REVIEW 3 major objections 4 minor 30 references
On the complementary Arthur representations and unitary dual for p-adic classical groups
T0 review · 3 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash
Pith's one-line read For p-adic symplectic and split odd orthogonal groups, a complementary Arthur representation is unitary precisely when each reducible generalized Speh factor occurs with even multiplicity.
desk verdict A substantial conjecture proven with a credible strategy, but the combinatorial core leaves enough 'direct computations' omitted that the proof is conditional as written. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The central object is the extended Z-segment, a triple $([A,B],l,\eta)$ where $[A,B]$ is a consecutive block of integers, $l$ is a nonnegative integer at most half the block length, and $\eta$ is a sign (with a relation when $l$ equals half the length). These segments parameterize the rows of an extended multi-segment, and the paper defines an interval as a consecutive set of such segments under a total order, with two intervals adjacent when they meet at a boundary. The key machinery is the non-vanishing set $NV(e_1,-)(\Delta_2)$ of segments $e'$ for which the ordered pair $(e_1,e')$ satisfies the combinatorial non-vanishing criterion, together with the row-exchange operator $R$ that swaps comparable segments. Lemma 4.10 shows these sets are intervals and that adjacency is preserved; this lets the proof decompose the unitary induction $u_\rho(a,b)\rtimes\pi(E)$ into summands whose component-group characters alternate sign, so a reducible induction produces both characters and hence a non-scalar intertwining operator.
What would settle it
A single explicit adjacent pair $e_1,e'_1$ and a segment $\Delta_2$ for which $NV(e_1,-)(\Delta_2)$ and $NV(e'_1,-)(\Delta_2)$ are not adjacent would contradict Lemma 4.10(ii). Since Theorem 6.6, and then the non-unitarity direction of Theorem 6.3, rests directly on that statement, such a pair would invalidate the proof of the parity criterion.
Extended reading notes
Core claim
Theorem 1.3 (announced as Theorem 6.3) asserts that for $G_n=\mathrm{Sp}_{2n}(F)$ or $\mathrm{SO}_{2n+1}(F)$, a representation $\pi\in\Pi_{\psi}$ with $\psi\in\Psi^+_{\mathrm{unit}}(G_n)$ is unitary exactly when $\pi$ lies in $\Sigma_{A+,u}(G_n)$. In concrete terms, decomposing $\pi=\times_{i\in I_{\mathrm{nu}}} u_{\rho_i}(a_i,b_i)|\cdot|^{x_i}\rtimes\pi_A$, unitarity is equivalent to the parity condition: for each $i$ such that $u_{\rho_i}(a_i,b_i)\rtimes\pi_A$ is reducible, the number of $j$ with $\rho_j\cong\rho_i$, $a_j=a_i$, and $b_j=b_i$ is even. If all such inductions are irreducible, the representation is automatically unitary.
Load-bearing premise
The load-bearing premise is that the non-vanishing sets $NV(e_1,-)(\Delta_2)$ and $NV(-,e_2)(\Delta_1)$ are always intervals and that adjacency of extended Z-segments is preserved under these set-valued maps and under row exchanges; the text verifies this through case checks, several of which are described as straightforward or omitted.
Editorial extensions
If this is right
- For $G_n=\mathrm{Sp}_{2n}(F)$ or $\mathrm{SO}_{2n+1}(F)$, the unitary complementary Arthur representations are exactly $\Sigma_{A+,u}(G_n)$, so $\Pi_{A+,u}(G_n)=\Sigma_{A+,u}(G_n)$.
- Every local component at a finite place of a discrete automorphic representation of split $\mathrm{Sp}_{2n}$ or $\mathrm{SO}_{2n+1}$ obeys the parity constraint, with no appeal to the generalized Ramanujan conjecture.
- For irreducible self-dual cuspidal automorphic representations of $\mathrm{GL}_N$ of orthogonal or symplectic type, any non-tempered Speh factor of the corresponding type that is not in the tempered part must occur with even multiplicity among the non-tempered factors.
- In the low-rank cases $N=2$ and $N=3$, this forces temperedness: orthogonal-type $\mathrm{GL}_2$ cuspidal representations have tempered local components at every finite place, and self-dual ramified $\mathrm{GL}_3$ local components of the form $\chi|\cdot|^x \times 1 \times \chi|\cdot|^{-x}$ must have $x=0$ unless $\chi=1$.
- As a consequence, the closure of Arthur representations equals the unitary Arthur representations for these groups, $\Pi_{A}(G_n)=\Pi_{A+,u}(G_n)$, completing one step toward the unitary-dual conjecture.
Reading between the lines
- The interval/adjacency calculus is likely reusable: any family of representations whose reducibility is read off from extended multi-segments, not only Arthur packets of symplectic and odd-orthogonal groups, may obey the same parity principle as long as the non-vanishing sets are intervals.
- If a proof of Lemma 4.10 can be given by explicit closed formulas instead of omitted case checks, the non-unitarity direction would become more robust and the criterion would be easier to verify computationally for larger parameters.
- For self-dual cuspidal automorphic representations of $\mathrm{GL}_N$ with $N>3$, the same localization argument should force even-multiplicity constraints on more non-tempered forms; the $N=4$ list in the paper is a natural place to test the pattern.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper proves an explicit characterization of the unitary complementary Arthur representations for split symplectic groups Sp_2n(F) and split odd special orthogonal groups SO_{2n+1}(F) over a non-Archimedean local field F of characteristic zero. The main theorem (Theorem 6.3) states that a representation in the enlarged Arthur-type set Π_{A+}(G_n) is unitary if and only if it lies in the set Σ_{A+,u}(G_n), i.e. the multiplicities of certain generalized Speh factors whose induced representation is reducible are even. The proof combines Arthur's local intertwining relation, Mœglin's construction of Arthur packets as reformulated by Atobe, a non-unitarity criterion of Muić and Tadić, and a new combinatorial theory of intervals and adjacency on extended Z-segments developed in Section 4. As applications, the paper derives constraints on the local components of irreducible self-dual cuspidal automorphic representations of GL_N, with concrete consequences for N=2,3,4.
Significance. If the proof is correct, the paper proves Conjecture 1.2 from the authors' program [HJLLZ24], giving a precise description of Π_{A+,u}(G_n) for the two largest families of split classical groups. This is a substantive step toward the broader conjecture that the unitary dual is the closure of the Arthur-type representations. The paper also yields falsifiable constraints on automorphic local components that do not require the generalized Ramanujan conjecture, which is a genuinely useful application. The combinatorial framework of intervals and adjacency on extended Z-segments is a new technical contribution that is likely to be reusable. However, the decisive combinatorial lemmas in Section 4 are not fully proved in the manuscript; several key statements are dismissed as straightforward or deferred to omitted direct computations. Since these lemmas are load-bearing for the non-unitarity direction of Theorem 6.3, the present version is conditional on their verification.
major comments (3)
- [Lemma 4.10, §4.2] Lemma 4.10(ii) and the NV(−,e2) half of (i)–(iii) are not proved. The proof explicitly says 'Part (ii) follows from a case-by-case straightforward computation, which we omit' and 'We omit the analogous verification of these statements for NV(−,e2)(Δ1)'; Lemma 4.10(iii) Case (c) is closed with 'We omit the rest of the verification, which is similar to Case (b).' These statements are not optional: Lemma 4.10 is used in the proof of Proposition 4.20(1) and hence in Theorem 6.6, which underlies the base case of the non-unitarity argument. The manuscript itself flags these as omissions, so the proof of Theorem 6.3 is conditional on assertions that remain unchecked.
- [Lemmas 4.17 and 4.19, §4.4] The proof of Lemma 4.17, after reducing to the case e3 = ([A3,A3],0,η3), ends with 'the proof proceeds by a direct computation, which we omit.' Lemma 4.19(1)-(3) are all deferred with the same phrase: 'These statements follow from direct computations, which we omit.' These lemmas are used in the proof of Proposition 4.20 through Observations (i) and (ii), and Proposition 4.20(2) is invoked by Theorem 6.8. Thus the induction step for the non-unitarity direction of Theorem 6.3 rests on unverified combinatorial identities. A rigorous submission needs either complete proofs of these case checks or a machine-checkable verification.
- [Proposition 4.20(2), §4.4] In the proof of Proposition 4.20(2), the text states: 'It is possible that [Ee,e′]†† is empty' and then immediately asserts 'Suppose [Ee,e′]†† is empty for some (equivalently, for all) (e,e′) ∈ NVE(S) × NVE(S′).' The asserted equivalence is not justified. If emptiness occurs only for some pairs, the contradiction derived from Conditions (b) and (c) does not follow, and one would need an additional argument to reduce to the non-empty case. Since Proposition 4.20(2) is essential for Theorem 6.8, this is a second load-bearing gap in the same combinatorial engine.
minor comments (4)
- [§6.3] In the proof of Theorem 6.3, the text 'for any i ∈ IInu,nsd' has an extra 'I': it should be 'i ∈ Inu,nsd'.
- [Corollary 6.7] The final sentence of the proof says 'This gives a contradiction to the existence of Π + k−1 or Π + k−1'; the second expression should be Π − k−1.
- [Lemma 4.6] Lemma 4.6(c) is used in Proposition 4.20(2), but its proof is only one line: 'These are straightforward consequences of the definitions.' Since the proposition is load-bearing, expanding the proof of Lemma 4.6(c) would improve verifiability.
- [Definition 4.7(2)] The definition of ǫ in Definition 4.7(2) uses a lift ηi ∈ {±1}; Remark 4.8(1) asserts independence of the lift, but an explicit verification would help the reader, especially because Lemma 4.10 relies on this convention.
Circularity Check
No circular derivation: Theorem 6.3 is proved from Arthur's intertwining relations, Atobe's packet decompositions, and Mœglin's irreducibility results; the authors' earlier conjectures are quoted as motivation and application, not as proof inputs.
full rationale
I walked the derivation chain for Theorem 6.3. The unitary characterization is not assumed: Conjecture 1.2 from [HJLLZ24] is quoted as an open conjecture and then proved via a base case (Theorem 3.2 and Corollary 3.3, using Arthur's (A-LIR) and the Muić-Tadić criterion Lemma 3.1) and a reduction (§6.3) that uses Atobe's decomposition Theorem 5.12, Mœglin's irreducibility and multiplicity-freeness, Tadić's unitary classification, and the combinatorial Proposition 4.20. The equality Π_{A+,u}(Gn)=Σ_{A+,u}(Gn) is not definitional: Σ is a parity/reducibility condition while Π_{A+,u} is defined as Π_{A+}∩Π_u, so the theorem supplies real content in both directions. No fitted parameter is renamed as a prediction, and no uniqueness theorem from the authors' prior work is invoked to force the choice. Self-citations to [HJLLZ24] appear in the introduction (Conjectures 1.1 and 1.2, Example 5.1, Theorem 1.7) and are not used as inputs in Sections 4–6; the central result is derived from external results by Arthur, Mœglin, Tadić, and Atobe. The text itself flags omitted case checks in Lemmas 4.10(ii)/(iii), 4.17, and 4.19; that makes the submitted proof conditional on those combinatorial verifications, but it is a completeness/correctness concern, not a circularity. I therefore find no significant circularity.
Assumptions & free parameters
assumptions (6)
- standard math Local Langlands correspondence for GL_n over non-Archimedean local fields identifies bounded L-parameters with irreducible unitary supercuspidal representations.
- domain assumption Arthur's endoscopic classification for quasi-split symplectic and odd orthogonal groups: local Arthur packets, component groups, and the local intertwining relation [Art13, Theorems 2.2.1 and 2.4.1].
- domain assumption Moeglin's construction of local Arthur packets by extended multi-segments, including irreducibility in Theorem 2.1 and multiplicity freeness ([Moe11b, Proposition 5.1], [Moe11]).
- domain assumption Tadic's classification of the unitary dual of GL_n ([Tad86]) and the structure theorem for weakly real representations ([Tad09, Theorem 4.2]).
- domain assumption Atobe's reformulation: row exchanges on extended multi-segments preserve the associated representation ([Ato22a, Theorem 4.3]) and the decomposition theorem for unitary inductions ([Ato22b, Theorem 4.4]).
- domain assumption Xu's nonvanishing theorem [Xu21a, Theorem A.3] for Moeglin's parametrization.
Cite this review
Pith. "Pith review of On the complementary Arthur representations and unitary dual for p-adic classical groups." pith.science (2026). https://pith.science/paper/S5TYSXI3
@misc{pith2026250511381,
author = {Pith},
title = {Pith review of: On the complementary Arthur representations and unitary dual for p-adic classical groups},
year = {2026},
howpublished = {\url{https://pith.science/paper/S5TYSXI3}},
note = {Machine review of arXiv:2505.11381}
}
abstract
In [HJLLZ24], we proposed a new conjecture on the structure of the unitary dual of connected reductive groups over non-Archimedean local fields of characteristic zero based on their Arthur representations and verified it for all the known cases on the unitary dual problem. One step towards this conjecture involves the question whether certain complementary Arthur representations are unitary. In this paper, we give an explicit characterization of the complementary Arthur representations for symplectic and split odd special orthogonal groups. As applications, we obtain interesting constraints on local components of irreducible self-dual cuspidal automorphic representations of $\mathrm{GL}_N$, especially when $N=2,3$.
Reference graph
Works this paper leans on
-
[1]
Arthur, Unipotent automorphic representations: conjectures
J. Arthur, Unipotent automorphic representations: conjectures. Ast \'e risque . pp. 13-71 (1989), Orbites unipotentes et repr \' e sentations, II
1989
-
[2]
Arthur, The endoscopic classification of representations: Orthogonal and Symplectic groups
J. Arthur, The endoscopic classification of representations: Orthogonal and Symplectic groups. Colloquium Publication Vol. 61, 2013, American Mathematical Society
2013
-
[3]
Atobe, Construction of local A-packets
H. Atobe, Construction of local A-packets. J. Reine Angew. Math. , vol. 2022, no. 790, 2022, pp. 1-51
work page 2022
-
[4]
Atobe, On the socles of certain parabolically induced representations of p-adic classical groups
H. Atobe, On the socles of certain parabolically induced representations of p-adic classical groups. Represent. Theory . 26 pp. 515-541 (2022)
2022
-
[5]
H. Atobe, The set of local A-packets containing a given representation , Journal f\"ur die reine und angewandte Mathematik (Crelles Journal), vol. 2023, no. 804, (2023), pp. 263-286
work page 2023
- [6]
-
[7]
H. Atobe and A. M \'i nguez, Unitary dual of p -adic split _ 2n+1 and _ 2n : The good parity case (and slightly beyond). Preprint. 2025. arXiv:2505.09991
arXiv 2025
-
[8]
B. Bo s njak and A. Stadler, Irreducibility of the parabolic induction of essentially Speh representations and a representation of Arthur type over a p -adic field. Preprint. 2025. arXiv:2504.01226
Show all 30 references
-
[9]
W. T. Gan, B. H. Gross, and D. Prasad, Symplectic local root numbers, central critical L values, and restriction problems in the representation theory of classical groups. Ast \'e risque . pp. 1-109 (2012), Sur les conjectures de Gross et Prasad. I
2012
-
[10]
Harris and R
M. Harris and R. Taylor, The geometry and cohomology of some simple Shimura varieties. (Princeton University Press, 2001)
2001
-
[11]
Hazeltine, D
A. Hazeltine, D. Jiang, B. Liu, C.-H. Lo, and Q. Zhang, Arthur representations and unitary dual for classical groups , arXiv:2410.11806
-
[12]
Hazeltine, B
A. Hazeltine, B. Liu, and C.-H. Lo, On the local theta correspondence and intersection of local Arthur packets for even orthogonal groups. Preprint. 2025
2025
-
[13]
Henniart, Une preuve simple des conjectures de Langlands pour GL(n) sur un corps p-adique
G. Henniart, Une preuve simple des conjectures de Langlands pour GL(n) sur un corps p-adique. Invent. Math. 139 (2000), 439-455
2000
-
[14]
Jantzen, On supports of induced representations for symplectic and odd-orthogonal groups
C. Jantzen, On supports of induced representations for symplectic and odd-orthogonal groups. Amer. J. Math. 119, 1213-1262 (1997)
1997
-
[15]
Lapid, G
E. Lapid, G. Mui \'c , M. Tadi \'c . On the generic unitary dual of quasisplit classical groups. Int. Math. Res. Not. 26, 1335–1354 (2004)
2004
-
[16]
M glin, Paquets d'Arthur pour les groupes classiques; point de vue combinatoire
C. M glin, Paquets d'Arthur pour les groupes classiques; point de vue combinatoire. (2006), arXiv:math/0610189v1
2006 arXiv
-
[17]
M glin , Sur certains paquets d'Arthur et involution d'Aubert-Schneider-Stuhler g \'e n \'e ralis \'e e
C. M glin , Sur certains paquets d'Arthur et involution d'Aubert-Schneider-Stuhler g \'e n \'e ralis \'e e. Represent. Theory 10 , (2006), 86-129
2006
-
[18]
M glin, Formes automorphes de carr \'e int \'e grable non cuspidales
C. M glin, Formes automorphes de carr \'e int \'e grable non cuspidales. Manuscripta Math. 127 (2008), 411-467
2008
-
[19]
M glin , Paquets d'Arthur discrets pour un groupe classique p -adique
C. M glin , Paquets d'Arthur discrets pour un groupe classique p -adique. Automorphic forms and L-functions II. Local aspects , 179-257, Contemp. Math., 489 , Israel Math. Conf. Proc., Amer. Math. Soc., Providence, RI , (2009)
2009
-
[20]
M glin , Holomorphie des op \'e rateurs d'entrelacement normalis \'e s \`a l'aide des param \`e tres d'Arthur
C. M glin , Holomorphie des op \'e rateurs d'entrelacement normalis \'e s \`a l'aide des param \`e tres d'Arthur. Canad. J. Math. 62 (2010), no. 6, 1340-1386
2010
-
[21]
M glin , Multiplicit \'e 1 dans les paquets d'Arthur aux places p -adiques
C. M glin , Multiplicit \'e 1 dans les paquets d'Arthur aux places p -adiques. On certain L -functions, 333-374, Clay Math. Proc., 13 , Amer. Math. Soc., Providence, RI, (2011)
2011
-
[22]
M glin, Image des op \'e rateurs d'entrelacements normalis \'e s et p \^ o les des s \'e ries d'Eisenstein
C. M glin, Image des op \'e rateurs d'entrelacements normalis \'e s et p \^ o les des s \'e ries d'Eisenstein. Adv. Math. 228 (2011), 1068-1134
2011
-
[23]
Mui \'c , M
G. Mui \'c , M. Tadi \'c . Unramified unitary duals for split classical p-adic groups; the topology and isolated representations. On certain L-functions . Clay Math. Proc., vol. 13, pp. 375-438. Amer. Math. Soc. , Providence (2011)
2011
-
[24]
Sarnak, Notes on the generalized Ramanujan conjectures
P. Sarnak, Notes on the generalized Ramanujan conjectures. Harmonic analysis, the trace formula, and Shimura varieties , 659-685. Clay Math. Proc., 4 American Mathematical Society, Providence, RI, 2005
2005
-
[25]
Scholze, The local Langlands correspondence for _n over p-adic fields
P. Scholze, The local Langlands correspondence for _n over p-adic fields. Invent. Math. 192, (2013), 663-715
2013
-
[26]
Tadi \'c , Classification of unitary representations in irreducible representations of general linear group (non-Archimedean case)
M. Tadi \'c , Classification of unitary representations in irreducible representations of general linear group (non-Archimedean case). Ann. Sci. \'E cole Norm. Sup. (4)19(1986), no.3, 335-382
1986
-
[27]
Tadi \'c , An external approach to unitary representations
M. Tadi \'c , An external approach to unitary representations. Bull. Amer. Math. Soc. (N.S.) . 28, 215-252 (1993)
1993
-
[28]
Tadi \'c , On reducibility and unitarizability for classical p-adic groups, some general results
M. Tadi \'c , On reducibility and unitarizability for classical p-adic groups, some general results. Canad. J. Math. 61, 427-450 (2009)
2009
-
[29]
Xu, Nonarchimedean components of non-endoscopic automorphic representations for quasisplit Sp(N) and O(N)
B. Xu, Nonarchimedean components of non-endoscopic automorphic representations for quasisplit Sp(N) and O(N) . Math. Z. 297, 885-921 (2021)
2021
-
[30]
Xu, A combinatorial solution to M glin's parametrization of Arthur packets for p -adic quasisplit Sp(N) and O(N)
B. Xu, A combinatorial solution to M glin's parametrization of Arthur packets for p -adic quasisplit Sp(N) and O(N) . J. Inst. Math. Jussieu . 20, 1091-1204 (2021)
2021
Reviewed August 15, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.