Monodromy results for abelian surfaces and K3 surfaces with bad reduction

Tejasi Bhatnagar

REVIEW 4 major objections 5 minor 19 references

Monodromy results for abelian surfaces and K3 surfaces with bad reduction

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read For ordinary abelian surfaces and K3 surfaces with type II supersingular reduction, the inertia subgroup has finite index in the p-adic monodromy image.

desk verdict The abelian-surface part is a plausible but incomplete extension of Igusa; the K3/Shimura theorem rests on an unproved p-adic bridge between the K3 and Kuga-Satake representations. read the letter →

arxiv 2411.16865 v1 pith:GSFMS2EO submitted 2024-11-25 math.NT math.AG

Tejasi Bhatnagar This is my paper · ORCID

classification math.NTmath.AG MSC 11G1811G2511F8014G2014J28

keywords p-adicmonodromyabeliansurfacesK3badreductionRaynaudextensionKuga-SatakevarietyHeckeorbitfinitenesssupersingular

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 proves a local p-adic monodromy theorem in characteristic p for ordinary abelian surfaces and ordinary K3 surfaces whose reduction is bad. For abelian surfaces with semi-abelian reduction whose abelian quotient is a supersingular elliptic curve, it shows that the inertia subgroup has finite index in the total image of the p-adic Galois representation; the same conclusion is proved for K3 surfaces with type II supersingular reduction. If the theorem is right, these are the higher-dimensional analogues of the classical one-dimensional supersingular monodromy result, and they imply that p-power Hecke orbits reduce to only finitely many isomorphism classes over the residue field. The proof works by describing the p-power torsion through Raynaud uniformization and, on the K3 side, transferring the question to the Kuga-Satake abelian variety.

What carries the argument

Raynaud uniformization carries the argument: a semistable abelian variety over K is a quotient $Z/M$ of a semi-abelian rigid-analytic group $Z$ by a lattice $M$, so p-power torsion points can be chased through the extension $0 \to T \to Z \to B \to 0$. For an abelian surface, $B$ is an elliptic curve, and the Galois action on the two torsion generators is a $2 \times 2$ matrix whose behaviour is governed by whether $B$ is ordinary or supersingular. On the K3 side, the Kuga-Satake construction converts the K3 crystal into a high-dimensional abelian variety, and Corollary 8.3 shows that in type II degeneration its Raynaud extension has the same shape, with abelian quotient isogenous to $d/2$ copies of one elliptic curve. Toroidal compactifications and the variation of mixed Hodge structures on the boundary supply this Raynaud description in characteristic p.

What would settle it

Take an explicit ordinary K3 surface over $\mathbb{F}_q((t))$ with type II supersingular reduction, compute the image of inertia inside $\mathrm{SO}_n(\mathbb{Z}_p)$ from its crystalline Frobenius and monodromy operator, and check that it has finite index in the full Galois image; a counterexample would be an inertia image of infinite index, or a mismatch between the ramification filtration of $\rho_X$ and that predicted by the Kuga-Satake quotient relation.

Watch

Extended reading notes

Core claim

The central claim is Theorem 10.1: for an ordinary K-point X of an orthogonal Shimura variety of signature $(n,2)$, the p-adic monodromy representation $ ho_X$ (and the representation $ ho_{KS}$ of the Kuga-Satake abelian variety) is controlled by the boundary stratum. If X has type II reduction and the reduction is ordinary, inertia acts unipotently; if the boundary point is supersingular, the inertia subgroup has finite index in the full Galois image; if X has type III reduction, the representation is trivial. In the abelian surface case this appears as Theorem 1.2, where semi-abelian reduction with ordinary elliptic quotient is unipotent, supersingular elliptic quotient gives finite index, and total degeneration gives trivial image. The argument reduces the K3 statement to an explicit description of the Raynaud extension of the Kuga-Satake abelian variety as an extension of a torus by a product of copies of one elliptic curve.

Load-bearing premise

The load-bearing premise is that the p-adic Galois representation attached to the K3 surface is a quotient of the representation attached to its Kuga-Satake abelian variety, an assertion stated in the paper without a detailed proof or a specific p-adic comparison.

Editorial extensions

If this is right

For ordinary abelian surfaces over K with semi-abelian reduction, the p-adic monodromy representation is unipotent when the elliptic quotient is ordinary, has finite-index inertia image when the quotient is supersingular, and is trivial when the surface totally degenerates.
For ordinary K3 surfaces with type II reduction, inertia is unipotent in the ordinary boundary case and has finite index in the whole Galois image in the supersingular boundary case; type III reduction gives trivial monodromy.
The p-power Hecke orbit of an ordinary point with type II supersingular reduction reduces to finitely many isomorphism classes over the residue field, as stated in Corollary 10.2.
The same finite-index conclusion holds for the Kuga-Satake abelian variety, so any statement depending only on the Kuga-Satake representation inherits the result.

Reading between the lines

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

If the finite-index statement survives base change to finite extensions of K, it gives an open-image-type theorem for the p-adic monodromy of ordinary points in the supersingular boundary locus, a property that is usually required for further arithmetic applications.
The Raynaud-extension recipe suggests a route to higher dimensions: prove a suitable higher-dimensional version of the one-dimensional supersingular ramification theorem for ordinary abelian varieties, and the same argument would likely go through for abelian varieties of any dimension with semi-abelian reduction.
The quotient-of-representations assumption could be checked by comparing the ramification filtrations of $\rho_X$ and $\rho_{KS}$ on a concrete ordinary K3 family; if the comparison only holds up to a finite extension, the finiteness conclusion should still hold but the index statement would need to be formulated for that extended field.

Share X Bluesky LinkedIn Reddit HN

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

Tejasi has taken on a real problem and has a sensible plan: reduce the bad-reduction monodromy questions to the one-dimensional elliptic curve case via Raynaud uniformization and the Kuga-Satake construction. If the main theorems were right, they would answer Chai's question in dimension two and give the first char-p unit-root monodromy results for type II/III K3s. I want that paper to exist.

What is genuinely useful: the calculation in Section 8 of the weight filtration on the Clifford algebra is clean. Proposition 8.2, which identifies the abelian quotient of the KS Raynaud extension as a product of d/2 copies of an elliptic curve, is a nice observation and the right kind of input. The abelian-surface setup in Section 5 is also natural, and reducing to an elliptic curve quotient is the obvious thing to try.

But the proofs as written do not establish the main claims. Three soft spots, in increasing order of severity.

First, Lemma 5.4, needed for the supersingular case, asserts without proof that Gal(K(z(1),y(n))/K) is non-abelian. That is exactly the fact that rules out unramified subextensions, so it is not a minor omission. The lemma's statement also says "ordinary elliptic curve with supersingular reduction," which is contradictory in the usual terminology, though the intended meaning is clear.

Second, the transfer from the complex boundary to characteristic p in Section 8.1 is asserted. The diagram at the end is not an argument that the Raynaud extension of KS(X) over F_q((t)) is governed by the mod p reduction of the universal mixed Hodge structure. This may be repairable using Madapusi Pera's integral models, but the paper doesn't supply the comparison.

Third, and most seriously, the Kuga-Satake bridge. Section 1.0.3 and Theorem 10.1 state that the Galois representation of the K3 surface is a quotient of the representation attached to the Kuga-Satake abelian variety. The classical construction gives a Hodge-theoretic realization of the K3 cohomology inside the algebra of endomorphisms of H^1(KS), or in the Clifford algebra, not as a direct quotient of H^1(KS). One needs a p-adic comparison theorem to get a Galois-equivariant quotient map from T_p(KS) to the K3 Galois representation, and no construction or citation is given. Without this, Theorem 10.1(1)(b) does not follow.

I should say the paper isn't a bluff. The strategy is coherent and the author has read the relevant literature. But the load-bearing assertions are unproved, and the K3 side as it stands is unsupported.

Bottom line: this deserves a serious referee because the questions are important and the plan is plausible. I would send it to review, but I would expect the referee to demand either a real p-adic KS comparison and a proof of Lemma 5.4, or a major reframing. As it stands, I wouldn't cite the K3 theorems, and I'd be cautious about the abelian-surface case until Lemma 5.4 is fixed.

Referee Report

4 major / 5 minor

Summary. The paper studies p-adic Galois monodromy representations coming from p-power torsion of abelian surfaces and K3 surfaces over the equal-characteristic local field K = F_q((t)), in cases of semi-stable or bad reduction. The main theorems state that for ordinary abelian surfaces with semi-abelian reduction whose abelian quotient is supersingular, the inertia image has finite index in the Galois image (Theorem 1.2), and analogously for ordinary K3 surfaces and orthogonal Shimura varieties with type II supersingular reduction (Theorem 10.1). Type III reduction is asserted to yield trivial Galois image, and type II ordinary reduction to yield unipotent inertia. As an application, the paper claims finiteness of reductions of p-power Hecke orbits (Corollary 10.2). The strategy combines Raynaud uniformization, Igusa's elliptic-curve monodromy theorem, and the Kuga-Satake construction, with boundary mixed Hodge structures used to describe the Raynaud extension of the Kuga-Satake abelian variety.

Significance. If the main theorems are correct, the paper would give a function-field analogue of the KLSS finiteness result and a substantial extension of Igusa's and Chai's local monodromy results to abelian surfaces and K3 surfaces with bad reduction. The proposed strategy is natural: it identifies the right external anchors (Igusa, Raynaud and Bosch-Lutkebohmert, Kisin, Madapusi Pera) and it is not circular, since no fitted parameters or self-referential derivations are involved. The paper is also honest that the abelian-surface case is the base input and that the K3 case is meant to follow from the Kuga-Satake construction. However, as submitted, the central reductions are incomplete at two load-bearing points: the proof of Lemma 5.4 asserts rather than proves the key ramification statement, and the K3/Shimura part relies on an unproved p-adic comparison between the Galois representation of a K3 surface and that of its Kuga-Satake abelian variety. A further characteristic-p transfer from complex mixed Hodge structures is also asserted rather than established.

major comments (4)

[§5.2.2, Lemma 5.4] The proof of total ramification in Lemma 5.4 ends with 'But this is a contradiction as we know that Galois group of K(z(1), y(n))/K is not abelian.' The non-abelianness of this Galois group is precisely the content that needs to be established; it is asserted without argument or citation. Since Proposition 5.3 and Theorem 1.2(1)(b) rely on this lemma, the proof of the abelian-surface theorem is incomplete at a load-bearing point.
[§5.2.1] In the ordinary case B0 ordinary, the text says 'it follows from Lemma 5.4 below that once we attach the z(n) to K(y(n)), the field extension K(z(n),y(n))^sep over K(y(n)) must be totally ramified.' But Lemma 5.4 assumes that B is an 'ordinary elliptic curve over K with supersingular reduction,' so it does not apply when B0 is ordinary. The step that proves Theorem 1.2(1)(a) therefore needs a different argument.
[§1.0.3 and proof of Theorem 10.1] The assertion that 'the Galois representation of the Kuga-Satake abelian variety is a lift of the representation associated to the K3 surface' is stated without proof, and the proof of Theorem 10.1 repeats it as 'the monodromy group ... is a quotient of the monodromy of the Kuga-Satake abelian variety.' The classical Kuga-Satake construction gives a Hodge-theoretic correspondence in characteristic 0; it does not by itself produce a Galois-equivariant p-adic quotient map for an equal-characteristic local field with bad reduction. A construction or a precise p-adic comparison theorem is needed before any K3/Shimura conclusion follows.
[§8.1, Corollary 8.3] Corollary 8.3 transfers the mixed-Hodge-theoretic boundary computations of §8.0.2–8.0.4 (made over C/Q) to the Raynaud extension over K=F_q((t)) by asserting that the diagram of boundary components 'commutes over Z_p (and hence mod p)' and that the universal Raynaud extension extends from C to the canonical integral model. No comparison is provided that justifies this transfer for a point with bad reduction in equal characteristic. Since Corollary 8.3 is the basis for the structure of Z_KS used in §9, this is a load-bearing gap for Theorem 10.1.

minor comments (5)

[Throughout] The paper repeatedly writes 'bases' where 'basis' is meant, and 'Siegal' for 'Siegel'; these typos should be corrected throughout.
[Proposition 1.4] Proposition 1.4 defines d = dim(X), but the corollary and §9 use d for the dimension of KS(X); with d = dim(X) the exponent d/2 in G^{d/2}_m is ambiguous and inconsistent with the later notation.
[§2, proof of Theorem 2.1] The quantities A1(t) and A2(t) are introduced without definition, and the claim that one may assume v_K(A1)=1 'when E is the universal elliptic curve' does not explain why the conclusion transfers to an arbitrary ordinary elliptic curve with supersingular reduction over K; since Igusa's theorem is available as an external result, this should be framed as a sketch.
[§5.2.2, Proposition 5.3] The sequence is written as 'y(1), y(1), y(2), ... and z(1), z(1), z(2), ...'; the duplicate terms should be corrected.
[Corollary 10.2] The phrase 'the reduction of their p-power Hecke orbit of X is finite' is grammatically ambiguous, and 'p-power Hecke orbit' is not defined before use.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the derivation chain uses external theorems and genuine reduction arguments; the main gaps are unproved bridge assertions, not circular ones.

full rationale

The paper's central results are not obtained by fitting parameters, by renaming known outputs, or by self-referential definitions. The abelian-surface theorem (Theorem 1.2) is reduced to Igusa's theorem on elliptic curves with supersingular reduction, Tate/Raynaud uniformization for degenerating abelian varieties, and a direct study of the Galois action on p-power torsion; these are external or newly proven inputs, and the conclusions are not built into those inputs. The K3/Shimura theorem (Theorem 10.1) is intended to follow from the Kuga-Satake abelian-variety computation via the assertion that the K3 monodromy representation is a quotient of the Kuga-Satake representation. That assertion is stated without proof in Sections 1.0.3 and 10, and the transfer from complex mixed Hodge structures on toroidal boundaries to the characteristic-p Raynaud extension is asserted in Section 8.1; these are gaps in justification rather than circular reductions, because the finite-index conclusion for the quotient would not be identical to the finite-index conclusion for the Kuga-Satake variety by definition. There are no fitted parameters renamed as predictions, no load-bearing self-citations by the author, and no uniqueness theorem imported from the author's own prior work to force a choice. The citations to KLSS, MP16, MP19, Kis10, BL91, and Ray71 are external sources; even where a cited result is analogous, the present claim is not asserted to hold merely because of that citation. Accordingly, the paper is not circular, though its K3-side argument is incomplete as written.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

All foundational inputs are prior published results; the paper's new content is the combination and application. No free parameters or new entities are introduced. The central fragility is not circularity but the unproved bridging assumptions, in particular the Kuga-Satake monodromy lift and the characteristic p transfer.

assumptions (6)

standard math Igusa's theorem on monodromy of an elliptic curve with supersingular reduction (Theorem 2.1)
Used as the base case in the proofs of Proposition 5.3 and Proposition 9.3 for the y(n) torsion points.
domain assumption Raynaud's uniformization theorem for abelian varieties with semi-stable reduction (Theorem 4.1, Ray71, BL91)
Gives the decomposition A = Z/M and the rigid analytic extension 0 to T to Z to B to 0 used throughout Sections 5, 8, and 9.
domain assumption Existence of canonical integral models and toroidal compactifications for GSpin Shimura varieties (Kis10, MP16, MP19)
Used to transfer Kuga-Satake and boundary Raynaud data to characteristic p in Section 8.1.
domain assumption Kuga-Satake construction and its integral extension (Section 6.3, MP16)
Associates an abelian scheme to the GSpin Shimura variety; the p-adic monodromy lift property is asserted rather than established.
domain assumption Ogus's theorem that Kuga-Satake abelian varieties of ordinary K3 surfaces are ordinary (Lemma 6.1, Ogu84 Theorem 7.8)
Used to conclude that the elliptic curve E in the Raynaud extension of KS(X) is ordinary when X is ordinary.
domain assumption Mixed Hodge structure on the toroidal boundary describes the universal Raynaud extension (Section 7.1.1, MP19)
Underlies the computation of the Raynaud extension of the Kuga-Satake abelian variety in Section 8.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Monodromy results for abelian surfaces and K3 surfaces with bad reduction." pith.science (2026). https://pith.science/paper/GSFMS2EO

@misc{pith2026241116865,
  author       = {Pith},
  title        = {Pith review of: Monodromy results for abelian surfaces and K3 surfaces with bad reduction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GSFMS2EO}},
  note         = {Machine review of arXiv:2411.16865}
}

read the original abstract

The purpose of this paper is to prove a local p-adic monodromy theorem for ordinary abelian surfaces and K3 surfaces with bad reduction in characteristic p. As an application, we get a finiteness result for the reduction of their Hecke orbits in the case of type II supersingular reduction.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

19 extracted references · 18 canonical work pages

[1]

4, 653--698

Siegfried Bosch and Werner L\" u tkebohmert, Degenerating abelian varieties, Topology 30 (1991), no. 4, 653--698. 1133878

work page 1991
[2]

Jan Hendrik Bruinier and Shaul Zemel, Special cycles on toroidal compactifications of orthogonal S himura varieties , Math. Ann. 384 (2022), no. 1-2, 309--371. 4476226

work page 2022
[3]

Reine Angew

Ching-Li Chai, Local monodromy for deformations of one-dimensional formal groups, J. Reine Angew. Math. 524 (2000), 227--238. 1770609

work page 2000
[4]

, Methods for p -adic monodromy , J. Inst. Math. Jussieu 7 (2008), no. 2, 247--268. 2400722

work page 2008
[5]

Ana Caraiani and Peter Scholze, On the generic part of the cohomology of non-compact unitary S himura varieties , Ann. of Math. (2) 199 (2024), no. 2, 483--590. 4713019

work page 2024
[6]

Math., vol

Torsten Ekedahl, The action of monodromy on torsion points of J acobians , Arithmetic algebraic geometry ( T exel, 1989), Progr. Math., vol. 89, Birkh\"auser Boston, Boston, MA, 1991, pp. 41--49. 1085255

work page 1989
[7]

Gross, Ramification in p -adic L ie extensions , Journ\' e es de G \' e om\' e trie A lg\' e brique de R ennes ( R ennes, 1978), V ol

Benedict H. Gross, Ramification in p -adic L ie extensions , Journ\' e es de G \' e om\' e trie A lg\' e brique de R ennes ( R ennes, 1978), V ol. III , Ast\' e risque, vol. 65, Soc. Math. France, Paris, 1979, pp. 81--102. 563473

work page 1978
[8]

Jun-ichi Igusa, On the algebraic theory of elliptic modular functions, J. Math. Soc. Japan 20 (1968), 96--106. 240103

work page 1968

Show all 19 references

[9]

Katz, p -adic properties of modular schemes and modular forms , Modular functions of one variable, III ( P roc

Nicholas M. Katz, p -adic properties of modular schemes and modular forms , Modular functions of one variable, III ( P roc. I nternat. S ummer S chool, U niv. A ntwerp, A ntwerp, 1972), Lecture Notes in Math., vol. Vol. 350, Springer, Berlin-New York, 1973, pp. 69--190. 447119

1972
[10]

Mark Kisin, Integral models for S himura varieties of abelian type , J. Amer. Math. Soc. 23 (2010), no. 4, 967--1012. 2669706

2010
[11]

Shankar, and Padmavathi Srinivasan, Finiteness of reductions of hecke orbits, Journal of the European Mathematical Society

Mark Kisin, Yeuk Hay Joshua Lam, Ananth N. Shankar, and Padmavathi Srinivasan, Finiteness of reductions of hecke orbits, Journal of the European Mathematical Society
[12]

Sigma 4 (2016), Paper No

Wansu Kim and Keerthi Madapusi Pera, 2-adic integral canonical models, Forum Math. Sigma 4 (2016), Paper No. e28, 34. 3569319

2016
[13]

Keerthi Madapusi Pera, Integral canonical models for spin S himura varieties , Compos. Math. 152 (2016), no. 4, 769--824. 3484114

2016
[14]

, Toroidal compactifications of integral models of S himura varieties of H odge type , Ann. Sci. \' E c. Norm. Sup\' e r. (4) 52 (2019), no. 2, 393--514. 3948111

2019
[15]

Arthur Ogus, F -isocrystals and de R ham cohomology. II . C onvergent isocrystals , Duke Math. J. 51 (1984), no. 4, 765--850. 771383

1984
[16]

a t Bonn, Mathematisches Institut, Bonn, 1990, Dissertation, Rheinische Friedrich-Wilhelms-Universit\

Richard Pink, Arithmetical compactification of mixed S himura varieties , Bonner Mathematische Schriften [Bonn Mathematical Publications], vol. 209, Universit\" a t Bonn, Mathematisches Institut, Bonn, 1990, Dissertation, Rheinische Friedrich-Wilhelms-Universit\" a t Bonn, Bon...

1990
[17]

473--477

Michel Raynaud, Vari\' e t\' e s ab\' e liennes et g\' e om\' e trie rigide , Actes du C ongr\`es I nternational des M ath\' e maticiens ( N ice, 1970), T ome 1, Gauthier-Villars \' E diteur, Paris, 1971, pp. 473--477. 427326

1970
[18]

Stefan Schreieder and Andrey Soldatenkov, The K uga- S atake construction under degeneration , J. Inst. Math. Jussieu 19 (2020), no. 6, 2165--2182. 4167005

2020
[19]

Salim Tayou, Picard rank jumps for k3 surfaces with bad reduction, To appear in Algebra and Number theory, arXiv:2203.09559 (2022)

2022 arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [1]

4, 653--698

Siegfried Bosch and Werner L\" u tkebohmert, Degenerating abelian varieties, Topology 30 (1991), no. 4, 653--698. 1133878

work page 1991

[2] [2]

Jan Hendrik Bruinier and Shaul Zemel, Special cycles on toroidal compactifications of orthogonal S himura varieties , Math. Ann. 384 (2022), no. 1-2, 309--371. 4476226

work page 2022

[3] [3]

Reine Angew

Ching-Li Chai, Local monodromy for deformations of one-dimensional formal groups, J. Reine Angew. Math. 524 (2000), 227--238. 1770609

work page 2000

[4] [4]

, Methods for p -adic monodromy , J. Inst. Math. Jussieu 7 (2008), no. 2, 247--268. 2400722

work page 2008

[5] [5]

Ana Caraiani and Peter Scholze, On the generic part of the cohomology of non-compact unitary S himura varieties , Ann. of Math. (2) 199 (2024), no. 2, 483--590. 4713019

work page 2024

[6] [6]

Math., vol

Torsten Ekedahl, The action of monodromy on torsion points of J acobians , Arithmetic algebraic geometry ( T exel, 1989), Progr. Math., vol. 89, Birkh\"auser Boston, Boston, MA, 1991, pp. 41--49. 1085255

work page 1989

[7] [7]

Gross, Ramification in p -adic L ie extensions , Journ\' e es de G \' e om\' e trie A lg\' e brique de R ennes ( R ennes, 1978), V ol

Benedict H. Gross, Ramification in p -adic L ie extensions , Journ\' e es de G \' e om\' e trie A lg\' e brique de R ennes ( R ennes, 1978), V ol. III , Ast\' e risque, vol. 65, Soc. Math. France, Paris, 1979, pp. 81--102. 563473

work page 1978

[8] [8]

Jun-ichi Igusa, On the algebraic theory of elliptic modular functions, J. Math. Soc. Japan 20 (1968), 96--106. 240103

work page 1968