{"id":"0cda3f4c-100e-445e-9cad-ac841670acf9","arxiv_id":"2411.13085","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The Penrose transformation is constructed on general non-classical flag domains, and for large weights it gives isomorphisms between higher automorphic cohomology and automorphic forms on Hermitian symmetric domains.","lead":"This paper defines Penrose transformations for cohomology of homogeneous line bundles on flag domains of Hermitian type and proves injectivity and isomorphism conditions. As a corollary, higher automorphic cohomology on non-classical flag domains is identified with spaces of automorphic forms for large tensor powers.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 3.8 is false as stated for Sp(4): the compact root e1-e2 pairs to zero with beta=e1+e2, so Theorems 3.7, 4.3, and the injectivity step of Theorem 4.8 do not cover the Sp(4) case.","rationale":"The reader's weakest assumption was the dependence on the companion preprint [22], particularly Proposition 2.2 used in Lemma 3.8. That dependency is real and worth stating, but my stress-test found a more definite internal problem: Lemma 3.8 is false as written for Sp(4), and Theorems 3.7, 4.3, and the injectivity part of Theorem 4.8 rely on it without excluding Sp(4). This is not a disagreement with external consensus; it is a concrete failure of a stated lemma in a case the paper itself discusses later. The example section even records the vanishing pairing (e1+e2, e1-e2) = 0, so the contradiction with Lemma 3.8 is immediate. I do not claim the final isomorphism theorem is false, only that the proof as written does not cover Sp(4) and therefore the theorem overclaims. The companion-paper dependency remains a secondary concern. Since the central claim may be salvageable by excluding Sp(4) or by a separate argument, the appropriate verdict is CONDITIONAL rather than REJECT, matching the reader's verdict but for a different, more concrete reason.","tokens_in":35542,"tokens_out":17778,"duration_ms":163776,"concrete_test":"In the Sp(4) root system of Example 6.4, compute (e1+e2, e1-e2) explicitly. It equals 0, so Lemma 3.8 fails for beta = e1+e2. Then verify that Theorem 4.8's condition (41) for this beta reduces to (beta, alpha) > 0 for some alpha in Delta^c_+, which is impossible because the only compact root is e1-e2 and the pairing is 0. If no separate Sp(4) argument is provided, the statement of Theorem 4.8 must be restricted to exclude Sp(4) or amended.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The paper's main isomorphism Theorem 4.8 uses Theorem 4.3 to obtain injectivity of the Penrose transformation for the canonical-bundle weights. Theorem 4.3 is Theorem 3.7, whose proof is exactly Lemma 3.8: for every beta in Delta^{nc,1}_+, there exists alpha in Delta^c_+ with (beta, alpha) > 0. Lemma 3.8 is stated for all simple GR of Hermitian type, with no exclusion. However, Example 6.4 (Sp(4)) itself gives Delta^c_+ = {e1-e2}, Delta^{nc,1}_+ = {2e1, e1+e2}, and observes (e1+e2, e1-e2) = 0. Hence for beta = e1+e2 there is no alpha in Delta^c_+ with (beta, alpha) > 0, so Lemma 3.8 is simply false. Consequently Theorem 3.7 and Theorem 4.3 are unproved for Sp(4), and Theorem 4.8's Step 1, which reduces condition (41) to exactly this pairing, does not establish injectivity for Sp(4). Remark 3.9(2) and Lemma 3.10 show the authors are aware that Sp(4) is exceptional, but Theorem 4.8 is not restricted or given an alternative Sp(4) argument. This is an internal inconsistency in the proof, not merely a gap in an imported statement; it affects the central claim as stated.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a Penrose transformation for non-classical flag domains D = G_R/T with G_R of Hermitian type, comparing cohomology of homogeneous line bundles on D with those on a diffeomorphic classical flag domain D'. The main results give sufficient conditions for injectivity of the transformation (Theorems 3.5, 4.2), an isomorphism criterion on compact quotients (Theorem 4.7), an application to automorphic forms on Hermitian symmetric domains (Theorem 4.8), and cup-product maps toward TDLDS (Theorem 5.2). The proofs use a correspondence space W, an incidence variety I, root-system computations, Bott-Borel-Weil vanishings, and results from the authors' companion paper [22].","tokens_in":35825,"tokens_out":7249,"duration_ms":73273,"significance":"If the main theorems are correct, the paper provides a substantial generalization of earlier SU(2,1) and Sp(4) Penrose transformations, linking higher automorphic cohomology of non-classical flag domains to classical automorphic forms. The explicit root-system conditions in Theorems 3.5 and 4.7 are useful and falsifiable, and the applications to canonical-bundle weights (Theorem 4.8) and cup products (Theorem 5.2) are interesting. The paper is not machine-checked and depends on several external or companion results, but the overall structure is coherent and the examples are concrete.","major_comments":[{"comment":"Lemma 3.8 is stated for every simple real group of Hermitian type, but Example 6.4 (Sp(4)) contradicts it: for Δ^c_+ = {e_1-e_2} and Δ^{nc,1}_+ = {2e_1, e_1+e_2}, taking β = e_1+e_2 gives (β, e_1-e_2) = 0, so no α ∈ Δ^c_+ has (β, α) > 0. Consequently Theorems 3.7 and 4.3, which reduce to Lemma 3.8, are not proved for Sp(4), and Theorem 4.8's Step 1, which uses Theorem 4.3 for injectivity, does not cover Sp(4). Since the main theorems are stated without excluding Sp(4), the central claim is overbroad; the paper should either explicitly exclude Sp(4), or supply a separate argument for that case.","section":"Section 3, Lemma 3.8 and Example 6.4"},{"comment":"The injectivity results rely essentially on unstated companion results: Theorem 1.1 and Theorem 1.2 of [22] provide the classical twin D' and its complex structure, while Proposition 2.2 of [22] is used in Lemma 3.8 for roots β with β+γ not a root. These results are neither restated nor proved in the present paper. This is acknowledged in the introduction, but as a referee I cannot verify the foundation of Theorem 3.5, Theorem 4.2, and hence Theorem 4.8 without access to the companion results or at least a precise statement of the needed propositions as clearly marked assumptions.","section":"Section 3 and Section 4, dependence on [22]"},{"comment":"In the proof of Theorem 3.5, after equation (42), the assertion that [F_σ ω^{nc,1}] lies in H^q_DR(W, π_I^* Ω^•_{π_D}(L_μ)) is not immediate: Ω^q_π has a filtration, and membership in the π_I^*Ω^•_{π_D} part is a property of the class, not just of the form's wedge type. Proposition 2.6 gives surjectivity onto that subspace, but one must first justify that the class is in that subspace. Please spell out the filtration step and how the class [F_σ ω^{nc,1}] is identified with the pullback of a class on I.","section":"Section 3, proof of Theorem 3.5, around Proposition 2.6"}],"minor_comments":[{"comment":"The phrase 'or equivalently (λ+ρ_c, α) ≤ 0' is not an equivalence in general; (λ, α) < 0 does not imply (λ+ρ_c, α) ≤ 0 unless (ρ_c, α) is controlled. Since the proof uses the first condition, please replace 'equivalently' by 'in particular' or prove the intended equivalence under the standing assumptions.","section":"Section 2, Theorem 2.7"},{"comment":"There are several typos and OCR artifacts: 'desecnd' in Section 4, 'dose' in Remark 2.9, 'T able' in the proof of Lemma 3.8, and 'K¨ahler' in the Introduction. The text also uses 'ba a' in the statement of Theorem 4.6. These should be corrected.","section":"Throughout"},{"comment":"Remark 3.9(2) and Example 6.4 show that Sp(4) is exceptional, but the theorem statements do not mention this exception. Please add an explicit caveat in the statements of Theorems 3.7, 4.3, and 4.8, or, preferably, broaden the proof to cover Sp(4) as described in the major comments.","section":"Sections 3-4, statements of Theorems 3.7, 4.3, 4.8"},{"comment":"The proof uses 'Lemma 4 of [25]' and 'Table 1 in [17]' without stating the exact root-system conventions. Since the k=2 case is delicate and now known to be exceptional, it would help the reader to state the relevant classification fact as a lemma in the paper.","section":"Section 3, Lemma 3.8 proof"}],"recommendation":"major_revision","confidential_remarks":"The paper's reliance on the companion preprint [22] is a risk for standalone verification; if the companion is not yet accepted, the editor may wish to ask for a statement of the exact results used. The Sp(4) counterexample to Lemma 3.8 is the most serious mathematical issue and should be resolved before publication; the theorem statements need to be amended or the proof strengthened. The paper fits the journal's scope in complex geometry and representation theory."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this paper deserves a serious referee, but the main theorems as stated are false for Sp(4), and that's a factual issue you should know before reading further.\n\nThe genuinely new part is the construction of Penrose transformations for all Hermitian-type non-classical flag domains, not just SU(2,1) and Sp(4). The incidence-variety vanishing theorem (Thm 2.7), the injectivity criterion (Thm 3.5), and the isomorphism theorem under Williams' Property W (Thm 4.7) form a coherent, useful framework. The application to automorphic forms on Hermitian symmetric domains (Thm 4.8) and the cup-product construction (Thm 5.2) are natural and potentially significant. If the technical core holds up, this fills a real gap noted in Green-Griffiths-Kerr.\n\nNow the soft spots, in proportion. The stress-test note is correct: Lemma 3.8 is false as stated for Sp(4). With Delta^c_+ = {e1-e2}, Delta^{nc,1}_+ = {2e1, e1+e2}, the root beta = e1+e2 pairs to zero with the only compact root. The lemma's proof even acknowledges Sp(4) as exceptional (Remark 3.9), and Example 6.4 shows the authors know their method fails there, but Theorems 3.7, 4.3, and 4.8 are not restricted. For the canonical-bundle weights in Theorem 4.8, condition (41) also fails for Sp(4). So the injectivity and isomorphism claims overclaim. This is not a small gap; it's an internal counterexample to a stated lemma, and it affects a case the paper explicitly claims to cover.\n\nA second issue, already flagged in the Pith Report: the paper leans on the companion preprint [22] for Theorem 1.1 and Proposition 2.2, including a root-system statement used to finish Lemma 3.8. Those results are not proved or restated. In a series that's acceptable, but it makes the paper's conclusions conditional on an unverified source.\n\nA milder concern: the end of Theorem 2.7 is terse; vanishing on cycles covering an open subset does not immediately imply global vanishing, and the identity-theorem/connectedness step is invoked without full detail.\n\nOverall, the architecture is sound for groups other than Sp(4), and the Sp(4) problem is fixable by excluding it or importing the known separate arguments. This is exactly the kind of paper peer review should catch in revision. Send it out, and tell the referee to check the root-system lemma carefully.","headline":"Smart generalization with a real overclaim: Lemma 3.8 fails for Sp(4), so the main isomorphism theorems don't cover a case they claim to.","tokens_in":36400,"tokens_out":4910,"would_cite":false,"duration_ms":45055,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["32M10","14M17","22E46","32L10"],"pacs":[],"model":"deepseek-v4-flash","headline":"Higher automorphic cohomology of certain line bundles on non-classical flag domains is isomorphic to automorphic forms on Hermitian symmetric domains.","keywords":["flag domains","Penrose transformation","automorphic cohomology","homogeneous line bundles","Hermitian symmetric domains","cycle spaces","discrete series","root systems"],"falsifier":"For a concrete group such as $SU(3,2)$ or $Sp(6,\\mathbb{R})$, compute the dimensions of $H^0(\\Gamma\\backslash B, \\omega_B^{\\otimes k/k_0})$ and $H^q(X, L_{\\mu_k})$ at several large $k$; if they ever differ at a level where the paper's conditions hold, the isomorphism theorem is false. Alternatively, exhibit a Hermitian-type root system in which a root $\\beta \\in \\Delta^{nc,1}_+$ satisfies $(\\beta, \\alpha) \\le 0$ for all compact roots $\\alpha$, which would refute Lemma 3.8 and the injectivity theorem.","tokens_in":35303,"feed_emoji":"📐","tokens_out":18189,"duration_ms":141046,"temperature":0.7,"pith_summary":"The paper proves that the Penrose transformation—a classical bridge between sheaf cohomology and solutions of field equations—applies to a wide class of non-classical flag domains $D = G_{\\mathbb{R}}/T$ with $G_{\\mathbb{R}}$ of Hermitian type, relating the cohomology of homogeneous line bundles on $D$ to that on a diffeomorphic classical flag domain $D'$. Under explicit root-system inequalities, the transformation is injective; under an additional regularity condition, it is an isomorphism on compact quotients $X = \\Gamma \\backslash D$. The consequence the authors emphasize is that for line bundles built from the canonical bundle of the Hermitian symmetric domain $B = G_{\\mathbb{R}}/K$, the higher automorphic cohomology groups $H^q(X, L_{\\mu_k})$ are isomorphic, for all large $k$, to the group $H^0(\\Gamma \\backslash B, \\omega_B^{\\otimes k/k_0})$ of automorphic forms on $B$. This lends an arithmetic structure to cohomology groups that previously had none. The same isomorphism is then used to construct cup-product homomorphisms into the dual of a totally degenerate limit of discrete series.","feed_headline":"Penrose transform maps flag-domain cohomology to automorphic forms","feed_subtitle":"This reduction gives the cohomology groups an arithmetic structure from classical automorphic forms.","key_machinery":"The load-bearing objects are the correspondence space $W = G_{\\mathbb{C}}/T_{\\mathbb{C}}$ (an open Stein subset of the enhanced flag variety) with its two holomorphic submersions to $D$ and $D'$, and the incidence variety $I = \\{(x,u) \\in D \\times U : x \\in Z_u\\}$, where $Z_u$ are the compact cycles in $D$ and $U$ is the cycle space. The EGW theorem identifies the cohomology of a line bundle $L_\\lambda$ on $D$ with the relative de Rham cohomology $H^*_{DR}(W, \\Omega^\\bullet_\\pi(L_\\lambda))$ of global sections on $W$, because $W$ is Stein and the fibers are contractible. The transformation itself is multiplication by the left-invariant $q$-form $\\omega^{nc,1} = \\omega_{-\\beta_1} \\wedge \\cdots \\wedge \\omega_{-\\beta_q}$, which is $d_\\pi$-closed with values in $L_{-2\\rho^{nc,1}}$ and pairs the two de Rham complexes. Two vanishing inputs make the argument work: the vanishing theorem on $I$ (Theorem 2.7), which uses the Bott–Borel–Weil theorem on each cycle and upper semi-continuity of cohomology to control neighboring cycles, and the root-system lemma (Lemma 3.8)—every $\\beta \\in \\Delta^{nc,1}_+$ has positive inner product with some compact root—which turns the injectivity hypothesis into a checkable condition on weights and is proved by classification of Hermitian symmetric spaces together with a root-system vanishing statement from the companion preprint.","core_discovery":"Let $D = G_{\\mathbb{R}}/T$ be a non-classical flag domain with $G_{\\mathbb{R}}$ of Hermitian type. A companion construction gives a classical flag domain $D'$ on the same underlying manifold, with complex structures differing only along a subspace $p^1_{\\pm}$; in root terms $\\Delta_+ = \\Delta^c_+ \\cup \\Delta^{nc,1}_+ \\cup \\Delta^{nc,2}_+$ and $\\Delta'_+ = \\Delta^c_+ \\cup (-\\Delta^{nc,1}_+) \\cup \\Delta^{nc,2}_+$. The paper defines a Penrose transformation $P : H^0(D', L_{\\mu'}) \\to H^q(D, L_\\mu)$ for weights with $\\mu + \\rho = \\mu' + \\rho'$ by passing through the correspondence space $W$ and multiplying by the left-invariant closed $q$-form $\\omega^{nc,1} = \\omega_{-\\beta_1} \\wedge \\cdots \\wedge \\omega_{-\\beta_q}$, where $q = \\#\\Delta^{nc,1}_+$. The main injectivity theorem (Theorem 3.5) asserts that $P$ is injective whenever, for every $\\beta \\in \\Delta^{nc,1}_+$, there is a compact root $\\alpha \\in \\Delta^c_+$ with $(\\alpha, \\mu' - \\beta) < 0$. The proof uses a new vanishing theorem on the incidence variety $I$: if $(\\lambda, \\alpha) < 0$ for some compact root $\\alpha$, then $H^0(I, \\pi_D^* L_\\lambda) = 0$, obtained by the Bott–Borel–Weil vanishing theorem on the cycles $Z_u$ and by upper semi-continuity of cohomology dimensions as the cycle moves. Passing to compact quotients $X = \\Gamma \\backslash D$, the same argument gives injectivity (Theorem 4.2), and Theorem 4.7 upgrades it to an isomorphism when $\\mu + \\rho$ is regular in the Weyl chamber of $\\Delta^c_+ \\cup \\Delta^{nc,1}_+ \\cup (-\\Delta^{nc,2}_+)$ and the pairing inequality $(\\mu' + 2\\rho'_{nc}, -\\Delta'^{nc}_+) > 0$ holds. Theorem 4.8 applies this to the line bundles $L_{\\mu'_k} = \\omega_B^{\\otimes k/k_0}$ pulled back to $D'$, with $L_{\\mu_k} \\to D$ defined by $\\mu_k = \\mu'_k - 2\\rho^{nc,1}$: for all sufficiently large $k$ the Penrose transformation is an isomorphism, so $H^q(X, L_{\\mu_k}) \\cong H^0(\\Gamma \\backslash B, \\omega_B^{\\otimes k/k_0})$.","pith_inferences":["A natural extension, not pursued in the paper, is to make the threshold $N$ in Theorem 4.8 effective by tracking the inner-product inequalities; this would turn the abstract isomorphism into a computable family of isomorphisms.","If Proposition 2.2 of the companion preprint could be replaced by a direct proof from the classification, the injectivity theorem would become self-contained and the dependency on the companion preprint would be removed.","The cup-product homomorphisms target a TDLDS representation; the surjectivity question raised in Problem 1 could be tested by computing the dimensions of the source and target in explicit examples such as $SU(3,2)$.","The isomorphism suggests that the arithmetic structure on $H^q(X, L_{\\mu_k})$ could support automorphic or motivic interpretations, connecting non-classical period domains to the arithmetic theory of automorphic forms."],"forward_implications":["If $G_{\\mathbb{R}}$ is of Hermitian type and $D$ is non-classical, then for every sufficiently large $k$ the Penrose transformation is an isomorphism, so $H^q(X, L_{\\mu_k}) \\cong H^0(\\Gamma\\backslash B, \\omega_B^{\\otimes k/k_0})$.","The injectivity criterion (7) is a purely root-system condition; for weights satisfying $\\mu'$ orthogonal to all compact roots, injectivity holds unconditionally (Theorem 3.7).","On compact quotients, the isomorphism holds exactly when the weight chamber condition (10) and the pairing inequality (11) hold, which together are equivalent to Williams' Property W for the weight $\\mu + \\rho$.","The cup-product theorem gives explicit homomorphisms $H^0(\\Gamma\\backslash B, \\omega_B^{\\otimes k/k_0}) \\times H^0(\\Gamma\\backslash \\overline{B}, \\omega_{\\overline{B}}^{\\otimes k/k_0} \\otimes L_{\\lambda_0}) \\to H^d(X, L_{-\\rho})^*$ for $SU(s+1,s)$ and a variant with an extra weight factor $L_{\\beta}$ for $Sp(6,\\mathbb{C})$.","The quotient $X$ admits no $\\partial\\bar{\\partial}$-structure and is not in the Fujiki class, while $X'$ is projective; the isomorphism transfers arithmetic structure from the projective side to the non-algebraic side."],"supporting_citations":[{"why":"Companion preprint that proves the new complex structure $D'$ (the classical twin) and supplies the root-system vanishing statement used in Lemma 3.8.","marker":"[22]"},{"why":"Monograph that develops the correspondence-space framework, the EGW cohomology identification, and the earlier $SU(2,1)$ and $Sp(4)$ cases that this paper generalizes.","marker":"[12]"},{"why":"EGW theorem identifying sheaf cohomology on the flag domain with relative de Rham cohomology on the Stein correspondence space; this is the map that defines the Penrose transformation.","marker":"[9]"},{"why":"Theorem giving the dimension of automorphic cohomology as a multiplicity when Property W holds; this dimension equality turns injectivity into isomorphism in Theorem 4.7.","marker":"[27]"},{"why":"Lemma characterizing the weights satisfying Property W in terms of inner products; Theorem 4.7 uses it to verify the dimension equality.","marker":"[26]"},{"why":"Upper semi-continuity of cohomology dimensions in an analytic family, used in the proof of the vanishing theorem on the incidence variety.","marker":"[21]"},{"why":"Classification of Hermitian symmetric spaces, invoked in Lemma 3.8 to control the Dynkin diagrams of the simple factors.","marker":"[25]"},{"why":"Table of root-string lengths used in Lemma 3.8 to rule out the $k=2$ case for the $\\beta$-string.","marker":"[17]"}],"fun_headline_variants":["Penrose transform maps flag-domain cohomology to automorphic forms","Flag-domain cohomology reduced to automorphic forms","Penrose transform links flag domains to automorphic forms","From flag domains to automorphic forms via Penrose transform"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire construction depends on the companion preprint's two results—that every non-classical flag domain with Hermitian-type group has a diffeomorphic classical twin with the prescribed complex structure, and that a certain root-counting vanishing statement holds; if either fails, the injectivity and isomorphism theorems collapse.","fun_headline_variants_meta":{"raw":{"variants":["Penrose transform maps flag-domain cohomology to automorphic forms","Flag-domain cohomology reduced to automorphic forms","Penrose transform links flag domains to automorphic forms","From flag domains to automorphic forms via Penrose transform"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001248,"raw_usage":{"total_tokens":5282,"prompt_tokens":1270,"completion_tokens":4012,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":886,"completion_tokens_details":{"reasoning_tokens":3945}},"tokens_in":886,"tokens_out":4012,"duration_ms":36631,"temperature":1.0,"reasoning_tokens":3945,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T16:51:02.083750+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For a concrete group such as $SU(3,2)$ or $Sp(6,\\mathbb{R})$, compute the dimensions of $H^0(\\Gamma\\backslash B, \\omega_B^{\\otimes k/k_0})$ and $H^q(X, L_{\\mu_k})$ at several large $k$; if they ever differ at a level where the paper's conditions hold, the isomorphism theorem is false. Alternatively, exhibit a Hermitian-type root system in which a root $\\beta \\in \\Delta^{nc,1}_+$ satisfies $(\\beta, \\alpha) \\le 0$ for all compact roots $\\alpha$, which would refute Lemma 3.8 and the injectivity theorem.","supporting_citations":[{"cited_title":"Vanishing Theorems and Complex Structures on Non-Classical Flag Domains","cited_arxiv_id":"2405.16536","evidence_quote":"Companion preprint that proves the new complex structure $D'$ (the classical twin) and supplies the root-system vanishing statement used in Lemma 3.8."},{"cited_title":"Green, P","cited_arxiv_id":null,"evidence_quote":"Monograph that develops the correspondence-space framework, the EGW cohomology identification, and the earlier $SU(2,1)$ and $Sp(4)$ cases that this paper generalizes."},{"cited_title":"Eastwood, S","cited_arxiv_id":null,"evidence_quote":"EGW theorem identifying sheaf cohomology on the flag domain with relative de Rham cohomology on the Stein correspondence space; this is the map that defines the Penrose transformation."},{"cited_title":"Williams, The n-cohomology of limits of discrete series, J","cited_arxiv_id":null,"evidence_quote":"Theorem giving the dimension of automorphic cohomology as a multiplicity when Property W holds; this dimension equality turns injectivity into isomorphism in Theorem 4.7."},{"cited_title":"Williams, Discrete series multiplicities in L2(Γ \\G) (II)","cited_arxiv_id":null,"evidence_quote":"Lemma characterizing the weights satisfying Property W in terms of inner products; Theorem 4.7 uses it to verify the dimension equality."},{"cited_title":"Kodaira and D","cited_arxiv_id":null,"evidence_quote":"Upper semi-continuity of cohomology dimensions in an analytic family, used in the proof of the vanishing theorem on the incidence variety."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Classification of Hermitian symmetric spaces, invoked in Lemma 3.8 to control the Dynkin diagrams of the simple factors."},{"cited_title":"Humphreys, Introduction to Lie Algebras and Representation Theory , Springer-Verlag New York, (1972)","cited_arxiv_id":null,"evidence_quote":"Table of root-string lengths used in Lemma 3.8 to rule out the $k=2$ case for the $\\beta$-string."}],"review_version":1}