{"id":"5da3a793-0310-442a-b19c-418b24fb4018","arxiv_id":"2507.17604","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A canonical generalized Levi-Civita connection is defined for any pair (G, div), and its full curvature is decomposed into classical metric, three-form, and dilaton-like data.","lead":"This paper constructs a canonical torsion-free metric connection for any generalized metric and divergence operator on an exact Courant algebroid, and computes its curvature in terms of ordinary Riemannian data. The result gives a standardized tool-kit of generalized curvature invariants, including two new scalars, for use in generalized geometry and supergravity.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The construction hinges on Lemma 2.2; the proof is compressed and the trace map is implicit, but explicit unpacking shows the nondegeneracy claim is correct, so the central argument stands.","rationale":"The reader identified Lemma 2.2 as the weakest assumption, and I agree that it is the most load-bearing algebraic step. However, after unpacking the proof, the nondegeneracy claim appears correct: the parity decomposition works, the trace map respects the odd/even splitting, and the resulting kernel is a direct sum of definite subspaces. I therefore found no mathematical flaw that would change the ACCEPT verdict. The residual issues are expositional: the trace map is not explicitly defined in the text, and the proof of Proposition 3.4 omits some eps factors in the trace-free calculation. These are worth fixing in revision but do not undermine the central claim. The paper has independent support from the comparison with the NS-NS supergravity formulas and the vector-deformed supergravity equations, which would be unlikely to match if the sign or structure of the canonical connection and its curvature were wrong.","tokens_in":17535,"tokens_out":42520,"duration_ms":454348,"concrete_test":"Compute the Gram matrix of the kernel of tr explicitly for V = R^{p,q} with (p,q) = (1,1), (1,2), (2,2) using the decomposition (2.4)-(2.5), with the trace convention tr(S)(v) = sum_i eps_i S(u_i, v, u_i). Verify that the Gram matrix is nondegenerate and that its odd- and even-parity blocks are respectively negative and positive definite. This directly tests the only algebraic fact on which the existence of S_alpha in Theorem 2.1 rests.","verdict_should_be":"UNCHANGED","load_bearing_attack":"No significant objection identified. The existence and uniqueness of the canonical connection D^{G,div} in Theorem 2.1 depends on Lemma 2.2: if the kernel of the trace map tr: so(E)^{(1)}_G -> E* were degenerate, the orthogonal complement K=(ker tr)^perp would not be a complement and the section S_alpha with prescribed trace need not exist. The proof of Lemma 2.2 is compressed at the step where the equation dS=0 'decouples into four independent equations' and at the final splitting of the trace kernel into odd and even parts. Working through the linear algebra, with the trace convention implicit in Proposition 3.4 (equivalently tr(S)(v) = sum_i eps_i S(u_i, v, u_i), which differs by a sign from the first-two contraction on the first prolongation), the claimed parity decomposition is valid: the odd-parity summands of (2.4) are negative definite, the even-parity summands of (2.5) are positive definite, and the trace map sends odd pieces to L* and even pieces to P*, so the kernel is a direct sum of a negative-definite and a positive-definite subspace. The orthogonality of S to ker tr in Proposition 3.4 also checks out once the trace-free identity is written with the correct eps factors. Thus the reader's flagged assumption is real but sound; the weakness is presentational rather than mathematical.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs, for an exact Courant algebroid with a semi-Riemannian generalized metric G and a divergence operator div, a canonical generalized Levi-Civita connection D^{G,div}, thereby resolving the non-uniqueness of torsion-free metric generalized connections. It then computes the generalized Riemann tensor of D^{G,div} and decomposes it into classical data: the ordinary Riemann tensor of the underlying metric, the closed three-form H, and a section e encoding the divergence difference. The main curvature formulas appear in Theorems 4.1 and 4.2 (pure- and mixed-type components), with corollaries for the full generalized Ricci tensor, the generalized Ricci tensor, two scalar curvatures, and a newly defined generalized Kretschmann scalar. Section 5 compares the results with the physics literature, recovering known expressions for NS-NS supergravity and vector-deformed supergravity.","tokens_in":17791,"tokens_out":8632,"duration_ms":84695,"significance":"If the results are correct, this is a substantial contribution to generalized geometry: it turns the non-uniqueness of generalized Levi-Civita connections into a canonical choice, makes the generalized Riemann tensor an invariant of the pair (G, div), and provides explicit master formulas that can be used in applications. The curvature computations are carried out in detail and are benchmarked against the existing literature [1, 11, 14], which gives strong independent consistency checks. The new scalar-valued invariants, including the generalized Kretschmann scalar, are potentially useful for future work on curvature invariants and supergravity applications. The main caveat is the compressed proof of Lemma 2.2, which is load-bearing for the existence and uniqueness of the canonical connection; the algebraic argument is sketched rather than fully demonstrated. This is a presentational gap in an otherwise sound construction.","major_comments":[{"comment":"The proof of nondegeneracy of the kernel of the trace map tr: so(E)^{⟨1⟩}_G → E* is too compressed at the key step. The sentence \"It clear that the equation ∂S = 0 decouples into four independent equations corresponding to the tensor power of L\" is not a proof: one must explicitly show that the four components of ∂S with different L-degree vanish independently, and that the subspaces so(V)^{⟨1⟩}_{odd} and so(V)^{⟨1⟩}_{even} are definite with respect to the relevant inner product. In addition, the splitting ker(tr: so(V)^{⟨1⟩}_{odd} → L*) ⊕ ker(tr: so(V)^{⟨1⟩}_{even} → P*) requires an explicit verification that the trace of an odd-parity element takes values in L* and that of an even-parity element in P*. Because the existence of the complementary subbundle K = (ker tr)⊥ and the unique section S_α with prescribed trace depend on this lemma, the proof must be completed and presented in detail.","section":"Lemma 2.2, §2"}],"minor_comments":[{"comment":"The notation \"TM = TM ⊕ T*M\" is confusing because the symbol TM is used both for the generalized tangent bundle and for the ordinary tangent bundle. Please use a distinct notation, e.g. \\mathbb{T}M or E, for the generalized tangent bundle.","section":"§2, first paragraph of proof of Theorem 2.1"},{"comment":"The classification line reads \"MSc classification\"; this should be \"MSC classification\".","section":"Abstract and header"},{"comment":"There is a typo \"connnection\" in the sentence \"The Riemann tensor of the canonical generalised Levi-Civita connnection\".","section":"Introduction, page 6"},{"comment":"The phrase \"It clear that the equation ∂S = 0 decouples\" should read \"It is clear that the equation ∂S = 0 decouples\".","section":"Lemma 2.2, proof"},{"comment":"In the expression for 4Rc(G, div)±_μν, the meaning of the ± and ∓ signs (one for each chirality) should be stated explicitly, since the left-hand side contains a superscript ± that is not defined in the text.","section":"Section 5, index notation for the generalized Ricci tensor"},{"comment":"The non-triviality check (|Rm^D|^2_G)|_{H=0,e=0} = 4|Rm|^2_g is stated without derivation. A short justification, or a reference to the relevant formulas, would be helpful.","section":"Proposition 4.6"}],"recommendation":"major_revision","confidential_remarks":"The paper is mathematically strong and the central construction appears sound. The main issue is the proof of Lemma 2.2, which is load-bearing and currently too terse; this should be expanded before publication. The definition of the generalized Riemann tensor in equation (3.1) is taken from the same-authors preprint [2], which is not yet published; the editor may wish to ensure that [2] is publicly available or that the definition is self-contained. Otherwise the paper fits the journal well and is likely to be of significant interest to the generalized geometry and supergravity communities."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read the whole thing, including the compressed Lemma 2.2 proof. The central construction holds up. The paper does something genuinely useful: it resolves the non-uniqueness of generalized Levi-Civita connections by selecting one canonically from the pair (G, div), and then proves that the generalized Riemann tensor of that connection is an invariant of the pair. The decomposition into ordinary Riemann curvature, the closed three-form H, and the section e = 2(X+ξ) is the advertised master formula, and the follow-up toolkit (full Ricci, generalized Ricci, two scalar curvatures, Kretschmann) is exactly what people doing applications will reach for. The comparisons with [1], [11], and [10] check out; the match with NS-NS supergravity and vector-deformed supergravity is real evidence that the signs and factors are right. Credit where due: the pure-type formulas for general divergence, the second scalar curvature, and the Kretschmann invariant are new, and the paper is honest about which pieces appear in earlier work.\n\nSoft spots are mild. The proof of Lemma 2.2, which carries the existence of the canonical connection, is too compressed at the 'decouples into four independent equations' step. I worked through the linear algebra with the trace convention from Proposition 3.4, and the claimed parity decomposition is correct: odd pieces sit in negative-definite summands, even pieces in positive-definite ones, and the trace maps them to L* and P* respectively. So the weakness is presentational, not mathematical. A referee should ask for a few more lines there, not a new proof. The other caveat is external: the generalized Riemann tensor is imported from the companion preprint [2]. That is not circular, and the special-case comparisons give independent anchors, but an external published source for (3.1) would tighten things.\n\nMath, data, citations: all look solid. No code or data, which is normal for pure math. The reader's flagged assumption is real and sound.\n\nWho is this for? People working in generalized geometry and in supergravity/string theory applications. It deserves a serious referee. I would send it out with a request to expand Lemma 2.2 and to note the preprint dependence. I would also cite it.","headline":"A careful, genuinely useful canonical construction that resolves non-uniqueness of generalized Levi-Civita connections; the key nondegeneracy lemma is compressed but correct, and the paper deserves refereeing.","tokens_in":18328,"tokens_out":2004,"would_cite":true,"duration_ms":19358,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53D18","83C10"],"pacs":[],"model":"deepseek-v4-flash","headline":"For every generalized metric and divergence operator on an exact Courant algebroid, this paper constructs a canonical generalized Levi-Civita connection whose curvature is an invariant of the pair.","keywords":["generalised geometry","Courant algebroids","generalised Levi-Civita connection","generalised metric","divergence operator","generalised Riemann tensor","generalised Einstein equations","supergravity"],"falsifier":"Compute the Gram matrix of the kernel of $\\mathrm{tr}: \\mathrm{so}(V)^{\\langle 1\\rangle}\\to V^*$ on a pseudo-Euclidean space with a mixed-signature metric (positive and negative definite parts both nonzero, e.g. signature $(2,2)$) and check whether it is degenerate; a degenerate kernel would exhibit a nonzero $S$ with $\\mathrm{tr}(S)=0$ but $\\langle S,S\\rangle=0$, breaking the claimed uniqueness of $S_\\alpha$ and invalidating $D^{\\mathcal G,\\mathrm{div}}$ for arbitrary $\\mathrm{div}$. The paper's own proof asserts the kernel is a direct sum of two definite subspaces, so this is a direct numerical test.","tokens_in":17315,"feed_emoji":"","tokens_out":9944,"duration_ms":86203,"temperature":0.7,"pith_summary":"This paper solves a well-known ambiguity in generalized geometry: generalized Levi-Civita connections are not unique, so curvature quantities built from them depend on a choice. The authors construct, for each exact Courant algebroid equipped with a (semi-Riemannian) generalized metric $\\mathcal G$ and a divergence operator $\\mathrm{div}$, a single canonical torsion-free metric connection $D^{\\mathcal G,\\mathrm{div}}$. Its generalized Riemann tensor is therefore an invariant of the pair $(\\mathcal G,\\mathrm{div})$, not of an arbitrary choice. The paper's main formulas express that tensor, and the associated full Ricci tensor, Ricci tensor, and three scalar invariants, entirely in terms of ordinary Riemann curvature, the closed three-form $H$ of the Courant algebroid, and the section $e = 2(X+\\xi)$ that records the difference between $\\mathrm{div}$ and the metric divergence. A reader should care because these master formulas make generalized geometry directly usable in gravitational and string-theoretic settings, recovering standard supergravity equations and extending them to vector-deformed supergravity.","feed_headline":"Generalized geometry gets a canonical Levi-Civita connection","feed_subtitle":"Curvature becomes an invariant of the pair (metric, divergence), with explicit formulas in classical data.","key_machinery":"The load-bearing construction is the trace-map orthogonal projection inside the first generalized prolongation, the space of tensors $A\\in (\\mathrm{so}(E)_{\\mathcal G})^{\\langle 1\\rangle}\\subset E^*\\otimes \\Lambda^2 E^*$ satisfying the algebraic condition that the cyclic sum $\\partial A$ vanishes. For a pair $(\\mathcal G,\\mathrm{div})$, the canonical connection is $D^{\\mathcal G,\\mathrm{div}}=D^0+S$, where $S$ is the unique section of this prolongation perpendicular to the kernel of the trace map $S\\mapsto \\mathrm{tr}(S)$ and with prescribed trace $\\alpha=\\mathrm{div}-\\mathrm{div}^{\\mathcal G}$; Lemma 2.2 supplies the nondegeneracy of the kernel that makes $S$ well-defined. The curvature computation then reduces to using the closed form $S=(\\chi^{e_+}_+ + \\chi^{e_-}_-)/(d-1)$ and expanding $\\mathrm{Rm}^{D^0+S}$ into $D^0$-derivatives of $S$ plus algebraic terms quadratic in $S$, producing the master formulas of Theorems 4.1 and 4.2.","core_discovery":"The central claim is that for any exact Courant algebroid with generalized metric $\\mathcal G$ and divergence operator $\\mathrm{div}$, there exists a canonical generalized Levi-Civita connection $D^{\\mathcal G,\\mathrm{div}}$, obtained as $D^0+S$, where $D^0$ is the extension of the Levi-Civita connection of the induced metric with metric divergence, and $S$ is the unique section of the first generalized prolongation $\\mathrm{so}(E)^{\\langle 1\\rangle}_{\\mathcal G}$ that has trace $\\mathrm{div}-\\mathrm{div}^{\\mathcal G}$ and is orthogonal to the trace-free tensors. Its generalized Riemann tensor $\\mathrm{Rm}^{D}$ decomposes into pure-type and mixed-type components expressed in closed form by Theorems 4.1 and 4.2 in terms of the ordinary Riemann tensor, the closed three-form $H$, and the section $e$ with $\\langle e,\\cdot\\rangle=\\mathrm{div}^{\\mathcal G}-\\mathrm{div}$. Tracing these formulas yields the full generalized Ricci tensor, the generalized Ricci tensor, and three scalar invariants, two of which are new (the second generalized scalar curvature and the generalized Kretschmann scalar). In the metric-divergence case $e=0$ the formulas recover a known class of generalized Levi-Civita connections from the literature, and in the dilaton and vector-deformed cases they match, respectively, the NS-NS supergravity equations and the field equations of vector-deformed supergravity.","pith_inferences":["The canonical connection gives a preferred representative of the equivalence class of generalized Levi-Civita connections with fixed divergence, which may make generalized Ricci flow and generalized Einstein equations into well-posed evolution problems rather than gauge-choice-dependent ones.","The explicit dependence on $e$ suggests a systematic classification of flat pairs $(\\mathcal G,\\mathrm{div})$: solving $\\mathrm{Rm}^{D}=0$ as equations for $g$, $H$, and the vector field $X$ and one-form $\\xi$ could characterize exact Courant algebroids admitting Ricci-flat generalized geometry.","The new scalar invariants, particularly the generalized Kretschmann scalar, invite applications to curvature singularity and extendability questions in generalized geometry, a direction the paper notes only in passing.","Because the construction works for semi-Riemannian metrics of any signature, it may be directly relevant to timelike T-duality and other signature-changing settings in gravitational theories."],"forward_implications":["Any invariant computed from $D^{\\mathcal G,\\mathrm{div}}$, in particular its generalized Riemann tensor and the new scalar invariants, is a genuine invariant of the pair $(\\mathcal G,\\mathrm{div})$; no curvature component has to be discarded for depending on a connection choice.","The master formulas reduce generalized curvature computations to classical data: ordinary Riemann curvature, the closed three-form $H$, and the section $e$; one no longer needs to track the full generalized connection.","Tracing the formulas produces explicit expressions for the full generalized Ricci tensor, the generalized Ricci tensor, and three scalar invariants, ready for use in generalized Einstein equations.","Under the standard dilaton identification $\\xi=2\\,d\\varphi$, $X=0$, the formulas reproduce the NS-NS supergravity field equations; under the compatibility conditions of vector-deformed supergravity, vanishing of the second generalized scalar curvature is part of the field equations.","The generalized Kretschmann scalar $|\\mathrm{Rm}^{D}|^2_{\\mathcal G}$ is a new invariant that in the flat case $H=0$, $e=0$ reduces to $4|\\mathrm{Rm}|^2_g$, so it can serve as a gravitational singularity probe in generalized geometry."],"supporting_citations":[{"why":"Defines semi-Riemannian generalized metrics and the isomorphism conditions used to set up the pair $(\\mathcal G,\\mathrm{div})$.","marker":"[12]"},{"why":"Supplies the definition of the first generalized prolongation used in Lemma 2.2 and in the construction of $S$.","marker":"[13]"},{"why":"Provides the comparison generalized connection; the canonical construction reproduces its formulas in the metric-divergence case.","marker":"[1]"},{"why":"Establishes that mixed-type Ricci components and the generalized scalar curvature are independent of the chosen connection with fixed divergence, used to identify invariants.","marker":"[14]"},{"why":"The supergravity-as-generalized-geometry reference whose NS-NS action and field equations are recovered in the dilaton specialisation.","marker":"[11]"},{"why":"The vector-deformed supergravity reference whose field equations are matched and partly explained by the new curvature invariants.","marker":"[10]"}],"fun_headline_variants":["Canonical generalized Levi-Civita connection defined","Unique connection from metric and divergence in exact Courant algebroids","Generalized Riemann tensor decomposed into classical data","New curvature invariants from canonical generalized connection","Resolving non-uniqueness with a canonical generalized connection"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole construction stands on the algebraic claim, proved in Lemma 2.2, that the kernel of the trace map on the first generalized prolongation $\\mathrm{so}(E)^{\\langle 1\\rangle}_{\\mathcal G}$ is nondegenerate; if that kernel were degenerate for some signature or dimension, the unique section $S$ and hence the canonical connection would not exist for arbitrary divergence operators.","fun_headline_variants_meta":{"raw":{"variants":["Canonical generalized Levi-Civita connection defined","Unique connection from metric and divergence in exact Courant algebroids","Generalized Riemann tensor decomposed into classical data","New curvature invariants from canonical generalized connection","Resolving non-uniqueness with a canonical generalized connection"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000239,"raw_usage":{"total_tokens":1568,"prompt_tokens":1056,"completion_tokens":512,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":672,"completion_tokens_details":{"reasoning_tokens":435}},"tokens_in":672,"tokens_out":512,"duration_ms":5239,"temperature":1.0,"reasoning_tokens":435,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T14:45:27.311559+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the Gram matrix of the kernel of $\\mathrm{tr}: \\mathrm{so}(V)^{\\langle 1\\rangle}\\to V^*$ on a pseudo-Euclidean space with a mixed-signature metric (positive and negative definite parts both nonzero, e.g. signature $(2,2)$) and check whether it is degenerate; a degenerate kernel would exhibit a nonzero $S$ with $\\mathrm{tr}(S)=0$ but $\\langle S,S\\rangle=0$, breaking the claimed uniqueness of $S_\\alpha$ and invalidating $D^{\\mathcal G,\\mathrm{div}}$ for arbitrary $\\mathrm{div}$. The paper's own proof asserts the kernel is a direct sum of two definite subspaces, so this is a direct numerical test.","supporting_citations":[{"cited_title":"Classification of generalized einstein metrics on 3-dimensional lie groups","cited_arxiv_id":null,"evidence_quote":"Defines semi-Riemannian generalized metrics and the isomorphism conditions used to set up the pair $(\\mathcal G,\\mathrm{div})$."},{"cited_title":"Springer,","cited_arxiv_id":null,"evidence_quote":"Provides the comparison generalized connection; the canonical construction reproduces its formulas in the metric-divergence case."},{"cited_title":"Ricci flow on courant algebroids.Communications in Contemporary Mathematics, page 2550037, 2025","cited_arxiv_id":null,"evidence_quote":"Establishes that mixed-type Ricci components and the generalized scalar curvature are independent of the chosen connection with fixed divergence, used to identify invariants."}],"review_version":1}