REVIEW 3 major objections 6 minor 12 references
Composition series of a class of induced representations built on discrete series
T0 review · 3 major / 6 minor · reviewed 2026-08-14 · deepseek-v4-flash
Pith's one-line read The paper establishes an explicit multiplicity-free composition series, with $3^k$ terms indexed by subsets of the segments, for a class of induced representations built on discrete series of classical p-adic groups.
desk verdict The main composition-series formula looks right, but the advertised counts 2^k and 3^k are false when any segment has b_i=-1/2, a case the paper explicitly allows; the fix is to restate the counts with 2^a 3^b. 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 engine of the proof is a family of intertwining operators that permute the segments $\Delta_i$ and eventually replace them by their contragredient segments $\widetilde{\Delta}_i$. For each $l$, the image $V_l$ of the sum of intertwinings from the representations attached to subsets $X$ of size $l$ gives a filtration $\{0\}=V_{-1}\subseteq V_0\subseteq\cdots\subseteq V_k=\prod_{i\in S}\delta(\Delta_i)\rtimes\sigma$, and Theorem 4.1 identifies each quotient $V_l/V_{l-1}$ with the direct sum of the Langlands quotients indexed by subsets $X$ of size $l$. The irreducibility condition (C2) is what makes the segment permutations isomorphisms and keeps the counting of subrepresentations exact; the Tadić formula for Jacquet modules is then used to show each candidate Langlands quotient occurs with multiplicity one.
What would settle it
A direct calculation for $k=2$ with one explicit segment family satisfying (C1)–(C2) would settle the claim: use the Tadić formula (2.1) to compute the semisimplification of $\delta(\Delta_1)\times\delta(\Delta_2)\rtimes\sigma$ and check that it has exactly nine irreducible subquotients, each with multiplicity one and the predicted Langlands quotient; a different length or a repeated quotient would disprove Theorem 4.1.
Extended reading notes
Core claim
The central claim is that a large class of induced representations of classical $p$-adic groups—those built by parabolically inducing a product of essentially square-integrable general-linear representations onto a discrete series representation—has a fully explicit, multiplicity-free composition series. If $\sigma$ is a discrete series obtained by extending a strongly positive discrete series $\sigma_{\mathrm{sp}}$ along segments satisfying the conditions (C1) and (C2), then for any further such family $\{\Delta_i: i \in S\}$ the representation $\prod_{i \in S}\delta(\Delta_i)\rtimes\sigma$ has exactly $3^{|S|}$ irreducible subquotients. In the Grothendieck group it equals $$\sum_{X \subseteq S}\sum_{\$\sigma$' \hookrightarrow \prod_{j \in S\setminus X}\delta(\Delta_j)\rtimes\$\sigma$}\mathrm{Lang}\Bigl(\prod_{i \in X}\delta(\Delta_i)\rtimes\$\sigma$'\Bigr),$$ and these Langlands quotients occur as the successive quotients of an explicit filtration by images of intertwining operators. The paper also derives the decomposition of the same inductions when the starting piece is a Langlands quotient rather than a discrete series.
Load-bearing premise
The argument stands or falls on the irreducibility condition (C2) for every pair of segments—both $\delta(\Delta_i)\times\delta(\Delta_j)$ and $\delta(\widetilde{\Delta}_i)\times\delta(\Delta_j)$ irreducible for $i\neq j$—together with the reduction in Proposition 2.6 asserting that every discrete series can be brought into this situation, a step the paper sketches rather than proves in detail.
Editorial extensions
If this is right
- In the irreducible case the standard representation $\delta(\Delta_1)\times\cdots\times\delta(\Delta_k)\rtimes\sigma_{\mathrm{sp}}$ has length $3^k$, so its semisimplification can be written down directly from the formula without recursion.
- Every irreducible subrepresentation of the induction is a discrete series extension of $\sigma_{\mathrm{sp}}$, and the theorem specifies exactly which Jordan blocks are added and which values of the $\epsilon$-function occur.
- The same filtration works when the base is a Langlands quotient rather than a discrete series (Corollary 4.3), extending the decomposition to a broader class of induced representations.
- Combined with known Jacquet-module formulas for strongly positive discrete series, the decomposition gives a direct route to the Jacquet modules of a large family of discrete series.
- The multiplicity-one property and the exact $3^k$ count give a concrete numerical check for any attempt to decompose these representations algorithmically.
Reading between the lines
- Beyond the paper, filling in the sketched reduction in Proposition 2.6 would extend the same $3^k$ formula to every discrete series from the Mœglin–Tadić classification, including those whose defining segments are linked.
- Beyond the paper, because the proof replaces linked segments by their union and intersection, the general linked case is likely describable by an inclusion–exclusion over segment intersections, with additional terms for each reducible pair.
- Beyond the paper, the method appears transferable to settings with an analogous discrete-series classification and Jacquet-module formula, such as metaplectic groups, though the paper itself treats only symplectic and orthogonal groups.
- Beyond the paper, a small-rank numerical test of Corollary 4.3 for $k=2$ would identify any boundary cases where the irreducibility assumption is doing more work than the proof makes visible.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper studies parabolically induced representations of classical p-adic groups of the form δ(Δ_1)×⋯×δ(Δ_k)⋊σ, where σ is a discrete series obtained by extending a strongly positive discrete series σ_sp by a family of segments satisfying conditions (C1) and (C2). Condition (C1) restricts the exponents and Jordan-block data of each segment; condition (C2) requires pairwise irreducibility of the GL-inducements δ(Δ_i)×δ(Δ_j) and δ(~Δ_i)×δ(Δ_j). The main theorem (Theorem 4.1) asserts a multiplicity-one decomposition in the Grothendieck group into Langlands quotients, indexed by subsets X⊆{1,…,k} and by discrete-series subrepresentations σ′ of the remaining product, together with a filtration whose successive quotients are the displayed Langlands quotients. The proof is an induction combining the Mœglin-Tadić classification of discrete series, Tadić's Jacquet-module formula, and results of Muić on generalized principal series. The introduction states a special case (Theorem 1.1) in which the number of irreducible subrepresentations is 2^k and the total length is 3^k.
Significance. If valid, the result gives an explicit composition series for a substantial class of standard representations appearing in the Mœglin-Tadić classification, and the filtration via intertwining operators is a potentially useful tool for analysing Jacquet modules of discrete series. The proof is a serious induction based on the Mœglin-Tadić classification and Tadić's formula, and the main decomposition formula (4.1) is precise and checkable. The paper would be a useful contribution to the representation theory of p-adic classical groups, provided the scope of the counting claims and a few proof steps are clarified.
major comments (3)
- [Section 3, (C1), and Theorem 1.1 / Corollary 4.4] The family in (C1) explicitly allows b_i=-1/2 (equivalently -b_i=1/2). For such a segment, Proposition 3.4 gives δ(Δ_i)⋊σ = σ2 + Lang(δ(Δ_i)⋊σ), so the length is 2, not 3, and Proposition 3.5 counts only 2^l irreducible subrepresentations with l=#{i: -b_i≠1/2}. Substituting this into the sum in (4.1) gives total length 2^r 3^{k-r}, where r=#{i: -b_i=1/2}, not 3^k. Theorem 1.1 and Corollary 4.4 assert 2^k irreducible subrepresentations and, in Theorem 1.1, length 3^k without excluding b_i=-1/2. If these statements are intended only for segments produced by the Mœglin-Tadić reduction steps of Proposition 2.6, where the lower exponents are non-positive, that restriction must be stated explicitly; if they are intended for all families satisfying (C1)-(C2), they are false as written.
- [Proposition 2.6] The proof contains the step 'It is not hard to check that condition (C1) remained valid' after replacing linked segments by unions/intersections and possibly taking contragredients. No argument is supplied that the new endpoints satisfy the parity, emptiness, and cuspidal-reducibility conditions in (C1) relative to σ_sp, nor that the algorithm terminates with the required (C2) irreducibility. Since Proposition 2.6 is cited in Corollary 4.4 as the link between the Mœglin-Tadić description and the hypotheses of Theorem 4.1, this is a load-bearing gap. Please provide a complete verification or explicitly restrict Corollary 4.4 to families that already satisfy (C1)-(C2).
- [Proposition 3.4, first case] The induction in the first case is described as an induction over card(Y), but the reducibility of δ([ν^{1/2}ρ,ν^cρ])⋊σ concerns the particular cuspidal ρ of the new segment. If no segment in Y has the same ρ, the 'minimal corresponding segment' [ν^{-b_j}ρ,ν^{c_j}ρ] is not defined and the reduction that removes the corresponding Jordan blocks from σ makes no reference to a ρ-segment. The argument should be formulated as an induction over the number of Y-segments with the given ρ (or should explain why segments with different ρ are irrelevant under (C2)). Without this clarification, the proof of the basic one-segment step is incomplete.
minor comments (6)
- [Theorem 1.1 and Corollary 4.4] The phrase 'there are 2k of them' should read 'there are 2^k of them' both in Theorem 1.1 and in Corollary 4.4; the superscript is essential to the counting claim.
- [Proposition 3.4, second case] The sentence defining σ3 and σ4 says 'Jord(σ2)=Jord(σ3)=...'; it should say 'Jord(σ3)=Jord(σ4)=...'.
- [Theorem 4.1 proof, equation (4.7)] In the display after (4.7), 'j∈X\S' should be 'j∈S\X'; the current notation is nonsensical because X is a subset of S.
- [Corollary 4.3, equation (4.9)] The summation condition 'card(X)=k' in (4.9) should be 'card(X)=l', consistent with the filtration index l in the statement.
- [Proposition 3.4, proof] The reference 'By Remark 3.2 and Proposition 4.2 of [5]' is problematic: the paper has only Remark 3.1, no Remark 3.2. The intended reference should be corrected.
- [Theorem 2.2(iii)] The notation 'Jord\{(a,ρ)(a−,ρ)}' should be 'Jord\{(a,ρ),(a−,ρ)}' with a comma separating the two pairs.
Circularity Check
No circularity: the composition series is proved from Mœglin–Tadić and Muić inputs by induction, and the sole self-citation is a peripheral special case.
full rationale
I walked the derivation chain of Theorem 4.1 and its supporting propositions. The main result is not obtained by assuming the target formula or by renaming a fitted parameter as a prediction. Theorem 4.1 is proved by induction whose base case is Proposition 3.4, which is itself derived from external results: Theorem 2.1 and 2.3 of Muić [7], Lemma 6.1 of Muić [8], and results of Mœglin–Tadić [5]. The induction step uses the semisimplifications of kernels of standard intertwinings, formula (4.3), and the induction hypothesis, then identifies the subrepresentation sums using Proposition 3.5, which is proved by a Jacquet-module multiplicity count via Frobenius reciprocity. The multiplicity-one assertion for each displayed Langlands quotient is again proved by a separate Jacquet-module computation, not by assuming the decomposition. No fitted parameter or input quantity is renamed as an output prediction. The only self-citation is Reference [2], cited in the introduction as the previously solved case of induction from two segments with cuspidal σsp; it is not used in the proof of Theorem 4.1 or Corollary 4.4, so it is not load-bearing. I therefore find no circular step. A separate, non-circular correctness concern is worth flagging: Remark 3.1 explicitly adds the possibility −b_i = 1/2 to the family, Proposition 3.4 then gives a one-segment decomposition of length 2 rather than 3 in that case, and Proposition 3.5 counts 2^l irreducible subrepresentations with l = card({i : −b_i ≠ 1/2}); yet Theorem 1.1 and Corollary 4.4 assert 2^k subrepresentations and length 3^k without excluding −b_i = 1/2. That is an internal inconsistency in the advertised counts, but it is not a circularity of the derivation.
Assumptions & free parameters
assumptions (3)
- domain assumption Mœglin-Tadić classification of discrete series of classical p-adic groups
- standard math Tadić's formula for Jacquet modules (Theorem 2.1)
- domain assumption The field F has characteristic different from two and the groups are symplectic or orthogonal
Cite this review
Pith. "Pith review of Composition series of a class of induced representations built on discrete series." pith.science (2026). https://pith.science/paper/LKPERAVL
@misc{pith2026190803818,
author = {Pith},
title = {Pith review of: Composition series of a class of induced representations built on discrete series},
year = {2026},
howpublished = {\url{https://pith.science/paper/LKPERAVL}},
note = {Machine review of arXiv:1908.03818}
}
read the original abstract
We have determined composition series of a class of induced representations appearing in Moeglin Tadi\'c classification of discrete series. The result is further used to determine composition series of certain representations induced from Langlands quotients. This should provide more information on decomposing standard representations as well as Jacquet modules of discrete series, which has application in automorphic forms.
Reference graph
Works this paper leans on
-
[1]
I. N. Bernstein and A. V. Zelevinsky, Induced representations of reductive p-adic groups. I , Ann. Sci. ´Ecole Norm. Sup. (4) 10 (1977), no. 4, 441-472
work page 1977
-
[2]
I. Ciganovi´ c, Composition series of a class of induced representations, a case of one half cuspidal reducibility, Pacific J. Math. 296 (2018), no. 1, 21-30
work page 2018
-
[3]
Mati´ c, First occurrence indices of tempered representations of me taplectic groups, Proc
I. Mati´ c, First occurrence indices of tempered representations of me taplectic groups, Proc. Amer. Math. Soc. 144 (2016), no. 7, 3157-3172
work page 2016
-
[4]
, Jacquet modules of strongly positive representations of th e metaplectic group ~Sp(n), Trans. Amer. Math. Soc. 365 (2013), no. 5, 2755-2778
work page 2013
-
[5]
C. Mœglin, Sur la classification des s´ eries discr` etes des groupes cla ssiques p-adiques: param` etres de Langlands et exhaustivit´ e. (French) [On th e classification of discrete series of classical p-adic groups: Langlands parameters and complet eness] J. Eur. Math. Soc. (JEMS) 4 (2002), no. 2, 143-200
work page 2002
-
[6]
Tadi´ c,Construction of discrete series for classical p-adic groups, J
, M. Tadi´ c,Construction of discrete series for classical p-adic groups, J. Amer. Math. Soc. 15 (2002), no. 3, 715-786
work page 2002
-
[7]
G. Mui´ c, Composition series of generalized principal series; the ca se of strongly positive dis- crete series , Israel J. Math. 140 (2004), 157-202
work page 2004
-
[8]
, Reducibility of generalized principal series , Canad. J. Math. 57 (2005), no. 3, 616-647
work page 2005
Show all 12 references
-
[9]
Tadi´ c,Structure arising from induction and Jacquet modules of rep resentations of classical p-adic groups , J
M. Tadi´ c,Structure arising from induction and Jacquet modules of rep resentations of classical p-adic groups , J. Algebra 177 (1995), no. 1, 1-33
1995
-
[10]
, A family of square integrable representations of classical p-adic groups in the case of general half-integral reducibilities Glas. Mat. Ser. III 37(57) (2002), no. 1, 21-57
2002
-
[11]
China Math
, On tempered and square integrable representations of class ical p-adic groups , Sci. China Math. 56 (2013), no. 11, 2273-2313
2013
-
[12]
A. V. Zelevinsky, Induced representations of reductive p-adic groups. II. On irreducible rep- resentations of GL(n), Ann. Sci. ´Ecole Norm. Sup. (4) 13 (1980), no. 2, 165-210. Email address : igor.ciganovic@math.hr (Igor Ciganovi´ c)Department of Mathematics, University of Zag...
1980
Reviewed August 14, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.