Torsors over moduli spaces of vector bundles over curves of fixed determinant

Indranil Biswas; Jacques Hurtubise

REVIEW 2 major objections 4 minor 19 references

Torsors over moduli spaces of vector bundles over curves of fixed determinant

T0 review · 2 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read The paper proves that the torsor of holomorphic projective connections on stable vector bundles of fixed rank and determinant over a compact Riemann surface is, up to a unique scaling constant, the same algebraic torsor as the torsor of…

desk verdict A plausible extension of the theta-characteristic torsor theorem, but the proof relies on an unstated coprimality hypothesis and a section-construction typo that a referee should catch. read the letter →

arxiv 2507.05690 v1 pith:HK25KK3W submitted 2025-07-08 math.AG math.DG

Indranil Biswas , Jacques Hurtubise This is my paper · ORCID

classification math.AGmath.DG MSC 14H6014D21

keywords stablebundlesprojectiveconnectionsmodulispacethetalinebundletorsorsQuillenmetricRiemannsurfacefixeddeterminant

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

This paper establishes that two torsors attached to the moduli space $M$ of stable vector bundles of fixed rank and determinant on a compact Riemann surface are the same algebraic object, up to a single scaling constant. One torsor, $N_C$, has as its fiber over a bundle $E$ the affine space of holomorphic projective connections on the projective bundle $P(E)$. The other, $\mathrm{Conn}(L)$, has as its fiber the affine space of holomorphic connections on a naturally defined line bundle $L$ over $M$, the determinant ($\theta$) line bundle built from $H^0$ and $H^1$ of $E\otimes F$ for a fixed semistable auxiliary bundle $F$. The main theorem constructs a unique algebraic isomorphism $I:\mathrm{Conn}(L)\to N_C$ that is affine-linear on each fiber with a single real number $\lambda_0$. The point of the identification is that it turns curve-level data, a projective connection on each stable bundle, into moduli-space data, a connection on the $\theta$ line bundle, and back.

What carries the argument

The machinery has two parts. First, the two torsors: $N_C$, whose fiber over $E$ is the affine space of holomorphic connections on the projective bundle $P(E)$, modeled on $H^0(X,\mathrm{ad}(E)\otimes K_X)=T^*_E M$, and $\mathrm{Conn}(L)$, defined as the inverse image of the constant function $1$ under the symbol map in the dual of the Atiyah exact sequence for $L$, the sheaf-theoretic sequence whose splittings are connections. Second, the classification mechanism: algebraic $T^*M$-torsors are classified by $H^1(M,T^*M)$, and under the isomorphism $H^1(M,T^*M)\cong H^2(M,\mathbb C)\cong \mathbb C$ each torsor is represented by the $\bar\partial$-derivative of a smooth section. The two smooth sections — the Quillen connection on $L$ and the Narasimhan–Seshadri connection on $P(E)$ — both have $\bar\partial$-derivatives proportional to the Atiyah–Bott Kähler form, so the two torsors sit in the same one-dimensional class; $H^0(M,T^*M)=0$ makes the isomorphism unique. The canonical algebraic section $\gamma$ over the open set $U$ comes from a canonical projective connection on $E\otimes F$, itself built from a section of a sheaf on the doubled diagonal $2\Delta\subset X\times X$.

What would settle it

Compute, for a concrete case such as genus $2$ and rank $3$ with a fixed determinant, the ratio of the $(1,1)$-forms $\partial\beta$ and $\omega_M$ at two distinct stable bundles $E$ in $M$; if the ratio is not constant, equation (4.16) fails and no single $\lambda_0$ can make the affine-linear isomorphism of Theorem 4.1 work.

Watch

Extended reading notes

Core claim

The central discovery is Theorem 4.1: over the moduli space $M$ of stable rank-$r$ vector bundles with fixed determinant $\xi$, the $T^*M$-torsor $\mathrm{Conn}(L)$ of holomorphic connections on the $\theta$ line bundle $L$ is algebraically isomorphic, as a fiber bundle over $M$, to the $T^*M$-torsor $N_C$ of holomorphic projective connections on the bundles $P(E)$. The isomorphism is unique once the scale is fixed: for every $E\in M$, every $z\in \mathrm{Conn}(L)_E$, and every cotangent vector $w\in T^*_E M$, it satisfies $I(z+w)=I(z)+\lambda_0 w$ with a unique real number $\lambda_0$. The proof classifies algebraic $T^*M$-torsors by $H^1(M,T^*M)$, identifies that group with $H^2(M,\mathbb C)\cong \mathbb C$, and shows that both torsors are represented by positive real multiples of the same Kähler form: the Quillen connection on $L$ has $c_1(\nabla_{L,Q})=\lambda\,\omega_M$, and the $\bar\partial$-derivative of the Narasimhan–Seshadri section of $N_C$ is $\nu\,\omega_M$. Uniqueness of the isomorphism follows from $H^0(M,T^*M)=0$, which the authors attribute to the unirationality of $M$.

Load-bearing premise

The proof assumes the moduli space is a complete projective variety whose two cohomology groups that classify these torsors are one-dimensional complex vector spaces; for the open locus of stable bundles alone, which may be affine, that assumption can fail and the identification is not established.

Editorial extensions

If this is right

The holomorphic projective connections on the stable bundles in $M$ are, up to the fixed scaling $\lambda_0$, in bijection with holomorphic connections on the theta line bundle $L$ over $M$.
The algebraic section $\gamma$ constructed on the open set $U$ (where $H^0(X,E\otimes F)=H^1(X,E\otimes F)=0$) is compatible with the global smooth Narasimhan–Seshadri section: after the isomorphism, $I\circ \nabla_{L,Q}=\beta$ on all of $M$.
Since $\mathrm{Conn}(L)$ and $N_C$ are isomorphic as $T^*M$-torsors, their spaces of algebraic sections over any open subset of $M$ agree.
Both torsors are represented by the same class in $H^1(M,T^*M)\cong \mathbb C$, so the ratio of the two proportionality constants $\lambda$ and $\nu$ determines $\lambda_0$.

Reading between the lines

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

A natural extension, not pursued in the paper, is to run the same two-forms-proportionality argument for moduli of principal $G$-bundles; the theorem would then give a general transfer principle from bundle-level projective connections to moduli-space connections.
Because the theta line bundle $L$ has a section vanishing exactly on $M\setminus U$, the isomorphism suggests that the canonical projective connection on $E$ degenerates in a controlled way as $E$ approaches the theta divisor; the paper constructs the canonical connection only on $U$.
The theorem proves existence and uniqueness of $\lambda_0$ but does not compute it; evaluating the ratio of the proportionality constants in (4.12) and (4.16) for a small genus would give its numerical value.

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

The paper's real new content is Proposition 3.1: for a fixed semistable F with χ(E⊗F)=0, there is a canonical projective connection on each stable E in the open set where H^0(E⊗F)=H^1(E⊗F)=0. The Lie algebra projection J that descends the connection on E⊗F to one on P(E) is a clean idea, and it genuinely extends the earlier theta-characteristic results of the same authors. Theorem 4.1 then identifies the torsor of such connections with the torsor of holomorphic connections on the determinant line bundle L, up to a single real scale factor. That is a nice statement.

The proofs are mostly self-contained and use standard machinery (Quillen metrics, Narasimhan-Seshadri, Atiyah-Bott). The torsor classification step is plausible.

The soft spots are real. First, the paper never states the coprime assumption gcd(r, deg ξ)=1. Without it, the moduli space M of stable bundles is not projective; it is a Zariski open subset of the projective moduli of semistable bundles. The proof of Theorem 4.1 uses H^1(M,T*M)=H^2(M,C)=C (equations (4.14)-(4.15)) to classify T*M-torsors and to make λ0 unique. For a non-projective open M, H^1(M,T*M) may well be zero, in which case both torsors are trivial and every λ0 works. This is not a minor quibble; the uniqueness of λ0 is part of the theorem. Adding gcd(r, deg ξ)=1 fixes it, and that is a standard hypothesis in this context, so the theorem likely holds in the coprime case.

Second, Section 4.1 constructs a section Γ of the line bundle L by using an exact sequence involving E, not E⊗F, even though L was defined via the determinant of cohomology of E⊗F. The claimed divisor of Γ (M\U) is therefore not justified as written. This seems to be a typo in the text—the construction would work if E were replaced by E⊗F throughout—but a referee will want it corrected.

The citation pattern is fine: the paper builds on the authors' earlier work, but the earlier results are cited as context, not used as a black box, and the new construction stands on its own.

Bottom line: the central idea is sound and the result is likely correct under standard coprimality assumptions, but the paper as written is not rigorous in its stated generality. It deserves a serious referee; a good referee would ask for a precise statement of the hypotheses and a fix to the Section 4.1 typo, after which it should be publishable.

Referee Report

2 major / 4 minor

Summary. The paper studies two algebraic T*M-torsors associated to the moduli space M of stable vector bundles of fixed rank and determinant on a compact Riemann surface. Fixing a semistable bundle F with identically zero Euler characteristic against the universal family, the authors first show (Propositions 2.1 and 3.1) that every E in a Zariski-open subset U has a canonical holomorphic projective connection, obtained from the canonical connection on P(E⊗F) by reduction of structure group. They then define the theta-type determinant line bundle L on M and consider the torsor Conn(L) of holomorphic connections on L. The main result, Theorem 4.1, asserts a unique real number λ0 and a unique algebraic isomorphism I: Conn(L)→N_C of torsors over M such that I(z+w)=I(z)+λ0 w for all cotangent vectors w. The proof compares the Quillen connection on L with the Narasimhan-Seshadri section of N_C, using the identifications H^0(M,T*M)=0 and H^1(M,T*M)=H^2(M,C)=C.

Significance. If established in the stated generality, the theorem gives a striking and natural identification between projective connections on the underlying curve and connections on the theta line bundle over moduli space, extending earlier theta-characteristic results to arbitrary degree and rank. The paper has genuine strengths: Proposition 2.1 is a self-contained Cauchy-kernel/adjunction construction, the torsor formalism is appropriate, and the proof relies on substantial but well-matched external results (Quillen, Narasimhan-Seshadri, Zograf-Takhtadzhyan). The final theorem is falsifiable in the sense that its uniqueness and scaling claims depend on concrete cohomological facts that can be checked. However, as written, the main theorem is not proved for the stated class of moduli spaces, and one auxiliary construction in Section 4.1 is incorrect, so the work needs substantial revision before it is ready for publication.

major comments (2)

[§3.1 and Theorem 4.1, Eqs. (4.14)–(4.15)] The proof of Theorem 4.1 relies on H^0(M,T*M)=0 and H^1(M,T*M)=H^2(M,C)=C, but these identifications are not justified for M as defined in (3.1). Without the coprime condition gcd(r,deg ξ)=1, the stable locus is a proper quasiprojective open subset of the projective moduli space of semistable bundles. For an affine such M one has H^1(M,T*M)=0, in which case both Conn(L) and N_C are algebraically trivial and any nonzero λ0 yields an isomorphism satisfying the affine formula, so the asserted uniqueness of λ0 fails. The authors should either add gcd(r,deg ξ)=1 and justify (4.14)–(4.15) using [DN] together with unirationality and Hodge theory, or prove the theorem for the projective moduli of semistable bundles and cope with its singularities.
[§4.1, Eqs. (4.4)–(4.7)] The construction of the section Γ uses the sheaves E⊗η*O_X(D0), whereas the line bundle L in (4.1) is defined as the determinant bundle of E⊗η*F. For a general semistable F these are unrelated. Moreover the kernel of R in (4.5) is ψ_*E, so R need not be an isomorphism over U, since H^0(E) does not have to vanish for E∈U. Thus Γ is not a section of L, and the claim that its divisor is exactly M\U is not established. This subsection is not used in the proof of Theorem 4.1, so the main argument can be repaired, but the present assertion should be corrected or removed.

minor comments (4)

[§4.1, after Eq. (4.1)] In the patching calculation, the exponent χ(E⊗E) should read χ(E⊗F); the text even invokes (3.4), which concerns E⊗F.
[§3.2, Proposition 3.1] The notation is overloaded: E and F denote both vector bundles on X and the associated PGL principal bundles, and this makes the reduction-of-structure argument harder to follow; distinct symbols would clarify it.
[Theorem 4.1, Eq. (4.16)] The expression ∂β for a C∞ section of the torsor N_C is used without a definition; a short explanation of how ∂ acts on sections of this torsor would make the cohomological comparison in (4.15) transparent.
[Introduction] The phrase 'shows that stability is not required' is ambiguous: Proposition 1.1 does not require stability, but Proposition 3.1 is stated for E in the stable moduli space; the scope of the claim should be spelled out.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: Theorem 4.1 is obtained by comparing two independent torsor classes via external curvature constants; self-citations are not load-bearing.

full rationale

The paper's central derivation is self-contained against external results (Quillen, Narasimhan-Seshadri, Zograf-Takhtadzhyan, Ivanov, Drezet-Narasimhan). Theorem 4.1 identifies the two T*M-torsors by comparing their classes in H^1(M,T*M)=C: Conn(L) has class lambda*omega_M from Quillen's curvature formula (4.12), and N_C has class nu*omega_M from (4.16), with the slope lambda0 determined as the ratio of these independent geometric constants. No constant is fitted to force a conclusion; lambda0 emerges from the comparison. The self-citations ([BH], [BB], [AzBi]) are contextual or corroborating only: [BH] and [BB] are cited as prior theta-characteristic results and as a heuristic in the closing paragraph, while [AzBi] is cited alongside Weil and Atiyah for existence of a connection on P(E). None is the sole justification of a load-bearing premise. The construction in Proposition 3.1 genuinely reduces a connection on E tensor F to one on P(E), and is not equivalent to the target torsor isomorphism. The main caveat in the text is a correctness gap rather than circularity: the proof of Theorem 4.1 uses (4.14)-(4.15), namely 'H^0(M, T*M) = 0, because M is unirational' and 'H^1(M, T*M) = H^2(M, C) = C', while M is introduced in (3.1) as the moduli space of stable bundles 'with rank(E) = r and V^r E = xi' without any coprime condition and is called 'smooth complex quasiprojective'. Without projectivity, the identification H^1(M,T*M)=H^2(M,C)=C is not automatic, and if H^1(M,T*M)=0 the uniqueness of lambda0 can fail. This affects the stated generality, but it is not a self-referential reduction. Accordingly, no circular step can be exhibited, and the circularity score is low.

Assumptions & free parameters 1 free parameters · 5 assumptions · 0 invented entities

The central claim rests on standard theorems (Serre duality, Kunneth, Faltings, Narasimhan-Seshadri, Quillen) plus an unflagged projectivity assumption on M and the scalar lambda0. No invented entities are introduced.

free parameters (1)

lambda0 = nu/lambda (not computed)
The scalar relating the two torsors in Theorem 4.1. Its existence is asserted but its value is left as a ratio of two undetermined positive constants from equations (4.12) and (4.16).

assumptions (5)

domain assumption M is a projective unirational variety with Pic(M)=Z and H^1(M,T^*M)=H^2(M,C)=C
Used in the proof of Theorem 4.1 (equations (4.14)-(4.15)) to classify T^*M-torsors and show uniqueness. The paper does not state the coprime rank/degree condition needed for the stable moduli to be projective.
standard math Faltings' theorem: every semistable bundle V has a W with H^0(V tensor W)=H^1(V tensor W)=0
Cited in Section 3.2 to guarantee existence of F and to show F is semistable.
domain assumption Narasimhan-Seshadri theorem, Quillen's determinant construction, and the identity c1(nabla_{L,Q}) = lambda * omega_M
External theorems cited for the C-infinity sections beta and nabla_{L,Q} and their curvature properties, used in Sections 3.1 and 4.2.
standard math Poincare adjunction formula, Serre duality, Kunneth formula
Used in the proof of Proposition 2.1 to construct the canonical projective connection.
domain assumption Equation (4.16): partial beta = nu * omega_M
Cited to [ZT] and [Iv]; used to show that the torsor class of N_C is a positive real number.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Torsors over moduli spaces of vector bundles over curves of fixed determinant." pith.science (2026). https://pith.science/paper/HK25KK3W

@misc{pith2026250705690,
  author       = {Pith},
  title        = {Pith review of: Torsors over moduli spaces of vector bundles over curves of fixed determinant},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HK25KK3W}},
  note         = {Machine review of arXiv:2507.05690}
}

abstract

Let ${\mathcal M}$ be a moduli space of stable vector bundles of rank $r$ and determinant $\xi$ on a compact Riemann surface $X$. Fix a semistable holomorphic vector bundle $F$ on $X$ such that $\chi(E\otimes F)= 0$ for $E \in \mathcal M$. Then any $E\in \mathcal M$ with $H^0(X, E\otimes F) = 0 = H^1(X, E\otimes F)$ has a natural holomorphic projective connection. The moduli space of pairs $(E,\, \nabla)$, where $E\, \in\, \mathcal M$ and $\nabla$ is a holomorphic projective connection on $E$, is an algebraic $T^*{\mathcal M}$--torsor on $\mathcal M$. We identify this $T^*{\mathcal M}$--torsor on $\mathcal M$ with the $T^*{\mathcal M}$--torsor given by the sheaf of connections on an ample line bundle over $\mathcal M$.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

19 extracted references · 16 canonical work pages

[1]

M. F. Atiyah, Complex analytic connections in fibre bundles, Trans. Amer. Math. Soc. 85 (1957), 181--207

1957
[2]

M. F. Atiyah and R. Bott, The Yang-Mills equations over Riemann surfaces, Philos. Trans. Roy. Soc. London Ser. A 308 (1983), 523--615

work page 1983
[3]

Azad and I

H. Azad and I. Biswas, On holomorphic principal bundles over a compact Riemann surface admitting a flat connection, Math. Ann. 322 (2002), 333--346

work page 2002
[4]

Beauville, M

A. Beauville, M. S. Narasimhan and S. Ramanan, Spectral curves and the generalised theta divisor, Jour. Reine Angew. Math. 398 (1989), 169--179

work page 1989
[5]

Ben-Zvi and I

D. Ben-Zvi and I. Biswas, Theta functions and Szeg o kernels, Int. Math. Res. Not. (2003), no. 24, 1305--1340

work page 2003
[6]

Bismut, H

J.-M. Bismut, H. Gillet and C. Soul\'e, Analytic torsion and holomorphic determinant bundles. I. Bott-Chern forms and analytic torsion, Comm. Math. Phys. 115 (1988), 49--78

work page 1988
[7]

Bismut, H

J.-M. Bismut, H. Gillet and C. Soul\'e, Analytic torsion and holomorphic determinant bundles. II. Direct images and Bott-Chern forms, Comm. Math. Phys. 115 (1988), 79--126

work page 1988
[8]

Bismut, H

J.-M. Bismut, H. Gillet and C. Soul\'e, Analytic torsion and holomorphic determinant bundles. III. Quillen metrics on holomorphic determinants, Comm. Math. Phys. 115 (1988), 301-351

work page 1988

Show all 19 references

[9]

Biswas and J

I. Biswas and J. Hurtubise, A canonical connection on bundles on Riemann surfaces and Quillen connection on the theta bundle, Adv. Math. 389 (2021), Paper No. 107918

2021
[10]

Deligne, \'Equations diff\'erentielles \`a points singuliers r\'eguliers , Lecture Notes in Mathematics, Vol

P. Deligne, \'Equations diff\'erentielles \`a points singuliers r\'eguliers , Lecture Notes in Mathematics, Vol. 163, Springer-Verlag, Berlin-New York, 1970

1970
[11]

Derryberry, Lax formulation for harmonic maps to a moduli of bundles, arXiv:2106.09781

R. Derryberry, Lax formulation for harmonic maps to a moduli of bundles, arXiv:2106.09781

arXiv
[12]

Dr\'ezet, and M

J.-M. Dr\'ezet, and M. S. Narasimhan, Groupe de Picard des vari\' e t\' e s de modules de fibr\' e s semi-stables sur les courbes alg\' e briques, Invent. Math. 97 (1989), 53--94

1989
[13]

Faltings, Stable G -bundles and projective connections, Jour

G. Faltings, Stable G -bundles and projective connections, Jour. Alg. Geom. 2 (1993) 507--568

1993
[14]

N. V. Ivanov, Projective structures, flat bundles and K\"ahler metrics on moduli spaces, Math. USSR-Sb. 61 (1988), 211--224

1988
[15]

Kobayashi, Differential geometry of complex vector bundles, Princeton University Press, Princeton, NJ, Iwanami Shoten, Tokyo, 1987

S. Kobayashi, Differential geometry of complex vector bundles, Princeton University Press, Princeton, NJ, Iwanami Shoten, Tokyo, 1987

1987
[16]

M. S. Narasimhan and C. S. Seshadri, Stable and unitary vector bundles on a compact R iemann surface, Ann. of Math. 82 (1965), 540--567

1965
[17]

D. G. Quillen, Determinants of Cauchy-Riemann operators on Riemann surfaces, Funct. Anal. Appl. 19 (1985), 37--41

1985
[18]

Weil, G\'en\'eralisation des fonctions ab\'eliennes, Jour

A. Weil, G\'en\'eralisation des fonctions ab\'eliennes, Jour. Math. Pures Appl. 17 (1938), 47--87

1938
[19]

P. G. Zograf and L. A. Takhtadzhyan, On the uniformization of Riemann surfaces and on the Weil-Petersson metric on the Teichm\"uller and Schottky spaces, Math. USSR-Sb. 60 (1988), 297--313

1988

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.

[1] [1]

M. F. Atiyah, Complex analytic connections in fibre bundles, Trans. Amer. Math. Soc. 85 (1957), 181--207

1957

[2] [2]

M. F. Atiyah and R. Bott, The Yang-Mills equations over Riemann surfaces, Philos. Trans. Roy. Soc. London Ser. A 308 (1983), 523--615

work page 1983

[3] [3]

Azad and I

H. Azad and I. Biswas, On holomorphic principal bundles over a compact Riemann surface admitting a flat connection, Math. Ann. 322 (2002), 333--346

work page 2002

[4] [4]

Beauville, M

A. Beauville, M. S. Narasimhan and S. Ramanan, Spectral curves and the generalised theta divisor, Jour. Reine Angew. Math. 398 (1989), 169--179

work page 1989

[5] [5]

Ben-Zvi and I

D. Ben-Zvi and I. Biswas, Theta functions and Szeg o kernels, Int. Math. Res. Not. (2003), no. 24, 1305--1340

work page 2003

[6] [6]

Bismut, H

J.-M. Bismut, H. Gillet and C. Soul\'e, Analytic torsion and holomorphic determinant bundles. I. Bott-Chern forms and analytic torsion, Comm. Math. Phys. 115 (1988), 49--78

work page 1988

[7] [7]

Bismut, H

J.-M. Bismut, H. Gillet and C. Soul\'e, Analytic torsion and holomorphic determinant bundles. II. Direct images and Bott-Chern forms, Comm. Math. Phys. 115 (1988), 79--126

work page 1988

[8] [8]

Bismut, H

J.-M. Bismut, H. Gillet and C. Soul\'e, Analytic torsion and holomorphic determinant bundles. III. Quillen metrics on holomorphic determinants, Comm. Math. Phys. 115 (1988), 301-351

work page 1988