99]"},{"why":"Gives the $n=2$ classification of typical representations, which is used as the base case in the rank-one induction.","marker":"[BM02]"},{"why":"Observed that the induced typical representation admits a complement whose irreducible subrepresentations are atypical; Proposition 2.3 generalizes this observation.","marker":"[Wil10]"},{"why":"Provides the derivative functors and filtration over finite fields used in the rank-one case to exhibit atypicality.","marker":"[Zel81]"},{"why":"Gives the restriction decomposition of an irreducible smooth $\\mathrm{GL}_2(F)$-representation to $\\mathrm{GL}_2(O_F)$, used to handle the last term in the rank-one proof.","marker":"[Cas73b]"},{"why":"Provides the construction of level-zero supercuspidal representations of $\\mathrm{GL}_2$ with prescribed central character, needed in the rank-one atypicality argument.","marker":"[BH06]"}],"fun_headline_variants":["All level-zero typical reps arise from one induced type","Exact typical representations for level-zero Bernstein components","Level-zero typical reps completely classified via induced types","Typical reps for level-zero components: full classification","Every level-zero typical rep sits inside the same induced type"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument leans on the previously proved uniqueness of typical representations for the basic cuspidal pieces, a theorem this paper cites rather than proves; if that uniqueness failed, the complement decomposition driving the induction would break.","fun_headline_variants_meta":{"raw":{"variants":["All level-zero typical reps arise from one induced type","Exact typical representations for level-zero Bernstein components","Level-zero typical reps completely classified via induced types","Typical reps for level-zero components: full classification","Every level-zero typical rep sits inside the same induced type"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000445,"raw_usage":{"total_tokens":2248,"prompt_tokens":944,"completion_tokens":1304,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":560,"completion_tokens_details":{"reasoning_tokens":1229}},"tokens_in":560,"tokens_out":1304,"duration_ms":10606,"temperature":1.0,"reasoning_tokens":1229,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:18:26.929185+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a level-zero component $s=[M_I,\\sigma_I]$ for a small $n$, list all irreducible subrepresentations of $\\operatorname{ind}_{P_I(1)}^{K_n}(\\tau_I)$, and check whether any other irreducible $K_n$-type is typical for $s$ by testing whether it occurs in $\\operatorname{res}_{K_n}i_{P_I}^{\\mathrm{GL}_n(F)}(\\sigma_I)$; finding such a $\\Gamma$ outside the listed set, or finding one whose multiplicity in the two sides of Corollary 3.3 differs, would refute the classification.","supporting_citations":[],"review_version":1}