{"id":"48c917a3-b9f3-44e6-8620-e038395de502","arxiv_id":"2607.15428","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A new cohomology class obstructs the existence of Dirac complements, and for Lie algebras carrying a definite invariant pairing, the diagonal in g⊕ḡ admits a Dirac complement exactly when g is abelian — covering all real compact semisimple Lie algebras with the Killing form.","lead":"This mathematics paper studies Dirac structures — integrable half-rank subbundles of Courant algebroids — and asks when each admits a complementary partner of the same kind, a question tied to Lie bialgebroids, Poisson groupoids, and deformation theory. The authors build a new cohomological obstruction to such complements and prove that for the diagonal Dirac structure attached to a Lie algebra with a definite invariant pairing, a complement exists only when the Lie algebra i","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Obstruction class's H³ label conflicts with Lemma 4.4's shifted grading; Lemma 4.2's cancellation is not reproducible, though Appendix A supplies an alternative proof.","rationale":"The reader's weakest assumption—the canonicity computation in Lemma 4.2—is real: the cancellation is not reproducible from the text, and the proof of Proposition 4.6 depends on it. However, the paper contains independent support (Appendix A, Prop A.7(4)) that establishes the required invariance for general curved L∞ algebras by a short argument; so the mathematical construction is very likely sound. The additional grading inconsistency is concrete and affects the statement of the central claim as written: under Lemma 4.4's shifted grading N_M is a degree-2 class, while Definition 4.8 and the examples use H³/natural grading. This does not invalidate the obstruction argument but requires a correction/relabelling. Therefore I keep the reader's CONDITIONAL verdict rather than rejecting or fully accepting.","tokens_in":27660,"tokens_out":30240,"duration_ms":291897,"concrete_test":"Compute the twisted curved DGLA of Proposition 3.3 for M'=graph(ω) and apply Appendix A, Prop A.7(4), to prove p(d_{M'})=p(d_M) on a_ab(L) without invoking Lemma 4.2's pointwise identity. Then regrade a_ab(L) exactly as in Lemma 4.4: for L=T*X, list the degree of H in Γ(∧³T*X) and the cohomological degree of [H]. If the Appendix proof closes and [H] lands in degree 2, replace every 'H³(a_ab(L))' by 'H²(a_ab(L))' (or explicitly declare the unshifted grading); if [H] lands in degree 3, the shift convention in Lemma 4.4 needs correction.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central object N_L is only well-defined if (i) d descends to a_ab(L) independently of the complement and (ii) the degree of N_M is unambiguous. On (i): Lemma 4.2's proof cannot be closed as written—the step '= 2⟨[ω,l],m1∧m2⟩ + ⟨ω,[l,m1∧m2]⟩ − ρ(l)⟨ω,m1∧m2⟩ = ⟨[ω,l],m1∧m2⟩' uses an unstated sign convention for the extended Axiom A2; a direct derivation with the natural graded Leibniz rule gives a different coefficient. This is exactly the step on which Proposition 4.6's canonicity rests. Mitigation: Appendix A, Prop A.7(4), proves the same quotient-level invariance for twisted curved L∞ algebras without this cancellation; the authors should explicitly route the main proof through it. On (ii): Lemma 4.4 sets a_ab(L)=Γ(∧•L)[1]/[·,·], so N_M∈Γ(∧³L) has degree 2, not 3; yet Definition 4.8 and Proposition 4.7 claim H³(a_ab(L)), while Example 5.10 uses the natural (unshifted) grading and H³_dR. The grading is switched silently, making the central theorem's statement ambiguous.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the Dirac complement problem for Courant algebroids of split signature. It shows that a lagrangian complement always exists (Prop. 2.19) and that, in an exact Courant algebroid, Dirac complements always exist locally (Prop. 2.24). The main new tool is an obstruction class: for a Dirac structure L, the author chooses a lagrangian complement M and considers the curved DGLA governing deformations of M. Quotienting Γ(∧• L)[1] by its commutator subalgebra gives a complex a_ab(L); the curvature N_M descends to a canonical class N_L, and if L admits a Dirac complement then N_L = 0 (Theorem 4.9). This class is computed in several examples, including twisted exact Courant algebroids and Courant algebroids over a point. In the final section, the authors prove that for a real quadratic Lie algebra with definite invariant pairing, the diagonal Δ ⊆ g⊕ḡ admits a Dirac complement if and only if g is abelian (Theorem 6.9), yielding Corollary 6.10 for compact semisimple Lie algebras. Appendix A extends the construction to curved L∞ algebras.","tokens_in":27822,"tokens_out":11310,"duration_ms":114360,"significance":"If the technical issues below are resolved, this is a valuable contribution. The obstruction class gives a new, computable invariant for the complement problem, and the definite-pairing theorem is a clean and surprising contrast with the complex semisimple case. The paper also contains useful structural results (Props. 2.19 and 2.24) and a nice generalization to curved L∞ algebras in Appendix A. However, the central definition of N_L currently has an unresolved grading ambiguity and a proof gap in the canonicity of the differential. These issues are load-bearing, so the paper needs revision before I can recommend acceptance.","major_comments":[{"comment":"There is a grading inconsistency in the definition of the obstruction class. Lemma 4.4 defines a_ab(L) = Γ(∧•L)[1] / [a(L),a(L)], so an element of Γ(∧^3 L) has degree 2, not degree 3. Yet Proposition 4.7 and Definition 4.8 place N_L in H^3(a_ab(L)), and Example 5.10 silently switches to the unshifted grading and to H^3_dR(X). This is not a notational nicety: the degree of the differential, the meaning of the quotient, and the statement of Theorem 4.9 all depend on the convention. Please fix the convention and state the shift explicitly; for Example 5.10, if the shifted grading is retained, the statement should be that N_L lives in H^2(a_ab(L)) ≅ H^3_dR(X).","section":"§4.1, Lemma 4.2"},{"comment":"The proof that d_M' = d_M + [ω,·] contains a cancellation that I could not reproduce. In the displayed computation near the end, the step \n  2⟨[ω,l],m1∧m2⟩ + ⟨ω,[l,m1∧m2]⟩ − ρ(l)⟨ω,m1∧m2⟩ = ⟨[ω,l],m1∧m2⟩\nrequires a specific extension of Axiom (A2) to multi-vectors, but the sign convention for that extension is not stated. With the standard graded Leibniz rule I obtain an extra term and not the claimed identity. Since Proposition 4.6 (and therefore the well-definedness of N_L and Theorem 4.9) depends on this lemma, this is a load-bearing gap. Appendix A, Proposition A.7(4), proves twist-invariance of the quotient by a short argument that avoids this cancellation; the authors should either supply the missing convention and complete the computation, or make the Appendix A argument the primary route for the DGLA case.","section":"§4.1, Lemma 4.2"},{"comment":"The proof of closedness of N_M uses 'Definition 3.1' to assert d_M N_M = 0. This is correct if one interprets N_M as the curvature R of the curved DGLA, but the reader has to track the sign between the Nijenhuis tensor and the curvature R. The paper does not state this sign explicitly, and later in Appendix A the authors themselves note that one must set ℓ₀ = −R (Remark A.3). Please add a sentence in §4.2 clarifying that the curvature element is ±N_M with the convention chosen in §3, so that the closedness statement follows from the curved-DGLA axioms.","section":"§4.2, Proposition 4.7"}],"minor_comments":[{"comment":"The phrase 'taking the orthonormal basis A_i = (e_i, e_i)' is imprecise: the A_i are isotropic in the ambient pairing on g⊕ḡ. What is meant is that the e_i form an orthonormal basis of g. Please rephrase.","section":"Example 5.5"},{"comment":"There is a typo: 'for A 2, A2 ∈ g' should presumably be 'for A1, A2 ∈ g'.","section":"Example 2.3"},{"comment":"If the shifted grading is retained, the sentence 'the cohomology H•(a_ab(L)) recovers H*_dR(X)' should be replaced by the explicit shift H^k(a_ab(L)) ≅ H^{k+1}_dR(X). This will prevent future readers from miscomputing the degree of N_L.","section":"Example 5.10"},{"comment":"The statement about averaging is somewhat vague. It would help to give a one-line example or to make explicit why the Maurer-Cartan equation is not compatible with averaging, since this is the point of the remark.","section":"Remark 6.2"}],"recommendation":"major_revision","confidential_remarks":"The paper is promising and the main theorems are likely correct, but the central invariant is not yet presented in a way I can certify. The grading ambiguity and the gap in Lemma 4.2 affect every later use of N_L. I recommend requiring a revised §4 with explicit grading conventions and a completed proof of Lemma 4.2, or an explicit reduction to Appendix A. No concerns about novelty or attribution beyond the usual. The use of [Bur+] in Remark 5.8 is appropriate and not an issue."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a serious paper. The core is new: an abelianized cohomology class N_L for a Dirac structure, built from the Nijenhuis tensor of any lagrangian complement, with Theorem 4.9 saying N_L = 0 is necessary for a Dirac complement. The class is canonical on the quotient a_ab(L), and the paper computes it in enough examples that it is not just formal: so(3), Heisenberg, ideals, twisted Poisson structures, and the Hopf fibration. Theorem 6.9 (definite invariant pairing implies the diagonal has a complement iff the Lie algebra is abelian) is clean, and Corollary 6.10 (compact semisimple real Lie algebras with Killing form have no diagonal complement) is a good contrast to the complex semisimple case. The local existence result, Proposition 2.24, and the N_L = 0 but no-complement example, Proposition 5.14, also do real work. I checked the main argument paths and found no fatal error and no overclaim: the paper is explicit that N_L often vanishes and that the converse of Theorem 4.9 is false.\n\nThe soft spots are real but addressable. Lemma 4.2 is the load-bearing canonicity computation (d_{M'} = d_M + [ω,·]) and as written it does not close: the final 2-to-1 cancellation depends on sign conventions for the extended Axiom (A2) that are never stated, so I could not reproduce the coefficient. The paper itself contains the fix: Appendix A, Proposition A.7(4), proves the same twist-invariance for general curved L∞ algebras by a short argument that only uses the definition of the quotient. The authors should either expand the Lemma 4.2 computation or explicitly route the main proof through the appendix. Second, the grading of the quotient is a source of genuine confusion: Lemma 4.4 defines a_ab(L) = Γ(∧•L)[1]/[a,a], which puts N_M ∈ Γ(∧³L) in degree 2, while Definition 4.8 and Proposition 4.7 state H³(a_ab(L)); Example 5.10 then uses the natural unshifted grading and H³_dR. That is a silent switch that should be fixed in the statement. Sign conventions for the coadjoint action in §3.2–§5.1 are also unspecified, so the examples are not verbatim reproducible until one convention for L_A on g* is stated.\n\nNone of this undermines the main results in my view. The quotient descent is standard, Theorem 6.9 rests on Jacobson and Bourbaki and checks out, and the computations I could verify are consistent. Anyone working on Dirac structures, Lie bialgebroids, or deformation theory gets value from this paper. It deserves a serious referee, not a desk rejection; I would send it out with a request to tighten Lemma 4.2 and to state the gradings and sign conventions explicitly.","headline":"A genuinely new and mostly sound obstruction class for Dirac complements, with a clean definite-pairing theorem; the main line survives a dense Lemma 4.2 that needs rewriting for reproducibility.","tokens_in":28546,"tokens_out":2161,"would_cite":true,"duration_ms":21946,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53D17","17B62"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that a Dirac structure admits a Dirac complement only if its canonical obstruction class vanishes, and that for a definite pairing the diagonal in g⊕ḡ is complementable only when g is abelian.","keywords":["Dirac structures","Courant algebroids","Dirac complements","obstruction class","abelianized cohomology","quadratic Lie algebras","Cartan-Dirac structure","curved L∞ algebras"],"falsifier":"Search g = so(3) with its definite invariant pairing for an orthogonal automorphism φ with no nonzero fixed vector and check whether graph(φ) is a Dirac complement to the diagonal; by Lemma 6.6 every Dirac structure in g⊕ḡ is such a graph, so a single fixed-point-free orthogonal automorphism would refute Theorem 6.9. The paper's own calculation shows why none exists: the Maurer-Cartan equation forces (2 + λ_1^2 + λ_2^2 + λ_3^2) A_1∧A_2∧A_3 = 0, which has no real solution.","tokens_in":27372,"feed_emoji":"🚫","tokens_out":12671,"duration_ms":118039,"temperature":0.7,"pith_summary":"Dirac structures unify closed 2-forms and Poisson bivectors; the paper asks when one can be complemented by a transverse Dirac structure. Such complements matter because they make a Courant algebroid into the double of a Lie bialgebroid and put the deformation theory of Dirac structures under a simpler differential graded Lie algebra. The paper establishes two main results. First, it defines a canonical cohomology class N_L for any Dirac structure L and proves that N_L must vanish if a Dirac complement exists, so nonvanishing gives new families with no complement. Second, for a quadratic Lie algebra with definite invariant pairing, it proves that the diagonal in g⊕ḡ admits a Dirac complement if and only if g is abelian; real compact semisimple Lie algebras with their definite invariant pairing therefore never admit one.","feed_headline":"Definite pairings block Dirac complements unless g is abelian","feed_subtitle":"The diagonal in g⊕ḡ pairs up only when g is abelian; a new class explains other failures.","key_machinery":"The machinery is the abelianized cohomology a_ab(L) = Γ(∧•L)[1] / [a(L),a(L)], with the canonical differential d induced by d_M for any lagrangian complement M. Quotienting by bracket terms removes the quadratic term from the curved Maurer-Cartan equation — the equation whose solutions describe Dirac complements — reducing it to a linear equation controlled by the class N_L = [N_M]. In the definite-pairing setting, the key mechanism is Lemma 6.6, a bijection between Dirac structures in g⊕ḡ and orthogonal automorphisms of g, which converts the complement problem into a question about fixed-point-free automorphisms.","core_discovery":"The central object is the obstruction class N_L = [N_M] in H^3(a_ab(L)). Here M is any auxiliary lagrangian complement, N_M is the Nijenhuis tensor measuring how far M is from being Dirac, and a_ab(L) is the quotient of Γ(∧•L)[1] by its commutator subalgebra. The paper shows this class is independent of M, and that any Dirac complement forces it to vanish. For abelian Dirac structures the converse also holds: N_L = 0 exactly when a complement exists. On the Lie-theoretic side, the paper classifies all Dirac structures in g⊕ḡ when the pairing on g is definite as graphs of orthogonal automorphisms of g, then shows the diagonal Δ admits a complement only if such an automorphism is fixed-point-","pith_inferences":["The paper itself observes that N_Δ = 0 for every diagonal while Theorem 6.9 forbids complements in the definite case; the definite-pairing rigidity is therefore invisible to the new class and will require a finer invariant.","The orthogonal-automorphism classification suggests a concrete route into the open indefinite case: search for orthogonal maps whose fixed-point behavior makes their graphs transverse to Δ, for instance on so(4) with a split-signature invariant pairing.","Because the quotient a_ab(L) frequently vanishes for regular Dirac structures, the useful range of N_L is likely concentrated on Dirac structures with large isotropy, such as the abelian core examples in Section 5.","The appendix's short proof of twist-invariance for curved L∞ algebras indicates that the obstruction class could be re-derived without the sign-sensitive computation of Lemma 4.2, yielding a more conceptual formulation for future applications."],"forward_implications":["Dirac structures with N_L ≠ 0 — including structures built from so(3), the Heisenberg algebra, and twisted Poisson structures with a support condition — are certified to admit no Dirac complement.","For abelian Dirac structures the obstruction class is a complete invariant: N_L = 0 is equivalent to the existence of a Dirac complement.","For real compact semisimple Lie algebras with a definite invariant pairing, the diagonal Δ in g⊕ḡ admits no Dirac complement, in sharp contrast with the complex semisimple case where the Cartan-Dirac structure does admit complements.","Since lagrangian complements always exist and local Dirac complements always exist in exact Courant algebroids, the obstruction class isolates a purely global, integrability-level reason for failure.","The same quotient-and-curvature construction works for arbitrary curved L∞ algebras, giving a general obstruction to the existence of Maurer-Cartan elements in deformation problems."],"fun_headline_variants":["Obstruction class halts Dirac complements","Nonabelian g blocks diagonal complement","Definite pairings cap Dirac complements","Dirac complement exists only for abelian g","New class kills Dirac complement candidates"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The entire obstruction-class construction depends on the claim that the differential on the abelianized complex is independent of the chosen auxiliary lagrangian complement; that independence is proven by a delicate, sign-heavy computation (Lemma 4.2), and if a sign convention is off the class N_L is not canonical — though the paper's appendix gives a shorter argument in a more general setting that mitigates the risk.","fun_headline_variants_meta":{"raw":{"variants":["Obstruction class halts Dirac complements","Nonabelian g blocks diagonal complement","Definite pairings cap Dirac complements","Dirac complement exists only for abelian g","New class kills Dirac complement candidates"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000249,"raw_usage":{"total_tokens":1355,"prompt_tokens":684,"completion_tokens":671,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":428,"completion_tokens_details":{"reasoning_tokens":608}},"tokens_in":428,"tokens_out":671,"duration_ms":6602,"temperature":1.0,"reasoning_tokens":608,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T23:29:23.658433+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Search g = so(3) with its definite invariant pairing for an orthogonal automorphism φ with no nonzero fixed vector and check whether graph(φ) is a Dirac complement to the diagonal; by Lemma 6.6 every Dirac structure in g⊕ḡ is such a graph, so a single fixed-point-free orthogonal automorphism would refute Theorem 6.9. The paper's own calculation shows why none exists: the Maurer-Cartan equation forces (2 + λ_1^2 + λ_2^2 + λ_3^2) A_1∧A_2∧A_3 = 0, which has no real solution.","supporting_citations":[],"review_version":1}