{"id":"a15a115b-b3f0-45fe-a714-3d13649edbaa","arxiv_id":"1908.11041","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"New flagged Littlewood-Richardson tableaux count GL_n to O_n branching multiplicities for arbitrary highest weights, with applications to generalized exponents of types B and D.","lead":"The paper finds a new tableaux-counting formula for how representations of the general linear group GL_n break down when restricted to the orthogonal group O_n. It is the first subtraction-free combinatorial branching rule of this kind for O_n, and it also yields new formulas for Lusztig t-weight multiplicities of types B and D.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 4.17 depends on Lemma 6.8, whose surjectivity proof omits Case 3 ('we leave it to the reader'); the omitted mixed case is load-bearing for the induction, and §6.3 inherits the gap.","rationale":"After reading the paper in good faith, the main formula is plausible and the subtraction-free formulation is a genuine extension. The reader's weakest_assumption points to the external spinor model and Howe-duality identification [21, Theorem 5.3]; I agree those are foundational, but they are cited from prior work and are not the point where this manuscript's argument is most exposed. The most load-bearing internal step is Lemma 6.8, because Theorem 4.4 is the bridge from the crystal model to the flagged LR count, and Theorem 4.17 is nothing more than Theorem 4.4 plus the established equality with the branching multiplicity. The explicit sentence 'We leave it to the reader' in Case 3 is an acknowledged omitted proof, and the surrounding induction cannot be checked from the text for that residue pattern. The n - 2μ'_1 < 0 reduction in Section 6.3 is sketched even more briefly and relies on the same lemma, so it does not provide independent support. This is a proof-completeness concern, not an accusation of error: the numerical example and stable range are consistent, and the gap may be fillable. The honest verdict remains conditional rather than accept, matching the reader's evaluation. I do not think the concern forces rejection, because no contradiction or counterexample has been found; it forces completion of the proof.","tokens_in":41779,"tokens_out":6631,"duration_ms":65362,"concrete_test":"Complete the omitted Case 3 of Lemma 6.8 by direct computation: take a configuration with rU_{2i+1}(a_i) < rU_{2i}(1) and rU_{2i+3}(a_{i+1}) > rU_{2i+2}(1), form T_{i+1}, T_i via (6.14)–(6.15), compute the four ˚-pairs (T_R*_{i+1}, L_T_i, R_T_{i+1}, T_L*_i), and check Definition 2.4(1)(i)–(iii). If the inequalities close, the omitted case is routine; if they do not, Lemma 6.8 and hence Theorem 4.4 fail in that case.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central branching formula (Theorem 4.17) is derived from the bijection in Theorem 4.4. The surjectivity half of Theorem 4.4 is Lemma 6.8, an induction on n in which the step T_{i+1} < T_i is split into four residue cases. Case 3, where rU_{2i+1}(a_i) < rU_{2i}(1) and rU_{2i+3}(a_{i+1}) > rU_{2i+2}(1), is not proved: the text says the proof is almost identical with Case 2 and leaves it to the reader. This is not a purely cosmetic abbreviation, because the displayed ˚-pairs and inequalities in Case 2, especially (6.17)–(6.21), are derived under the opposite first inequality. Without an explicit verification of Definition 2.4(1)(i)–(iii) in the mixed pattern, the induction establishing surjectivity is incomplete. The n - 2μ'_1 < 0 branch of the proof does not repair this: Lemma 6.9 invokes Lemma 6.8 and also asserts the key bound m_i ≤ L and the equality A_tail = T_tail with only a 'by construction' justification. Thus the proof of the main theorem is not complete as written, even though the formula may well be true and is supported by the stable-range recovery and Example 4.18.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a combinatorial branching formula from GL_n to O_n, expressing the multiplicity [V^λ_{GL_n}:V^μ_{O_n}] as a sum over even partitions δ of the number c^λ_{δμ} of Littlewood–Richardson tableaux of shape λ/δ with content μ^π satisfying the flag condition τ_j + n_j ≤ n+1 (Theorem 1.1 / Theorem 4.17). The proof proceeds through a spinor model of type D, a separation algorithm on highest-weight elements, and a bijection (Theorem 4.4) between l-highest weight elements and flagged LR tableaux. The paper also derives a combinatorial formula for Lusztig t-weight multiplicities for types B_n and D_n (Theorem 5.6) and shows that the branching formula reduces to Littlewood's stable restriction rule when ℓ(λ) ≤ n/2 (Corollary 4.13).","tokens_in":41984,"tokens_out":2961,"duration_ms":31061,"significance":"If correct, the main theorem provides a subtraction-free, manifestly positive formula for an orthogonal branching rule outside the stable range, resolving an open analogue of known symplectic results. The formula is not forced by normalization or by assuming the desired count: the flag condition (4.3) is derived from the separation algorithm, and the independent checks in Example 4.18 and the stable-range recovery in Corollary 4.13 give meaningful evidence. The application to generalized exponents is a natural and potentially useful byproduct. However, the paper's central claim rests on the surjectivity half of Theorem 4.4, whose proof is not complete as written.","major_comments":[{"comment":"The surjectivity proof of the main bijection Theorem 4.4 is incomplete: Lemma 6.8 splits the verification of Ti+1 < Ti into four residue cases, but Case 3 is disposed of with the sentence 'The proof of this case is almost identical with Case 2. We leave it to the reader.' This is load-bearing, not cosmetic. In Case 3 the inequalities are mixed (rU_{2i+1}(a_i) < rU_{2i}(1) and rU_{2i+3}(a_{i+1}) > rU_{2i+2}(1)), while the displayed ˚-pairs and inequalities in Case 2, notably (6.17)–(6.21), are derived under the opposite first inequality. An explicit verification of Definition 2.4(1)(i)–(iii) in the mixed pattern is therefore required before Lemma 6.8, and with it the surjectivity of Theorem 4.4 and the branching formula Theorem 4.17, can be regarded as proved.","section":"§6.2, Lemma 6.8"},{"comment":"Lemma 6.9, which handles the case n − 2μ'_1 < 0, asserts the key bound m_i ≤ L and the equality A_tail = T_tail with only a 'by construction' justification. These claims are not immediate from the displayed construction of B and A, and Lemma 6.9 also calls on Lemma 6.8, so it inherits the gap in Case 3. Since this lemma establishes well-definedness of the map in the negative branch, the proof of Theorem 4.4 for n − 2μ'_1 < 0 also needs additional detail.","section":"§6.3, Lemma 6.9"}],"minor_comments":[{"comment":"The name of the t-weight multiplicity is spelled 'Lustig' in the abstract and in Section 5.1; it should be 'Lusztig' to match the rest of the paper and the literature.","section":"Abstract and §5.1"},{"comment":"The arrow in the displayed bijection ψ : LR^{λ'}_{μ'ν'} → LR^λ_{μν^π} appears garbled as '/d47/d47'; this is likely a rendering issue, and the authors should ensure the published version displays a single bijective arrow.","section":"§2.2, display (2.2)"},{"comment":"In the sentence 'Then it is straightforward to check that for ξ,υ ∈ P(2)_8', the line break between 'c^{λ'}_{ξ'μ'} =' and the case values makes the two cases visually awkward; a displayed aligned equation would improve readability.","section":"§4.2, Example 4.18"}],"recommendation":"major_revision","confidential_remarks":"The missing Case 3 in Lemma 6.8 is a real proof gap, but it is localized and the surrounding evidence (the stable-range recovery, Example 4.18, and the detailed treatment of the other cases) suggests the main result is likely correct. I would encourage the editors to invite a revision that supplies the omitted case and clarifies the 'by construction' steps in Lemma 6.9, rather than rejecting the paper."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper supplies the missing orthogonal analogue of the Lecouvey–Lenart formula: a subtraction-free combinatorial rule for [V^λ_{GL_n} : V^μ_{O_n}] in terms of flagged Littlewood–Richardson tableaux. That is a real gap in the literature, and the formula is not a re-coordinatization of known results. The recovery of Littlewood's stable rule and the match with Enright–Willenbring in Example 4.18 are genuinely independent checks. The paper also gives the expected application to Lusztig t-weight multiplicities for B_n and D_n. Credit where due: the main construction is substantial, the separation algorithm is new for type D, and the statement is clean and almost certainly correct.\n\nThe soft spot is exactly where the stress-test points. Lemma 6.8 is the surjectivity half of Theorem 4.4, and Case 3 is left with \"we leave it to the reader.\" That is not just a cosmetic abbreviation. Case 2's displayed identities, especially (6.20) and (6.21), are derived under the inequality rU_{2i+1}(a_i) > rU_{2i}(1), and Case 3 flips that inequality while keeping the other one. So \"almost identical\" overstates the similarity; the ˚-pairs and admissibility conditions would have to be re-derived. Since the induction for n > 4 depends on all four cases, the proof of the main theorem is incomplete as written. The n - 2μ'_1 < 0 branch inherits this: Lemma 6.9 invokes Lemma 6.8 and then asserts the key bounds and tail equality with only \"by construction\" justification. These are not signs the theorem is false; the independent checks and the stable-range argument provide real confidence. But they are genuine gaps in the written proof.\n\nThis paper is for specialists in combinatorial representation theory, crystal bases, and branching rules. It deserves a serious referee, not a desk reject, and the referee should ask for a full proof of Case 3 and more detail in §6.3. If the authors can supply those, the paper is solid. If not, it remains an interesting conjecture with strong evidence. My own view: the formula is very likely correct, and the gaps are patchable, but the current version is conditional.","headline":"A genuinely new and likely correct subtraction-free branching formula for GL_n to O_n, but the proof as written has an omitted load-bearing case in Lemma 6.8 that needs to be supplied.","tokens_in":42604,"tokens_out":2679,"would_cite":true,"duration_ms":28359,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["17B37","22E46","05E10"],"pacs":[],"model":"deepseek-v4-flash","headline":"A new subtraction-free branching rule from GL_n to O_n, expressed by flagged Littlewood–Richardson tableaux, is proved and applied to Lusztig t-weight multiplicities.","keywords":["branching rule","Littlewood-Richardson tableaux","flag condition","spinor model","orthogonal group","generalized exponents","Lusztig t-weight multiplicity","crystal graphs"],"falsifier":"Take a specific pair outside the stable range, for instance n=8, λ=(5,4,4,3,2,2,0,0), μ=(2,2,2,1,1) from Example 4.18, compute the left side of Theorem 4.17 by an independent algebraic method (e.g., the formula in [2, Theorem 4]) and compare with the sum Σ c^λ_{δμ}; the two agree in the example, and a single disagreement for any pair would falsify the identity.","tokens_in":41483,"feed_emoji":"🧮","tokens_out":7948,"duration_ms":67786,"temperature":0.7,"pith_summary":"The paper proves a new combinatorial formula for the branching multiplicity of an irreducible orthogonal group representation in a general linear group representation. The formula expresses the multiplicity as a sum over even partitions of the number of Littlewood–Richardson tableaux satisfying a simple flag condition on their second-rightmost column. It is subtraction-free, valid for arbitrary λ with length at most n, and reduces exactly to Littlewood's restriction formula in the stable range ℓ(λ) ≤ n/2. As a byproduct, the authors obtain a new combinatorial realization of Lusztig's t-weight multiplicity at zero weight for types B_n and D_n.","feed_headline":"Flagged LR tableaux give new GL_n-to-O_n branching rule","feed_subtitle":"A subtraction-free formula for arbitrary λ, recovering Littlewood's restriction rule in the stable range.","key_machinery":"The machinery has three parts. First, the spinor model T(μ,n) is a crystal of admissible sequences of tableaux T_i of shapes λ(a_i,b_i,c_i) that realizes the type-D crystal B(Λ(μ)); it is the setting in which branching multiplicities become counts of l-highest weight elements. Second, the separation algorithm uses sliding operators S_j, each a specific composition of jeu de taquin moves, to move a column tail one position leftward; iterating it decomposes any l-highest weight element as a pair (H_{($δ^{1}$)^π}, U) with U a flagged Littlewood–Richardson tableau. Third, the bijection ψ, an anti-lattice analogue of a known bijection between conjugate-shape LR tableaux and their anti-content counterparts, identifies the conjugate-shape tableaux $LR^{{λ'}}$_{δ' μ'} with the anti-content tableaux LR^λ_{δμπ} and converts the flag condition τ_j + n_j ≤ n+1 into an equivalent condition on the first-row entries of the original tableau.","core_discovery":"The central result (Theorem 1.1, stated as Theorem 4.17) is the identity [V^λ_{GL_n}:V^μ_{O_n}] = Σ_{δ∈P(2)_n} c^λ_{δμ}, where c^λ_{δμ} counts Littlewood–Richardson tableaux U of shape λ/δ with content μ^π satisfying τ_j + n_j ≤ n+1 for 1 ≤ j ≤ μ'_2, with τ_j the entries of the second rightmost column of the companion tableau and n_j defined via a 'missing indices' sequence. The proof identifies the branching multiplicity with the number of l-highest weight elements in the type-D spinor model T(μ,n) (via Howe duality), then constructs a bijection from these elements to the flagged tableaux using a new combinatorial operation called separation. Separation is carried out by sliding operators that move column tails leftward while preserving type-A crystal equivalence, and a second bijection ψ translates the flag condition into the companion-tableau form. The formula vanishes, i.e., reduces to the classical Littlewood sum, exactly when ℓ(λ) ≤ n/2.","pith_inferences":["The flag condition τ_j + n_j ≤ n+1 is likely a tableau version of 'the column fits inside n boxes'; tracing it through the bijection ψ might connect it to the orthogonal tableaux models of Sundaram and King, though the paper leaves such a bijection open (Remark 5.8).","The separation algorithm, defined here only on highest-weight elements, has a natural extension to arbitrary crystal elements; a broader version is mentioned in the paper's Remark 3.22(1), and if carried out it could yield branching formulas for tensor products or for other reductive pairs.","The new formula is manifestly positive, so it gives an independent combinatorial proof of the positivity of these branching multiplicities; comparing it with the known alternating formula of [2] for special μ,ν may suggest new bijections between flagged LR tableaux and the terms in that alternating sum."],"forward_implications":["The formula gives a subtraction-free count of [V^λ_{GL_n}:V^μ_{O_n}] for every λ with ℓ(λ) ≤ n, beyond the stable range where Littlewood's product formula applies.","In the stable range ℓ(λ) ≤ n/2 the flagged set LR^λ_{δμπ} coincides with the full set LR^λ_{δμπ}, so the new sum is literally Littlewood's formula (Corollary 4.13).","For type B_n and D_n, the generalized exponents — equivalently the Lusztig t-weight multiplicities K_{μ0}(t) — are expressed as sums over distinguished tableaux D_n(μ) with weights measured by |φ+ρ|/2 (Theorem 5.6).","The same separation machinery is shown to yield analogous flagged-LR formulas for the GL_n-to-Sp_n and type-B/C branching rules (Remark 4.16)."],"supporting_citations":[{"why":"Supplies Theorem 2.7, the crystal isomorphism T(μ,n) ≅ B(Λ(μ)) for the spinor model of type D, which is the foundation for counting highest-weight elements.","marker":"[19]"},{"why":"Supplies the equality [V^λ_{GL_n}:V^μ_{O_n}] = c^μ_λ(d) (Theorem 5.3) and the stable-range bijection that the new proof extends.","marker":"[21]"},{"why":"Provides the Howe duality / see-saw pair (O_n, D_∞) identifying the branching multiplicity with the number of l-highest weight elements.","marker":"[29]"},{"why":"Supplies the type-C analogue of flagged LR tableaux, the companion-tableau idea, and the method for translating branching formulas into Lusztig t-weight multiplicities.","marker":"[22]"},{"why":"The classical Littlewood restriction formula, which the new theorem generalizes and which is recovered in the stable range.","marker":"[23]"},{"why":"Gives the algebraic (alternating) formula for branching multiplicities used in Example 4.18 to test and compare the new combinatorial count.","marker":"[2]"}],"fun_headline_variants":["Flagged LR tableaux give branching rule from GL_n to O_n","New branching rule for GL_n to O_n via flagged LR tableaux","Flagged LR tableaux generalize Littlewood's restriction formula","Branching GL_n to O_n via flagged tableaux","Lusztig t-weight multiplicity via flagged LR tableaux"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof relies on the equality between the branching multiplicity [V^λ_{GL_n}:V^μ_{O_n}] and the number of l-highest weight elements in the spinor model T(μ,n); if this identity (grounded in Howe duality and the crystal isomorphism of Theorem 2.7) failed, the flagged tableau count would not measure the true multiplicity.","fun_headline_variants_meta":{"raw":{"variants":["Flagged LR tableaux give branching rule from GL_n to O_n","New branching rule for GL_n to O_n via flagged LR tableaux","Flagged LR tableaux generalize Littlewood's restriction formula","Branching GL_n to O_n via flagged tableaux","Lusztig t-weight multiplicity via flagged LR tableaux"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001065,"raw_usage":{"total_tokens":4427,"prompt_tokens":869,"completion_tokens":3558,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":485,"completion_tokens_details":{"reasoning_tokens":3471}},"tokens_in":485,"tokens_out":3558,"duration_ms":24782,"temperature":1.0,"reasoning_tokens":3471,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T10:25:54.312553+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a specific pair outside the stable range, for instance n=8, λ=(5,4,4,3,2,2,0,0), μ=(2,2,2,1,1) from Example 4.18, compute the left side of Theorem 4.17 by an independent algebraic method (e.g., the formula in [2, Theorem 4]) and compare with the sum Σ c^λ_{δμ}; the two agree in the example, and a single disagreement for any pair would falsify the identity.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies Theorem 2.7, the crystal isomorphism T(μ,n) ≅ B(Λ(μ)) for the spinor model of type D, which is the foundation for counting highest-weight elements."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the equality [V^λ_{GL_n}:V^μ_{O_n}] = c^μ_λ(d) (Theorem 5.3) and the stable-range bijection that the new proof extends."},{"cited_title":"Wang, Duality in infinite dimensional Fock representations , Commun","cited_arxiv_id":null,"evidence_quote":"Provides the Howe duality / see-saw pair (O_n, D_∞) identifying the branching multiplicity with the number of l-highest weight elements."},{"cited_title":"Lecouvey, C","cited_arxiv_id":null,"evidence_quote":"Supplies the type-C analogue of flagged LR tableaux, the companion-tableau idea, and the method for translating branching formulas into Lusztig t-weight multiplicities."},{"cited_title":"Littlewood, On invariant theory under restricted groups , Philos","cited_arxiv_id":null,"evidence_quote":"The classical Littlewood restriction formula, which the new theorem generalizes and which is recovered in the stable range."},{"cited_title":"Enright, J","cited_arxiv_id":null,"evidence_quote":"Gives the algebraic (alternating) formula for branching multiplicities used in Example 4.18 to test and compare the new combinatorial count."}],"review_version":1}