Some properties of a Brauer class

Qixiao Ma

REVIEW 3 major objections 4 minor 18 references

Some properties of a Brauer class

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper proves that the Brauer class obstructing the relative Picard functor of a curve admits an involution of the second kind on its division algebra and splits at the generic point of the theta divisor.

desk verdict The involution theorem is not proved as written—the chosen pairing need not be σ-invariant—but the theta-splitting results are genuinely interesting and worth referee time. read the letter →

arxiv 1908.03131 v1 pith:MDUVTPRM submitted 2019-08-08 math.AG math.RA

Qixiao Ma This is my paper · ORCID

classification math.AGmath.RA MSC 14H6014K3014F2216K50

keywords BrauerclassPicardschemedivisionalgebrainvolutionofthesecondkindthetadivisorgeneralizeduniversalcurvetautologicallinebundle

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper studies the Brauer class $\alpha$ on the Picard scheme of a smooth proper curve that measures the obstruction to the relative Picard functor being representable by an actual line bundle on the product of the curve with its Picard scheme. It establishes that the division algebra representing $\alpha_0$ at the generic point of $\mathrm{Pic}^0_{X/k}$ has an involution of the second kind extending the natural involution induced by dualizing line bundles. It also proves that the restriction of the class to the generic point of the $\theta$ divisor in $\mathrm{Pic}^{g-1}_{X/k}$ is zero, and that in the universal genus-$g$ case the restriction to generalized $\theta$ divisors in $\mathrm{Pic}^0_{X/k}$ is zero. These splitting results say the obstruction disappears at height-one loci of the Picard scheme, so a tautological line bundle exists over the generic points of those divisors. The interest is that this gives concrete structural information about a natural Brauer class and its division algebra, including a property relevant to the open question of whether that division algebra contains an Azumaya order.

What carries the argument

The central object is the obstruction class $\alpha = d_0^{0,1}(1_{\mathrm{Pic}_{X/k}}) \in \mathrm{Br}(\mathrm{Pic}_{X/k})$ coming from the low-degree terms of the Leray spectral sequence for the projection $X \times \mathrm{Pic}_{X/k} \to \mathrm{Pic}_{X/k}$. On the generic point, $\alpha_0$ is represented by the endomorphism algebra of $H^0(X_{K^{\mathrm{sep}}}, \omega_{X_{K^{\mathrm{sep}}}} \otimes \mathcal{L})$ for the unique tautological line bundle $\mathcal{L}$. For the involution, the load-bearing identity is a $\sigma$-equivariant, Galois-equivariant perfect pairing between $H^0(\omega \otimes \mathcal{L})$ and $H^0(\omega \otimes \mathcal{L}^\vee)$, obtained from multiplication of sections into $H^0(\omega^{\otimes 2})$; Skolem–Noether converts this pairing into the required semilinear automorphism of the algebra. For the splitting theorems, the key mechanism is the projectivized direct image $\mathbb{P}(\pi'_*\mathcal{L}')$ over the $\theta$ divisor, with descent data that make the tautological line bundle descend to the total space; because $\pi'_*\mathcal{L}'$ has generic rank 1, the projectivization is birational and the descended bundle gives the splitting over $k(\Theta)$ (and over $k(\Theta_E)$ in the universal case).

What would settle it

Find a smooth proper curve $X$ and a semistable rank-2 bundle $E$ for which the generalized $\theta$ divisor $\Theta_E$ is nonreduced, and compute the generic corank of the connecting homomorphism $\delta'$ in the exact sequence of Section 5.4. If the corank is at least 2, then $\pi'_*(E'\otimes\mathcal{L}')$ has generic rank greater than 1, so the projectivization in Theorem 5.5 is not birational and the argument for splitting at $k(\Theta_E)$ breaks down.

Watch

Extended reading notes

Core claim

At the generic point of $\mathrm{Pic}^0_{X/k}$, the Brauer class $\alpha_0$ is represented by a division algebra $D$ over $K = k(\mathrm{Pic}^0_{X/k})$. The paper establishes three properties. First, $D$ carries an involution of the second kind extending the involution $\sigma$ on $\mathrm{Pic}^0_{X/k}$ induced by $L \mapsto L^\vee$. Second, the class $\alpha_{g-1}$ restricts to zero in $\mathrm{Br}(k(\Theta))$ for the $\theta$ divisor $\Theta \subset \mathrm{Pic}^{g-1}_{X/k}$. Third, under the universal-curve hypothesis, $\alpha_0$ restricts to zero in $\mathrm{Br}(k(\Theta_E))$ for the generalized $\theta$ divisor attached to a semistable rank-2 slope-$(g-1)$ vector bundle $E$. The splitting results are proved by descending a tautological line bundle over the projectivization of the direct image of the tautological bundle restricted to the divisor, and the descent is birational because the direct-image sheaf has generic rank 1.

Load-bearing premise

The load-bearing premise is that the direct-image sheaf on the theta divisor has generic rank 1, which the paper derives from reducedness of the divisor; in the generalized-theta case that reducedness is known only for universal curves and can fail for arbitrary curves or higher-rank bundles.

Editorial extensions

If this is right

Because $\alpha_{g-1}$ vanishes over $k(\Theta)$, a tautological line bundle exists over the generic point of the theta divisor, so the obstruction splits at that height-one point of $\mathrm{Pic}^{g-1}_{X/k}$.
The division algebra of $\alpha_0$ has an involution of the second kind extending the dual involution $L \mapsto L^\vee$, and the same construction gives an analogous involution for $\alpha_{g-1}$ extending $L \mapsto \omega_X \otimes L^\vee$.
For the universal genus-$g$ curve, $\alpha_0$ splits at the generic point of every generalized theta divisor $\Theta_E$ built from a semistable rank-2 bundle of slope $g-1$.
The proof's template shows that any height-one point of the Picard scheme where the restricted direct-image sheaf has generic rank 1 yields a splitting of the corresponding restriction of $\alpha$.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The paper's method suggests a broader criterion: whenever a divisor in a Picard scheme is reduced at its generic point and the relevant direct image has generic rank 1, the same projectivization-and-descent construction should split the Brauer class over the divisor's function field.
If the division algebra of $\alpha_0$ contains an Azumaya order compatible with the constructed involution, the paper's open question would be answered affirmatively; if not, the natural involutions could still help build an unramified division algebra without an Azumaya order, in the direction of the reference [AW14].
A testable next step is to compute the generic rank of $\pi'_*(E'\otimes\mathcal{L}')$ on a nonreduced generalized theta divisor, such as those cited from [HP15]; a rank jump would show exactly where the universal-curve hypothesis is needed.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Colleague,

The paper's flagship theorem, the second-kind involution on the division algebra at the generic point of Pic^0, is not proved as written. The stress-test note points at the real gap: Theorem 3.8 finds a K-point in the open set of perfect pairings, but σ-equivariance of the resulting pairing requires that point to be fixed by σ. A K-rational point in a σ-invariant open subset need not be σ-fixed; the example of G_m inside P^1 with x ↦ -x shows the issue is concrete. The proof would need an additional argument, for instance showing the open set has a σ-fixed point defined over the fixed field, to work.

That said, the paper is not sloppy. The splitting theorems are genuinely new, and the approach is coherent. The idea of testing whether the obstruction class dies at the generic point of the theta divisor by descending the projectivization of a direct image is elegant, and the descent lemmas (4.2, 4.3) are carefully written. The paper also deserves credit for raising the Azumaya-order question, which is a good research prompt.

The soft spots are concentrated in proof details. Lemma 3.4 asserts a Galois-equivariant isomorphism σ*L ≅ L^∨ from the triviality of σ*L ⊗ L with a one-sentence argument; a K-section of the tensor product does not automatically give a G-equivariant isomorphism after base change. This affects the setup of Proposition 3.6, though it may be fixable by choosing the section more carefully. Lemma 4.1's corank argument is compressed: reducedness of the theta divisor implies regularity at the generic point, but the step from corank ≥2 to non-regularity needs more spelling out. The generalized theta case in Section 5 explicitly relies on the universal-curve hypothesis for reducedness of Θ_E, and the paper is honest about that limitation, so I would not count that as a flaw beyond scope.

Overall, the main theorem likely needs nontrivial repair, but the splitting results may survive. This is a worthwhile preprint for anyone working on Brauer groups of Picard schemes, and it deserves a serious referee, though not acceptance as is. I would not cite the involution theorem without checking a corrected version; the splitting results might be citable after verification.

Recommendation: send to a good referee, but expect a report requiring substantive revision.

Referee Report

3 major / 4 minor

Summary. The manuscript studies the Brauer class α ∈ Br(Pic_{X/k}) obstructing the representability of the relative Picard functor for a smooth proper curve X over a field k. The author proves three main results: (i) the division algebra representing the restriction α_0 at the generic point of Pic^0_{X/k} admits an involution of the second kind extending the natural dual involution (Theorem 3.8); (ii) the class α_{g-1} splits at the generic point of the theta divisor Θ ⊂ Pic^{g-1}_{X/k} (Theorem 4.4); and (iii) under the universal-curve hypothesis, the class α_0 splits at the generic point of a generalized theta divisor Θ_E ⊂ Pic^0_{X/k} for a semistable rank-two slope-(g-1) vector bundle E (Theorem 5.5). The proofs use standard techniques: Leray spectral sequences, Brauer-Severi varieties, Galois descent, and projectivizations of direct images.

Significance. If the results are correct, they supply concrete structural information about an explicit Brauer class associated with the Picard scheme, a class that has been studied mainly through its period and index. The involution statement gives a positive answer for α_0 to a natural algebra-structure question, and the splitting results identify height-one loci where the obstruction vanishes, which is of interest for constructing Azumaya orders and for the geometry of theta divisors. The paper is careful to restrict the generalized-theta result to the universal curve, where reducedness of Θ_E is available. The manuscript also contains useful descent lemmas (Lemmas 4.2 and 4.3) that may be of independent use. However, as written, the proofs of the involution theorem contain several gaps that are central to the claim, so the paper needs substantial revision before the results can be considered established.

major comments (3)

[§3.4, Theorem 3.8] The proof of Theorem 3.8 chooses a K-rational point in the open set O of perfect pairings, but it does not justify that the corresponding linear form ℓ is fixed by the involution σ. The pairing u obtained as ℓ ∘ B is σ-equivariant only if ℓ(σ(w)) = σ(ℓ(w)) for all w, which is equivalent to ℓ being a σ-fixed point of the induced action on the dual projective space. The argument 'K is infinite' only guarantees a K-rational point, hence a G-invariant point, not a σ-fixed one. This is load-bearing because Proposition 3.6 explicitly requires a σ-equivariant pairing. The gap is repairable by observing that O, being defined over K and stable under σ, descends to a nonempty open subset of the projective space over K^σ, which must contain a K^σ-point since K^σ is infinite; but this argument is absent from the manuscript.
[Lemma 3.4] The proof of Lemma 3.4 asserts that σ^*L ⊗ L is the trivial line bundle on X_{K'} and then says 'Take any K-section of σ^*L ⊗ L would yield a G-equivariant isomorphism σ^*L ≅ L^∨.' Triviality over K' does not by itself provide a K-section, i.e., a G-invariant nonzero section. One must argue that the line bundle is trivial over X_K, for instance by the injectivity of Pic(X_K) → Pic(X_{K'}) (which follows from Hilbert 90 because H^0(X_{K'}, O^*) = K'^*), or by explicitly constructing the descent data. As written, the proof skips this step, and Proposition 3.6 relies on the existence of a G-invariant isomorphism φ: L^∨ → σ^*L.
[Theorem 3.8 and Lemma 3.5] Even if a σ-linear isomorphism Q: H^0(ω⊗L) → H^0(ω⊗L)^∨ is produced, the conclusion of Theorem 3.8 requires an involution of the second kind, i.e., a σ-semilinear anti-automorphism whose square is the identity. The manuscript proceeds 'by Proposition 3.6 and Lemma 3.5' without explaining how a σ-linear isomorphism of vector spaces yields an order-two anti-automorphism of the central simple algebra. The natural adjoint anti-automorphism attached to Q has square equal to an inner automorphism in general, unless Q satisfies a σ-Hermitian (or skew-Hermitian) condition. The proof does not verify this condition, nor does it invoke a standard result (such as the cohomological criterion in Lemma 3.1 or Albert's symmetrization theorem) to bridge the gap. Thus the existence of an involution extending σ is not established by the given argument.

minor comments (4)

[Lemma 5.2] The statement of Lemma 5.2 is garbled: 'the numerical class is always a multiple of 2g−2 / g. c. d(2g−2,d +g−1) Θ' should be written as a multiple of ((2g-2)/gcd(2g-2, d+g-1))·Θ, and the proof would benefit from a few more details relating the Néron-Severi class to the torsor class in H^1(k, Pic^0_{X/k}).
[Lemma 4.1] The sentence 'But T_U is reduced, so it is regular at the generic point' is imprecise: a reduced scheme is regular at its generic points, but this is being used for an irreducible divisor, so the wording should specify that T_U (or its component at the relevant generic point) is integral.
[Theorem 4.4] The application of Lemma 4.2 to F = π'_*(L') does not explicitly verify the hypotheses: the line bundle N on T_U ×_T T_U and the isomorphisms φ and β with the cocycle condition are not identified. The argument is plausible because the descent data come from the étale cover and the Brauer class, but the proof should spell out the data or cite the construction explicitly.
[General] There are numerous small typographical errors and missing spaces in the text (e.g., 'thetha' in the abstract, 'Kk^sep' for K k^sep, and inconsistent use of 'i' as an index starting from 0 in the matrices of Proposition 5.3). These do not affect the mathematics but should be corrected.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the proofs are deductive and rely on external cited results; no prediction is fitted or defined in terms of its conclusion.

full rationale

The paper derives properties of the Brauer class from standard external results in the Leray spectral sequence, Galois descent, Skolem–Noether, and theta-divisor theory, without fitting parameters or renaming empirical inputs. The central theorems (3.8, 4.4, 5.5) are not assumed as premises: the involutions are constructed from pairings, and the splitting statements are shown by descending tautological sheaves and exploiting birational projectivizations. All citations point to external literature ([Bos90], [Gir71], [KMRT98], [MV14], [Sch03], [Ray82], [HP15], etc.), with no load-bearing self-citation. The paper explicitly flags the restrictive hypothesis that generalized theta divisors are reduced (footnote 3, Lemma 5.4, Remark 4.5); this is a stated limitation rather than a circular assumption. Likewise, the possible gap in Theorem 3.8 about finding a sigma-fixed linear form is a proof-correctness concern, not a circularity: the theorem is not being used as an input to itself. There is no step in which an equation is equivalent to its own conclusion by construction.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The paper introduces no free constants and no new postulated entities. It relies on standard theorems from algebraic geometry and Galois cohomology; the most specialized inputs are the strong Franchetta theorem and Raynaud's theorem in the universal-curve setting.

assumptions (5)

domain assumption Representability of the relative Picard functor and the Leray spectral sequence construction of the obstruction class alpha (Bos90).
This is the foundational framework of the paper; the class alpha is defined through this background theory.
standard math The characterization of involutions of the second kind via corestriction in the Brauer group (KMRT98, 1.3).
Used in Lemma 3.1 to reduce the existence of an involution to a Brauer group vanishing condition.
domain assumption The strong Franchetta theorem for Pic(X) of the universal curve (Sch03).
Used in Lemma 5.2 and Proposition 5.3 to control line bundles and to construct semistable rank-2 vector bundles on the universal curve.
standard math Raynaud's theorem on vanishing of cohomology for semistable bundles and on the numerical class of generalized theta divisors (Ray82, 1.6.2, 1.8.1).
Used in Section 5 to ensure h^0(E_eta tensor L_eta)=0 and to prove reducedness of Theta_E in the universal case.
standard math Galois descent and Hilbert 90 for isomorphisms of vector spaces and algebras over fields (used in Lemmas 3.2 and 3.5).
Used to descend isomorphisms from the separable closure to the base field and to move between Brauer-Severi varieties and endomorphism algebras.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Some properties of a Brauer class." pith.science (2026). https://pith.science/paper/MDUVTPRM

@misc{pith2026190803131,
  author       = {Pith},
  title        = {Pith review of: Some properties of a Brauer class},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MDUVTPRM}},
  note         = {Machine review of arXiv:1908.03131}
}

abstract

Let $X$ be a smooth proper curve defined over a field $k$. The representability of the relative Picard functor is obstructed by a class $\alpha\in\mathrm{Br}(\mathrm{Pic}_{X/k})$. We show the associated division algebra on $\mathrm{Pic}^0_{X/k}$ has natural involutions. We show the class $\alpha$ splits at some height one points in $\mathrm{Pic}_{X/k}$.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

18 extracted references · 12 canonical work pages

[1]

Open problems on central simple algebras

Asher Auel, Eric Brussel, Skip Garibaldi, and Uzi Vishne. Open problems on central simple algebras. Transform. Groups , 16(1):219--264, 2011

work page 2011
[2]

Arbarello, M

E. Arbarello, M. Cornalba, P. A. Griffiths, and J. Harris. Geometry of algebraic curves. V ol. I , volume 267 of Grundlehren der Mathematischen Wissenschaften [Fundamental Principles of Mathematical Sciences] . Springer-Verlag, New York, 1985

1985
[3]

Unramified division algebras do not always contain A zumaya maximal orders

Benjamin Antieau and Ben Williams. Unramified division algebras do not always contain A zumaya maximal orders. Invent. Math. , 197(1):47--56, 2014

work page 2014
[4]

N\' e ron Models

Siegfried Bosch. N\' e ron Models . Springer Berlin Heidelberg, Berlin, Heidelberg, 1990

work page 1990
[5]

Derived categories of twisted sheaves on C alabi- Y au manifolds

Andrei Horia Caldararu. Derived categories of twisted sheaves on C alabi- Y au manifolds . ProQuest LLC, Ann Arbor, MI, 2000. Thesis (Ph.D.)--Cornell University

work page 2000
[6]

Cohomologie non ab\' e lienne

Jean Giraud. Cohomologie non ab\' e lienne . Springer-Verlag, Berlin-New York, 1971. Die Grundlehren der mathematischen Wissenschaften, Band 179

1971
[7]

Central simple algebras and Galois cohomology , volume 165

Philippe Gille and Tam \'a s Szamuely. Central simple algebras and Galois cohomology , volume 165. Cambridge University Press, 2017

work page 2017
[8]

Algebraic geometry , volume 52

Robin Hartshorne. Algebraic geometry , volume 52. Springer Science & Business Media, 1977

1977

Show all 18 references

[9]

Splitting Brauer classes using the universal Albanese

Wei Ho and Max Lieblich . Splitting Brauer classes using the universal Albanese . arXiv e-prints , page arXiv:1805.12566, May 2018

2018 arXiv
[10]

Hitching and Christian Pauly

George H. Hitching and Christian Pauly. Theta divisors of stable vector bundles may be nonreduced. Geom. Dedicata , 177:257--273, 2015

2015
[11]

The book of involutions

Max-Albert Knus, AA Merkurjev, Markus Rost, and Jean-Pierre Tignol. The book of involutions. number 44 in american mathematical society colloquium publications. American Mathematical Society, Providence, RI , 1998

1998
[12]

The P icard group of the universal P icard varieties over the moduli space of curves

Alexis Kouvidakis. The P icard group of the universal P icard varieties over the moduli space of curves. J. Differential Geom. , 34(3):839--850, 1991

1991
[13]

Algebra , volume 211 of Graduate Texts in Mathematics

Serge Lang. Algebra , volume 211 of Graduate Texts in Mathematics . Springer-Verlag, New York, third edition, 2002

2002
[14]

James S. Milne. \' E tale cohomology , volume 33 of Princeton Mathematical Series . Princeton University Press, Princeton, N.J., 1980

1980
[15]

Abelian varieties

David Mumford. Abelian varieties . Published for the Tata Institute of Fundamental Research By Hindustan Book Agency International distribution by American Mathematical Society, Mumbai New Delhi, 2008

2008
[16]

The P icard group of the compactified universal J acobian

Margarida Melo and Filippo Viviani. The P icard group of the compactified universal J acobian. Doc. Math. , 19:457--507, 2014

2014
[17]

Sections des fibr\'es vectoriels sur une courbe

Michel Raynaud. Sections des fibr\'es vectoriels sur une courbe. Bull. Soc. Math. France , 110(1):103--125, 1982

1982
[18]

The strong F ranchetta conjecture in arbitrary characteristics

Stefan Schr\"oer. The strong F ranchetta conjecture in arbitrary characteristics. Internat. J. Math. , 14(4):371--396, 2003

2003

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

Open problems on central simple algebras

Asher Auel, Eric Brussel, Skip Garibaldi, and Uzi Vishne. Open problems on central simple algebras. Transform. Groups , 16(1):219--264, 2011

work page 2011

[2] [2]

Arbarello, M

E. Arbarello, M. Cornalba, P. A. Griffiths, and J. Harris. Geometry of algebraic curves. V ol. I , volume 267 of Grundlehren der Mathematischen Wissenschaften [Fundamental Principles of Mathematical Sciences] . Springer-Verlag, New York, 1985

1985

[3] [3]

Unramified division algebras do not always contain A zumaya maximal orders

Benjamin Antieau and Ben Williams. Unramified division algebras do not always contain A zumaya maximal orders. Invent. Math. , 197(1):47--56, 2014

work page 2014

[4] [4]

N\' e ron Models

Siegfried Bosch. N\' e ron Models . Springer Berlin Heidelberg, Berlin, Heidelberg, 1990

work page 1990

[5] [5]

Derived categories of twisted sheaves on C alabi- Y au manifolds

Andrei Horia Caldararu. Derived categories of twisted sheaves on C alabi- Y au manifolds . ProQuest LLC, Ann Arbor, MI, 2000. Thesis (Ph.D.)--Cornell University

work page 2000

[6] [6]

Cohomologie non ab\' e lienne

Jean Giraud. Cohomologie non ab\' e lienne . Springer-Verlag, Berlin-New York, 1971. Die Grundlehren der mathematischen Wissenschaften, Band 179

1971

[7] [7]

Central simple algebras and Galois cohomology , volume 165

Philippe Gille and Tam \'a s Szamuely. Central simple algebras and Galois cohomology , volume 165. Cambridge University Press, 2017

work page 2017

[8] [8]

Algebraic geometry , volume 52

Robin Hartshorne. Algebraic geometry , volume 52. Springer Science & Business Media, 1977

1977