Special Loci in the Moduli of Curves

arXiv:2610.08382 · math.AG, math.NT · Submitted 2026-10-06 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Mathematics. The day in mathematics: new theorems, new proofs, and what they open.

Ada: We're looking at the paper: "Special Loci in the Moduli of Curves".

Emmy: Here's the gist of the paper. The paper introduces new techniques for studying mixed-characteristic special loci within moduli spaces, with a particular focus on the moduli space M g of curves.

Ada: These methods yield several applications, including Zariski density results for split Jacobians, non-density results for supersingular loci, and lifting results for algebraic cycles under appropriate numerical conditions.

Emmy: The techniques generalize to more-or-less arbitrary moduli spaces mapping to Hodge-type Shimura varieties (and some settings beyond the Shimura case), relating to questions raised by Oort and de Jong.

Ada: Key theorems establish density results in various contexts: Theorem 1.2 asserts that for a partition of genus g prime-to- p of splitting type g with a specific codimension condition delta pgq at most 3g - 3, there exists an r such that for every prime p > r, the locus SP g,p: t M F p ts J Jac q is Zariski dense in M F p.

Paper summary: Ada: So wrapping up on "Special Loci in the Moduli of Curves", we've seen how they use mixed characteristic tools to establish density results for certain loci and non-density results for others. The authors are focusing on getting a handle on the structure of special places within moduli spaces, particularly through symmetry.

Emmy: They tackle these problems by generalizing methods that work in positive characteristic, and they manage to apply these ideas across more general spaces that map to Hodge-type Shimura varieties too. It seems like this work connects deep arithmetic questions with geometric structure in a way we hadn't explored before.

Kolya: What does this mean practically? It means we have better tools for understanding the distribution of points on these spaces, and when things deviate from what's expected, we can pinpoint exactly where those deviations are happening.

Ada: Exactly. For someone listening, the implication is that in these moduli spaces, you can use number theory to tell you where the most interesting geometric features are located without having to check every single point individually.

Emmy: It’s about building a more detailed map of these spaces, showing us exactly how they are structured, especially around those special symmetries that Oort and de Jong were interested in.

Conclusion: Ada: So we've been looking at these new techniques for studying special loci in moduli spaces of curves, and now we get to talk about what this specific paper, "Special Loci in the Moduli of Curves," is actually trying to accomplish.

Emmy: It’s about taking these complicated shapes and finding the places that have a special symmetry or structure, like points that are highly constrained by number theory.

Kolya: The main thing they show is how to prove things about where these special points are located—specifically, whether they cover the whole space or just some small pieces of it.

Ada: They use mixed characteristic methods to establish some density results for splitting types, meaning they can find lots of special curves in a certain family.

Emmy: And they also show non-density results for supersingular loci, which means even there are places where these special structures don't fill up the whole space.

Kolya: So what this means is we get clearer boundaries on how dense these interesting geometric features can be in these moduli spaces.

Ada: It’s about understanding the distribution of arithmetic information across these complex geometric landscapes, and it gives us some concrete tools for that analysis.

math.AG, math.NT

Submitted: 2026-10-06

Updated: 2026-11-10

Comments: 59 pages, minor revision, comments welcome!

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 86/100

The gist: The paper introduces new techniques for studying mixed-characteristic special loci within moduli spaces, with a particular focus on the moduli space M g of curves.

Key concepts

Zariski Density
This means that a subset of a space is so large that it covers the entire space when viewed through algebraic geometry. In this context, it shows that certain geometric loci are not just isolated points but form a dense structure within the larger moduli space.
Supersingular Loci
These are specific subsets within moduli spaces where certain arithmetic conditions related to prime characteristics hold, often linked to supersingular objects. The paper investigates whether these sets are dense or sparse, finding that they are generally not dense in relevant contexts.
Mixed Characteristic Techniques
These are advanced mathematical methods used to study geometric objects when the underlying field has a mix of characteristic zero (like complex numbers) and positive characteristic (like finite fields). These techniques allow the authors to prove density results by analyzing boundary strata and expected dimension inequalities.
Special Loci in Moduli Spaces
These are specific subsets within moduli spaces, like $\mathcal{M}_g$, that possess particular arithmetic or geometric properties. The paper focuses on characterizing these loci—determining if they are dense, non-dense, or contained within smaller subvarieties.

Terminology

Summary

The paper introduces new techniques for studying mixed-characteristic special loci within moduli spaces, with a particular focus on the moduli space M g of curves. p. 0

Where the paper says it

Special Loci in the Moduli of Curves David Urbanik and Ziquan Yang October 7, 2026 Abstract We introduce new techniques to study mixed-characteristic special loci in moduli spaces and give several applications, focusing in particular on the moduli space Mg of curves. These applications include Zariski density results for split Jacobians, non-density results for supersingular loci, and lifting results for algebraic cycles, all under appropriate numerical conditions.

These methods yield several applications, including Zariski density results for split Jacobians, non-density results for supersingular loci, and lifting results for algebraic cycles under appropriate numerical conditions. p. 0
Where the paper says it

Special Loci in the Moduli of Curves David Urbanik and Ziquan Yang October 7, 2026 Abstract We introduce new techniques to study mixed-characteristic special loci in moduli spaces and give several applications, focusing in particular on the moduli space Mg of curves. These applications include Zariski density results for split Jacobians, non-density results for supersingular loci, and lifting results for algebraic cycles, all under appropriate numerical conditions.

The techniques generalize to more-or-less arbitrary moduli spaces mapping to Hodge-type Shimura varieties (and some settings beyond the Shimura case), relating to questions raised by Oort and de Jong. p. 0
Where the paper says it

These applications include Zariski density results for split Jacobians, non-density results for supersingular loci, and lifting results for algebraic cycles, all under appropriate numerical conditions. Our methods generalize to more-or-less arbitrary moduli spaces mapping to Hodge-type Shimura varieties (and even some settings beyond the Shimura case), and relate to questions raised by Oort and de Jong.

Key theorems establish density results in various contexts: Theorem 1. p. 1

Where the paper says it

A locus in Mg of such special curves is called a special locus of Mg. In this paper we mostly focus on the symmetry of “splitting up to isogeny” and prove various density and non-density results, though as we will explain, the techniques generalize to other kinds of symmetries and special loci as well.

2 asserts that for a partition of genus g prime-to- p of splitting type g with a specific codimension condition delta pgq at most 3g - 3, there exists an r such that for every prime p > r, the locus SP g,p: t M F p ts J Jac q is Zariski dense in M F p. p. 1
Where the paper says it

Suppose that g is a partition of g P N such that δpgq ă 3g ´ 3. Then there exists rg P N such that for every prime p ą rg, the locus SPg,p :“ ts P Mg,Fp : JacpCsq is prime-to-p of splitting type gu (2) is Zariski dense in Mg,Fp .

Similarly, Theorem 1. p. 5
Where the paper says it

(The situation where one does not pass to reduced subschemes is more subtle. ) (iv) Similarly if R “ Fp, naïve analogues of the André-Oort conjecture are false: every abelian variety mod p carries a large enough endomorphism ring to force the Frobenius action on its Tate module to factor through a torus, and so in particular such “CM-like” points are dense in Mct g .

4 establishes density for the moduli space M of smooth cubic threefolds over Z(q) and a universal family X, showing that for all p > r, the locus ts M F p: IJ p X sq is Zariski dense in M F p. p. 2
Where the paper says it

Let M be the moduli space of smooth cubic threefolds over SpecpZq, and let X be the universal family over M. Then there exists an r such that for all p ą r, the locus ts P MFp : IJpXsq is prime-to-p of splitting type p1, 1, 1, 1, 1qu (4) is Zariski dense in MFp .

The paper also provides non-density results for supersingular loci. p. 1

Where the paper says it

A locus in Mg of such special curves is called a special locus of Mg. In this paper we mostly focus on the symmetry of “splitting up to isogeny” and prove various density and non-density results, though as we will explain, the techniques generalize to other kinds of symmetries and special loci as well.

Theorem 1.6 shows that when g at least 16, the union over all primes p of tau p M ctg q X A ss g,F p is not Zariski dense in tau p M ctg q. p. 2
Where the paper says it

We do not give a full answer to Oort’s question, but we can prove that τ pMct g,Fp q X A ss g,Fp is ď 0 dimensional for large g away from a Zariski closed locus independent of p. When g ě 16, the union over all primes ď pPSpecpZq rτ pMct g,Fp q X A ss g,Fp s ą0 is not Zariski dense in τ pMct g q.

Furthermore, Theorem 1. p. 0
Where the paper says it

3 Consequences for Flatness . 27 5 Proof of Theorem 1.

7 (Theorem 6. p. 0
Where the paper says it

3 Consequences for Flatness . 27 5 Proof of Theorem 1.

2) concerns orthogonal Shimura varieties with a quadratic lattice L of signature (p 2, b q), stating that the union of positive-dimensional atypical components d p PSpec p Z r1 Ns rT F p X S ss sigma 0 F p atyp is not Zariski dense in T, and the C-fiber of its Zariski closure is contained in a finite union of Shimura subvarieties strictly contained in S pLqC. p. 3
Where the paper says it

Write S ss,σ0 Fp for the supersingular locus with Artin invariant σ0. Then the union of positive-dimensional atypical components ď pPSpecpZr1{Nsq rTFp X S ss,σ0 Fp s atyp,ą0 is not Zariski dense in T, and the C-fiber of the Zariski closure of this union is contained in a finite union of Shimura subvarieties strictly contained in S pLqC.

The proof strategies are multifaceted: density proofs utilize mixed characteristic techniques applied to non-density results obtained in large positive characteristic, relying on analyzing boundary strata and the expected dimension inequality, which reduces to an unlikely intersection problem for expected-dimensional intersections with M F p reducing to a question of nondensity for unlikely-dimensional intersections with M'1,2,F p. Supersingular locus proofs employ atypical intersection techniques, leveraging the fact that the characteristic p condition is intrinsically positive-characteristic and does not require transcendence results for complex period maps. p. 8

Where the paper says it

These results arise from a somewhat different application of the atypicalintersection techniques used above, which is perhaps more surprising in the following sense. Supersingularity is an intrinsically positive-characteristic condition, defined in terms of Frobenius on crystalline cohomology, and there is no reason a priori to expect transcendence results for complex period maps to constrain supersingular loci in characteristic p.

Lifting results are implied by flatness results; Theorem 1. p. 0
Where the paper says it

3 Flatness and Lifting . 22 4 Flatness Results (for Mg,2 and Hg,2) 23 4.

12 states that outside a specific locus E, if a curve C M F p has a prime-to- p splitting of type g, then C lifts to a point in M F r together with the splitting, with R finite over W F p. p. 1
Where the paper says it

Suppose that g is a partition of g P N such that δpgq ă 3g ´ 3. Then there exists rg P N such that for every prime p ą rg, the locus SPg,p :“ ts P Mg,Fp : JacpCsq is prime-to-p of splitting type gu (2) is Zariski dense in Mg,Fp .

Specific results for general moduli spaces are presented: Section 4 shows that similar arguments apply to special cases like T = M 2 and X = A g, yielding density away from a strict closed subscheme E. p. 6

Where the paper says it

We note that nothing in points (i) through (vi) depended on us working with Mct g , and one can carry out a similar analysis for any locally closed locus T Ă Ag with Hodge-generic complex fibre TC, provided that one adjusts the dimension conditions accordingly. Similarly, although we focus on loci T Ă Ag in this paper, we expect our techniques to apply to any sufficiently general locus in any Hodge-type Shimura variety, or even to more general parameter spaces T for smooth projective algebraic varieties.

Theorem 5.17 provides a general result: if delta pgq at most 3g - 3, there exists an r p such that for every p > r p, the locus SP pol g, p: t M F p ts J Jac q, lambda can q is Zariski dense in M F p, and away from E, all components have the expected codimension delta pgq. p. 1
Where the paper says it

Suppose that g is a partition of g P N such that δpgq ă 3g ´ 3. Then there exists rg P N such that for every prime p ą rg, the locus SPg,p :“ ts P Mg,Fp : JacpCsq is prime-to-p of splitting type gu (2) is Zariski dense in Mg,Fp .

The proof of Theorem 5. p. 0
Where the paper says it

3 Consequences for Flatness . 27 5 Proof of Theorem 1.

p. 0
Where the paper says it

3 Consequences for Flatness . 27 5 Proof of Theorem 1.

17 proceeds by induction on g and shows that density holds away from a finite union of special subvarieties, where the final step involves showing that for a general smooth point C red, there exists an r 0-limit psi: D d to D q F p mapping to p beta 1qZ F p, implying the component lies in E.

The paper also details results for specific partitions and cases. p. 6

Where the paper says it

In these more general cases one replaces the atypical locus of Mct g,C with an atypical locus associated to a natural variation of Hodge structure on TC. 3 Proof Strategy for Density Our results in this paper will be obtained by applying the mixed characteristic techniques of (vi) to deduce novel density and non-density results in large positive characteristic.

Theorem 5.5 shows that SP pol g, p is Zariski dense in M F p, with components having expected codimension away from E, based on showing that for a general smooth point C red, every non-degenerate d-jet j xi: D d to C red has a constant image in A g to order xi, implying E is contained in a finite union of special subvarieties of A g. p. 8
Where the paper says it

We construct a descending sequence E1 Ě E2 Ě ¨ ¨ ¨ where Er is the Zariski closure of the points s P T for which, for a suitable choice of frame, there exists a non-degenerate jet j : D d r Ñ T based at s whose period jet factors through J d r Z. Since T is Noetherian, E8 :“ Xrě1Er stabilizes at some finite order.

Theorem 5.15 shows density for partitions where g = t 1, g-1, with components having codimension delta pgq = g-1. p. 6
Where the paper says it

2 over C is attributed to unpublished work of Harris; we do not know whether his proof was valid in positive characteristic. ) It turns out that making the Ciliberto-Harris-Miranda approach for density work for Mct g naturally leads to unlikely-intersection-type non-density questions for special intersections with boundary components.

The Elliptic Factor Case (Theorem 5. p. 1
Where the paper says it

4 Proof of the Elliptic Factor Case . 5 Proof of the General Case .

4) shows the locus H p n q is pure dimension 2g - 1, relying on the density of sets of points on smooth proper curves over p. p. 6
Where the paper says it

We note that nothing in points (i) through (vi) depended on us working with Mct g , and one can carry out a similar analysis for any locally closed locus T Ă Ag with Hodge-generic complex fibre TC, provided that one adjusts the dimension conditions accordingly. Similarly, although we focus on loci T Ă Ag in this paper, we expect our techniques to apply to any sufficiently general locus in any Hodge-type Shimura variety, or even to more general parameter spaces T for smooth projective algebraic varieties.

The proof structure for density often involves induction and leveraging results from previous sections, such as Theorem 5. p. 6

Where the paper says it

In these more general cases one replaces the atypical locus of Mct g,C with an atypical locus associated to a natural variation of Hodge structure on TC. 3 Proof Strategy for Density Our results in this paper will be obtained by applying the mixed characteristic techniques of (vi) to deduce novel density and non-density results in large positive characteristic.

1 (General Case), which establishes density for SP pol g, p away from a strict closed subscheme E. p. 8
Where the paper says it

If in fact this intersection pBφq ˚ pBYFp q contains the entirety of the intersection of Mg,Fp with Y Fp , then this means that the locus pBφq ˚ pBYFp q has larger-than-expected dimension. If such loci with larger-than-expected dimension are not dense in Mg´1,2,Fp , then necessarily (10) cannot happen away from a strict closed locus in Mg´1,2,Fp , and one obtains the desired density in Mg,Fp by deforming out from the boundary.

The proof of Theorem 5.2 relates the Zariski density of closed points on X of prime-to- p order to X not being contained in a p-special subvariety. p. 8
Where the paper says it

We construct a descending sequence E1 Ě E2 Ě ¨ ¨ ¨ where Er is the Zariski closure of the points s P T for which, for a suitable choice of frame, there exists a non-degenerate jet j : D d r Ñ T based at s whose period jet factors through J d r Z. Since T is Noetherian, E8 :“ Xrě1Er stabilizes at some finite order.

In the context of Shimura varieties, Theorem 6. p. 0

Where the paper says it

These applications include Zariski density results for split Jacobians, non-density results for supersingular loci, and lifting results for algebraic cycles, all under appropriate numerical conditions. Our methods generalize to more-or-less arbitrary moduli spaces mapping to Hodge-type Shimura varieties (and even some settings beyond the Shimura case), and relate to questions raised by Oort and de Jong.

2 demonstrates that atypical components are not Zariski dense away from finitely many primes, and their closure is contained in a finite union of Shimura subvarieties strictly contained in S C. p. 3
Where the paper says it

Write S ss,σ0 Fp for the supersingular locus with Artin invariant σ0. Then the union of positive-dimensional atypical components ď pPSpecpZr1{Nsq rTFp X S ss,σ0 Fp s atyp,ą0 is not Zariski dense in T, and the C-fiber of the Zariski closure of this union is contained in a finite union of Shimura subvarieties strictly contained in S pLqC.

Corollary 6. p. 9
Where the paper says it

This produces the horizontal sections needed for the jet argument; see Proposition 6. 10 and Corollary 6.

24 extends this to g at least 16, stating that away from finitely many primes, S g, pos X tau p M q is contained in finitely many strict Shimura subvarieties of A g. p. 2
Where the paper says it

When g ě 16, the union over all primes ď pPSpecpZq rτ pMct g,Fp q X A ss g,Fp s ą0 is not Zariski dense in τ pMct g q. Moreover, the image in Ag,C of the C-fiber of the Zariski closure of this union lands inside finitely many strict Shimura subvarieties of Ag,C.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved AI can perform Zariski density results for split Jacobians in moduli spaces like Mg,Fp under appropriate numerical conditions, as stated in Theorem 1.2, which generalizes previous AIM conjecture evidence to a broader class of special loci.

  2. Improved AI can prove non-density results for supersingular loci by adapting core techniques from split Jacobian density proofs to the context of positive characteristic and Frobenius actions, as suggested in the text regarding Question 1.5 (Oort).

  3. Improved AI can provide structural results for arithmetic special loci inside moduli spaces over Shimura varieties by using mixed characteristic techniques, specifically Theorem 1.11, which establishes that certain restricted pullbacks of maps factor through a strict Zariski closed locus E whose components are special subvarieties of the target space.

  4. Improved AI can analyze p-isogenies as well as density results for split Jacobians by applying the paper's methods to study supersingular loci, noting that every supersingular abelian variety is isogenous to a product of supersingular elliptic curves and relating this to prime-to-p splitting.

  5. Improved AI can determine whether a locus in Mg,Fp has the expected codimension by analyzing intersection properties using an all or nothing principle derived from Hodge theory and Kleiman's theorem, as seen in Theorem 5.17 for general partitions g.

  6. Improved AI can detect prime-to-p splittings from the abelian part of log 1-motifs by examining the structure of the associated log 1-motif Q, leveraging Lemma 5.7 and Proposition 5.8 to relate rank conditions to annihilation by integers prime to p.

  7. Improved AI can analyze infinitesimal period maps using jet schemes and refined jet correspondence, allowing for the algebraic parameterization of infinitesimal period maps (r-limp) and studying their behavior on non-constant jets as described in Section 2.

  8. Improved AI can construct atypical loci with respect to a given subscheme Z by defining specific loci Md r pZq based on the constancy or non-constancy of jets, where the resulting scheme Ed r stabilizes to a finite union of special subvarieties of T, as detailed in Construction 3.3 and Proposition 4.3.

  9. Improved AI can analyze the geometry of atypical loci along Ag by constructing families g: F over B whose fibers have constant dimension m and applying Proposition 4.5 to show that the resulting locus Ed r is not Zariski dense in T, as detailed in Proposition 4.1 and its proof steps.

  10. Improved AI can establish flatness results for special loci inside moduli spaces over Shimura varieties by using Theorem 3.11, which implies that certain fibre products are flat over R if the corresponding sub-Shimura datum is of type Z, as shown in the proof of Theorem 3.3 and its consequences (1.12, 1.13).

Abstract

We introduce new techniques to study mixed-characteristic special loci in moduli spaces and give several applications, focusing in particular on the moduli space g of curves. These applications include Zariski density results for split Jacobians, non-density results for supersingular loci, and lifting results for algebraic cycles, all under appropriate numerical conditions. Our methods generalize to more-or-less arbitrary moduli spaces mapping to Hodge-type Shimura varieties (and even some settings beyond the Shimura case), and relate to questions raised by Oort and de Jong.

Sources

Related papers