REVIEW 2 major objections 4 minor 7 references
On the cohomology of line bundles over certain flag schemes II
T0 review · 2 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash
Pith's one-line read For n = p+f with 0 ≤ f ≤ p−2, this paper proves that H^d(Z, L) is the simple GL_{d+1}-module with partition (p−1+f, p−1, f+1).
desk verdict New f>0 cases of a cohomology theorem, but the proof leans on a citation whose scope the paper itself makes unclear. 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 load-bearing mechanism is a chain of identifications. A spectral sequence of composite functors reduces $H^i(\mu_{m,n})$ on $Z=G/Q$ to $H^{i-d+1}(V_P(\pi_{m,n}))$, the cohomology of a Weyl module for the maximal parabolic $P$ with highest weight $\pi_{m,n}=(m-n-1)\omega_1+n\omega_2$ (Lemma 1.2.1 and Proposition 1.2.2). For $n=p+f$, Doty's exact sequences (1.3.2)--(1.3.3) relate $V_P(\pi_n)$ to the simple $P$-module $L_P(\lambda_f)$, and applying $H^0$ together with Serre duality shows $H^{d-1}(\mu_n)\cong H^0(N_f)\cong H^d(\mu_n)$. A cited multiplicity-one statement makes this module nonzero, and the fact that any nonzero map $V(\lambda_f)\to H^0(\lambda_f)$ has image $L(\lambda_f)$ forces it to be the simple module. For $f=0$, Jantzen's sum formula computes the characters of the layers in a filtration, giving the alternating Schur-function identity.
What would settle it
Check whether [Liu19b], Cor. 4 states, or directly implies, that every weight $\lambda_f$ with $0\le f\le p-2$ has multiplicity one in $H^d(\mu_{p+f})$. If it covers only $\lambda_0$, then Proposition 1.3.1 for $f>0$ lacks its non-vanishing input. Alternatively, for $p=3$, $d=3$, $f=1$, compute the dimension of $H^d(\mu_4)$ by an independent method and compare it with the Weyl dimension of $L(\omega_1+2\omega_3)$.
Extended reading notes
Core claim
On the paper's own terms, it proves Proposition 1.3.1: for $G=\mathrm{SL}_{d+1}$, $d\ge 3$, and $n=p+f$ with $0\le f\le p-2$, both $H^{d-1}(\mu_n)$ and $H^d(\mu_n)$ are isomorphic to the simple $G$-module $L(\lambda_f)$, where $\lambda_f=f\,\omega_1+(p-2-f)\omega_2+(f+1)\omega_3$; in $\mathrm{GL}_{d+1}$ partition notation this is $(p-1+f,p-1,f+1)$. Equivalently, for the incidence variety $Z=G/Q$ and $\mathcal{L}$ the restriction of $\mathcal{O}(-n-d)\boxtimes\mathcal{O}(n)$, the group $H^d(Z,\mathcal{L})$ is irreducible with that highest weight. For $f=0$, the same methods plus Jantzen's sum formula yield Corollary 1.4.4: the sum of monomial symmetric functions $m_\lambda$ over partitions $\lambda$ of $2p-1$ dominated by $(p-1,p-1,1)$ equals $\sum_{i=0}^{p-2}(-1)^i S_{(p-1,p-1-i,1^{i+1})}$.
Load-bearing premise
The proof that $H^d(\mu_{p+f})$ is nonzero for $f>0$ invokes a cited multiplicity-one result from the first author's earlier paper, which the present paper's own Section 1.4 describes as covering only the endpoint cases $f=0$ and its dual; if that citation does not in fact cover all $f$, the irreducibility conclusion is unsupported.
Editorial extensions
If this is right
- For every prime $p$ and every $0\le f\le p-2$, the two adjacent cohomology groups $H^{d-1}(\mu_{p+f})$ and $H^d(\mu_{p+f})$ are the same irreducible module $L(\lambda_f)$; in particular neither is zero.
- The $f=0$ case gives explicit symmetric-function identities: $\sum_{\lambda\le(p-1,p-1,1)}m_\lambda=\sum_{i=0}^{p-2}(-1)^i S_{(p-1,p-1-i,1^{i+1})}$, with an analogous identity for the dominance order below $(p-1,1)$.
- The theorem extends the earlier $f=0$ result to the whole range $p\le n\le 2p-2$, and for $d=2$ the corresponding statement is already covered by earlier work of the first author.
- When $f>0$ and $d>p-f+1$, the module cannot be obtained from the $f=0$ case by translation functors, so the new cases lie genuinely outside the previously known ones.
Reading between the lines
- If the main theorem is correct, the same spectral-sequence-plus-Doty strategy is a plausible template for other integers $n$ near $p$, or for other maximal parabolics, where explicit simple cohomology modules may be describable by similar highest weights.
- The paper records the conjecture that the symmetric-function identity should hold for every integer $n\ge 2$, not only primes; testing it for small composite $n$ is a direct way to stress the boundary of the result.
- For $f>0$, the theorem implies that the cohomology module has some weight spaces of dimension greater than one, by Seitz's classification; a direct character computation for a small case such as $p=3$, $d=3$, $f=1$ would make this concrete.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper studies, over a field of characteristic p, the cohomology of the line bundle L = O(-n-d) ⊠ O(n) restricted to the incidence variety Z ⊂ P^d × (P^d)^*, for n = p+f with 0 ≤ f ≤ p−2. In the notation of SL_{d+1}, with µ_n = nω1 − (n+d)ω_d, the authors prove that H^d(µ_n) ≅ H^{d−1}(µ_n) ≅ L(λ_f), the simple module of highest weight fω1 + (p−2−f)ω2 + (f+1)ω3. The proof combines a composite-functor spectral sequence with Doty's exact sequences for the P-module V_P(π_n), and uses a multiplicity-one input from the first author's earlier paper [Liu19b]. For f = 0, Jantzen's sum formula is used to express the character of L(λ0), yielding as a by-product an identity between sums of monomial symmetric functions and alternating sums of Schur functions (Cor. 1.4.4).
Significance. If the proof is completed, the result is a genuine extension of [Liu19b] from f = 0 to the whole range f ≤ p−2, and the symmetric-function identity in Cor. 1.4.4 is a clean and checkable by-product. The paper is fairly self-contained apart from the cited results, and the Jantzen sum formula computation in Prop. 1.4.2 is explicit and detailed. The main weakness is external-citation dependence at the key non-vanishing step; because the cited corollary is not quoted in a form that the reader can verify, the crux of the proof is not checkable from the text alone.
major comments (2)
- [§1.3, proof of Prop. 1.3.1] The assertion "by [Liu19b], Cor. 4, λ_f has multiplicity 1 in H^d(µ_n)" is load-bearing: it is the only cited reason that H^d(µ_n) is non-zero and has a highest weight vector of weight λ_f. However, the paper's own Section 1.4 describes [Liu19b], Cor. 3 and 4 as proving only the weight-multiplicity structure of L(λ0) and L(λ0'), and no statement about H^d(µ_{p+f}) for general f is quoted there. If Cor. 4 of [Liu19b] does not cover f > 0, then the non-vanishing of H^d(µ_n) is unsupported and the conclusion H^d(µ_n) ≅ L(λ_f) does not follow. Please state the precise content of [Liu19b], Cor. 4 and either verify that it applies to all 0 ≤ f ≤ p−2 or supply a direct proof of the required multiplicity-one statement.
- [§1.3, Eq. (1.3.4)] The vanishing H^i(H^0_P(π_n)) = H^i(π_n) = 0 for all i ≥ 0 is used to obtain H^0(C) = 0 and the isomorphisms H^i(C) ≅ H^{i−1}(N_f), hence ultimately the vanishing of H^i(N_f) for i ≥ 1. As written, no proof or reference is given for this vanishing. The assertion is true: π_n + ρ has zero pairing with α_1∨, so π_n is singular for the dot action and has no dominant conjugate, and [Jan03], II.5.4 gives H^i(π_n) = 0 for all i. Please include this justification, or an explicit reference, at the point of (1.3.4).
minor comments (4)
- [Abstract and §1.3] The abstract and Introduction restrict to 0 ≤ f ≤ p−2, but §1.3 states Prop. 1.3.1 for 0 ≤ f ≤ p−1, with a separate definition of λ_{p−1}. Please state the exact range covered and, if f = p−1 is indeed proved, adjust the abstract and Introduction accordingly.
- [Abstract] In the abstract, the partition is written as λ_0 = (p−1+f, p−1, f+1); this should be λ_f. The notation λ_0 is also used in §1.4 for the f = 0 weight (p−2)ω2 + ω3, so the abstract's use is confusing.
- [Throughout] There are several typos that should be corrected: "rewrited" after (1.1.2), "i t" after (1.1.4), "applyi ng" before (1.3.5), and "proo f" at the end of §1.4.
- [Cor. 1.4.4] Cor. 1.4.4 is stated for each prime p without mentioning the hypothesis d ≥ 2p−2 that is used immediately before it. Please make the dimension assumption explicit, or state the identity as an identity in the ring of symmetric functions that is stable in d.
Circularity Check
The f>0 case of Proposition 1.3.1 hinges on a load-bearing self-citation whose stated scope appears limited to λ_0 and λ'_0.
-
self citation load bearing
[Section 1.3, proof of Proposition 1.3.1, paragraph following (1.3.8)]
"On the other hand, by [Liu19b], Cor. 4, λf has multiplicity 1 in H d(µn), which is therefore non-zero."
The proof needs a nonzero vector of weight λ_f in H^d(µ_n) to identify the module as L(λ_f). That fact is not proved or derived from the displayed exact sequences; it is imported from the first author's companion paper [Liu19b], Cor. 4. The present paper's own Section 1.4 summarizes [Liu19b], Cor. 3 and 4 as proving only the dominant-weight multiplicities of L(λ_0) and L(λ'_0), not of H^d(µ_{p+f}) for general f. If Cor. 4 does not cover f>0, the nonvanishing of H^d(µ_n) is unsupported and the final identification collapses. The load-bearing premise is thus a self-citation whose scope is not verified in this paper, rather than an independent derivation.
full rationale
The core computation is otherwise independent: Doty's exact sequences, the degeneration of spectral sequences, Kempf vanishing, Serre duality on G/Q and P/B, and Jantzen's sum formula are all external results applied with stated hypotheses. Proposition 1.2.2 and Corollary 1.2.3 derive the structure of H^i(µ_{m,n}) without fitting any parameter, and the by-product identity in Corollary 1.4.4 is obtained by equating two independently computed characters (Prop. 1.4.2 and [Liu19b], Cor. 3). The only step that resembles circularity is the use of [Liu19b], Cor. 4 to import the key multiplicity statement for f>0; because the paper's own §1.4 description suggests that corollary covers only λ_0 and λ'_0, this is a load-bearing self-citation rather than a fully verified external fact. Since the surrounding argument still carries substantial independent content and the target theorem is not merely a restatement of the cited result, the score is moderate.
Assumptions & free parameters
assumptions (4)
- standard math Jantzen sum formula (Jan03, II.8.19) computes the sum of the characters of the layers of the Jantzen filtration of a Weyl module.
- standard math Doty's exact sequences (1.3.2) and (1.3.3) describe the submodule structure of the P-module V_P(π_n) for n = p+f.
- domain assumption The vanishing H^i(π_n) = 0 for all i ≥ 0 (equation 1.3.4).
- domain assumption [Liu19b], Cor. 4: λ_f has multiplicity one in H^d(µ_n) for the range n = p+f with 0 ≤ f ≤ p-1.
Cite this review
Pith. "Pith review of On the cohomology of line bundles over certain flag schemes II." pith.science (2026). https://pith.science/paper/4ZEUKYLJ
@misc{pith2026190808432,
author = {Pith},
title = {Pith review of: On the cohomology of line bundles over certain flag schemes II},
year = {2026},
howpublished = {\url{https://pith.science/paper/4ZEUKYLJ}},
note = {Machine review of arXiv:1908.08432}
}
abstract
Over a field $K$ of characteristic $p$, let $Z$ be the incidence variety in $\mathbb{P}^d \times (\mathbb{P}^d)^*$ and let $\mathcal{L}$ be the restriction to $Z$ of the line bundle $\mathcal{O}(-n-d) \boxtimes \mathcal{O}(n)$, where $n = p+f$ with $0 \leq f \leq p-2$. We prove that $H^d(Z,\mathcal{L})$ is the simple $\operatorname{GL}_{d+1}$-module corresponding to the partition $\lambda_0 = (p-1+f,p-1,f+1)$. When $f= 0$, using the first author's description of $H^d(Z,\mathcal{L})$ and Jantzen's sum formula, we obtain as a by-product that the sum of the monomial symmetric functions $m_\lambda$, for all partitions $\lambda$ of $2p-1$ less than $(p-1,p-1,1)$ in the dominance order, is the alternating sum of the Schur functions $S_{p-1,p-1-i,1^{i+1}}$ for $i=0,\dots,p-2$.
Reference graph
Works this paper leans on
-
[1]
Stephen R.\,Doty, The submodule structure of certain Weyl modules for groups of type A_n , J. Algebra 95 (1985), no. 2, 373-383
work page 1985
-
[2]
Darij Grinberg, Petrie symmetric functions, arXiv:2004.11194
arXiv 2004
-
[3]
Jens Carsten Jantzen, Representations of algebraic groups (2nd ed.), Amer. Math. Soc, 2003
work page 2003
-
[4]
Linyuan Liu, Cohomologie des fibr\'es en droites sur _3/B en caract\'eristique positive: deux filtrations et cons\'equences, submitted (see also arXiv:1903.08758)
work page Pith review arXiv 1903
-
[5]
Linyuan Liu, On the cohomology of line bundles over certain flag schemes, submitted [to this journal]
-
[6]
Ian G.\,Macdonald, Symmetric Functions and Hall Polynomials (2nd ed.), Oxford Univ. Press, 1995
work page 1995
-
[7]
Gary Seitz, The maximal subgroups of classical algebraic groups, Mem. Amer. Math. Soc. 67 (1987), no. 365
work page 1987
Reviewed August 14, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.