$p$-adic hyperbolicity for Shimura varieties and period images

This paper establishes that rigid-analytic maps from punctured polydiscs into Shimura varieties or geometric period images with torsion-free level structures extend to the full polydisc over $p$-adic fields for sufficiently large primes $p$, provided the map's image intersects the good reduction locus, thereby proving a $p$-adic extension property and deducing applications to the algebraicity of such maps.

The Gist

Imagine you are an explorer trying to map a mysterious, vast landscape called Shimura Varieties. These aren't ordinary landscapes; they are complex mathematical worlds built from numbers and geometry, sitting at the intersection of algebra, geometry, and number theory.

For a long time, mathematicians knew how to navigate these worlds in the "complex" realm (using standard numbers like 3, $\pi$, and $i$). A famous mathematician named Borel proved in 1972 that if you have a map that works everywhere except a tiny hole in the middle of a region, you can always "fill in the hole" and extend the map to cover the whole area. It's like having a puzzle with one missing piece; Borel showed that the shape of the surrounding pieces guarantees the missing piece fits perfectly.

The Big Question:
Does this "filling the hole" rule also work in the p-adic world?

The p-adic world is a strange, alternative universe of numbers. Think of it as a landscape where distance is measured differently. Instead of numbers getting bigger as they move away from zero, they get "closer" if they share many factors of a specific prime number (like 5 or 7). It's like a fractal city where the streets loop back on themselves in a way that feels alien to our usual intuition.

Until now, no one knew if the "filling the hole" rule worked here. The authors of this paper—Bakker, Oswal, Shankar, and Yao—say: Yes, it does!

The Core Discovery: The "Magic Bridge"

The paper proves that if you have a rigid-analytic map (a very precise, rule-following path) in this p-adic world that goes around a hole (a punctured disk), you can almost always extend that path to cross the hole and continue smoothly.

Here is how they did it, using some creative metaphors:

1. The Two Types of Terrain: Good vs. Bad Reduction

Imagine the landscape has two types of soil:

• Good Reduction Soil: This is fertile, stable ground. If you walk on it, your footprints stay clear and predictable.
• Bad Reduction Soil: This is swampy, unstable ground. Walking here is messy; your footprints might get distorted or disappear.

The authors first proved a crucial rule: Your path cannot wander between the two. If your path starts on Good Soil, it stays on Good Soil. If it starts on Bad Soil, it stays on Bad Soil. It can't jump back and forth. This is like a river that either flows through a clear valley or a muddy swamp, but never switches mid-stream.

2. The "F-Crystal" Compass

To navigate, the explorers use a special tool called an F-crystal. Think of this as a magical compass that changes its direction based on the terrain.

• In the "Good Soil" areas, the compass behaves very predictably. The authors showed that if you shrink your map to a small enough area, the compass stops changing direction entirely. It becomes constant.
• If a compass is constant, it means the path you are walking is actually just a straight line (or a very simple curve) in disguise. This allows them to prove that the path can be extended across the hole without breaking.

3. The "Bad Soil" Problem and the "Retraction"

What if the path is stuck in the "Bad Soil" (the swamp)? This is harder.

• In the complex world, mathematicians had a trick called "Rapoport-Zink uniformization" (a fancy way of saying they could flatten the swamp into a simple map). But in the p-adic world, especially for the most exotic types of Shimura varieties, this trick doesn't exist yet.
• The Authors' Innovation: Instead of trying to flatten the swamp, they looked at the edges of the swamp. They realized that even in the bad soil, the "F-crystal compass" organizes itself into layers (like an onion). The outer layers of this onion are actually pulled from the "Good Soil" of the boundary (the edge of the map).
• By peeling back these layers, they could reduce the "Bad Soil" problem to a "Good Soil" problem. They essentially showed that even in the swamp, the path is being pulled toward a stable, predictable edge, allowing them to extend the map.

4. From One Hole to Many Holes

The paper starts by proving you can fill in a single hole (a 1-dimensional disk). But what if you have a whole grid of holes (a multi-dimensional polydisk)?

• The authors used a clever trick: If you can fill in any single line of holes, you can fill in the whole grid.
• They showed that if a map tries to be undefined at a specific point in a high-dimensional space, the "Good Soil" rules force it to be defined everywhere else. If it's defined everywhere else, the laws of p-adic geometry force it to be defined at that point too. It's like saying, "If a bridge is strong everywhere except one tiny spot, and the physics of the bridge demand it be strong everywhere, then that spot must also be strong."

Why Does This Matter?

This isn't just about filling holes in maps. It has a profound consequence called Algebraicity.

In mathematics, there are "analytic" objects (smooth, flexible shapes) and "algebraic" objects (rigid, equation-based shapes). Usually, proving that a flexible shape is actually a rigid equation is very hard.

• The Result: Because these maps can be extended so smoothly, the authors prove that any such map in the p-adic world is actually an algebraic map.
• The Analogy: Imagine you are drawing a curve with a flexible rubber band. You might think it's just a random squiggle. But this paper proves that if the rubber band follows these specific p-adic rules, it's actually a perfect, rigid circle drawn with a compass. You can write down an exact equation for it.

Summary

This paper is a tour de force of modern number theory. The authors took a difficult problem in a strange, non-intuitive number system (p-adic numbers), proved that paths in this system are incredibly rigid and predictable, and showed that even when the terrain looks messy ("bad reduction"), there is an underlying order that allows us to extend our maps and understand the geometry completely.

They didn't just find a new path; they built a bridge between the messy, p-adic world and the clean, algebraic world, showing that they are more connected than anyone thought.

Technical Summary

1. Problem Statement and Context

The paper addresses the p-adic extension and algebraicity problems for maps into Shimura varieties and geometric period images. This work serves as a p-adic analogue of classical theorems by Borel (1972) in the complex setting.

• The Complex Analogue (Borel): In the complex setting, Borel proved that any holomorphic map from a punctured polydisk $(D^\times)^a \times D^b$ into a Shimura variety extends to a map into its Baily-Borel compactification. A corollary is that any holomorphic map from a complex algebraic variety to a Shimura variety is algebraic (Borel's algebraicity theorem).
• The p-adic Challenge: Extending these results to the p-adic setting is non-trivial. Previous work (e.g., by Osipov, Shankar, Zhu, and others) established these results for abelian-type Shimura varieties using Rapoport-Zink (RZ) uniformizations. However, exceptional Shimura varieties and general geometric period images do not necessarily admit such uniformizations, leaving a gap in the theory.
• The Goal: The authors aim to prove p-adic extension theorems for all Shimura varieties (including exceptional ones) and geometric period images, without relying on RZ uniformizations. They also seek to establish algebraicity theorems for rigid-analytic maps in this context.

2. Methodology and Strategy

The authors develop a new strategy that bypasses the need for Rapoport-Zink uniformizations. Instead, they rely heavily on the theory of crystalline local systems, Fontaine-Laffaille modules, and prismatic cohomology.

The proof strategy proceeds in several logical stages:

A. Reduction to One-Dimensional Disks

The problem is first reduced to the case of a map $f: D^\times \to X^{an}$ (where $D^\times$ is the punctured unit disk). The authors prove that if the extension property holds for one-dimensional disks, it holds for higher-dimensional polydisks via meromorphic extension arguments.

B. Dichotomy of Reduction Types

A crucial first step is proving that the image of any map $f: D^\times \to X^{an}$ must lie entirely within either the good reduction locus or the bad reduction locus.

• This is established using a monodromy-theoretic description of the good reduction locus (analogous to the Neron-Ogg-Shafarevich criterion).
• The authors show that the rank of inertial invariants of the associated $\ell$-adic local system is constant on the image, forcing the image to stay within a specific boundary stratum (Baily-Borel stratum) if it is in the bad reduction locus.

C. The Good Reduction Case

If the image lies in the good reduction locus:

1. Extension of Local Systems: Using the p-adic Riemann-Hilbert correspondence (DLLZ23) and recent work by Du-Yao, the authors show that the crystalline local system extends from $D^\times$ to the full disk $D$.
2. Constancy of F-crystals: They utilize the theory of prismatic F-crystals. By shrinking the disk, they prove that the F-crystal associated with the Galois representation becomes constant (independent of the point in the disk).
3. Oort's Argument Generalization: They generalize an argument by Oort (2004) regarding F-crystals on $\mathbb{P}^1$. They show that if an F-crystal on a curve over $\mathbb{F}_p$ is pointwise constant and its associated isocrystal is constant, the map must be constant.
4. Conclusion: The map $f$ contracts the disk to a single residue disc, allowing the extension via the Riemann extension theorem.

D. The Bad Reduction Case (Shimura Varieties)

If the image lies in the bad reduction locus (near the boundary):

1. Stratification: The image is contained in the tube over a Baily-Borel stratum $T$.
2. Weight Filtrations: The authors construct a weight filtration on the log Fontaine-Laffaille module restricted to the boundary. This filtration is compatible with the monodromy operator.
3. Reduction to Boundary: They show that the associated graded object of the local system is pulled back from the boundary stratum (which is itself a lower-dimensional Shimura variety).
4. Induction: By induction on the dimension of the Shimura variety, the problem reduces to the good reduction case on the boundary stratum.

E. Higher Dimensions and Algebraicity

• Meromorphic Extension: Using the one-dimensional result, they prove that maps from polydisks extend meromorphically.
• Removal of Indeterminacies: They use the p-adic Riemann-Hilbert correspondence to show that the exceptional fibers in the resolution of indeterminacies must be contracted, yielding a regular extension.
• Algebraicity: Finally, they deduce that any rigid-analytic map from an algebraic variety to a Shimura variety (or period image) is the analytification of an algebraic map.

3. Key Contributions and Results

Main Theorem (Theorem 1.1)

Let $X$ be a Shimura variety or a geometric period image with torsion-free level structure, and $F$ a discretely valued p-adic field (for large enough $p$).

1. Shimura Varieties: Any rigid-analytic map $f: (D^\times)^a \times D^b \to X^{an}_F$ extends to a map $(D^\times)^a \times D^b \to X^{BB, an}_F$ (into the Baily-Borel compactification).
2. Geometric Period Images: If the image of $f$ intersects the good reduction locus (with respect to an integral canonical model), then $f$ extends to a map $D^{a+b} \to X^{an}_F$. (Note: If $X$ is proper, this hypothesis is vacuous).

Algebraicity Theorem (Theorem 1.2)

Any rigid-analytic map $f: M^{an} \to X^{an}$ defined over $F$ (where $M$ is an algebraic variety) is the analytification of an algebraic map $M \to X$.

Generalization (Theorem 1.4)

The results extend to smooth schemes over $W(\mathbb{F}_p)$ equipped with crystalline local systems satisfying specific positivity conditions (e.g., the Kodaira-Spencer map is immersive or the Griffiths bundle is ample). This covers a broader class of geometric objects beyond just Shimura varieties.

Technical Innovations

• Avoidance of Rapoport-Zink Uniformization: The proof works for exceptional Shimura varieties where RZ spaces are not known to exist.
• Prismatic F-crystals: The paper provides a robust application of prismatic cohomology to establish the constancy of F-crystals on disks, a key step in the p-adic extension.
• Logarithmic Structures: The use of log Fontaine-Laffaille modules and weight filtrations allows for a precise handling of the boundary behavior (bad reduction) without needing explicit uniformization.

4. Significance

• Unification: This work unifies the theory of p-adic hyperbolicity for both abelian and exceptional Shimura varieties, as well as general period images, under a single framework.
• New Tools: It demonstrates the power of prismatic cohomology and Fontaine-Laffaille theory in solving geometric extension problems, moving beyond the reliance on moduli interpretations (RZ spaces).
• Algebraicity: It establishes a strong p-adic GAGA-type theorem for Shimura varieties, confirming that p-adic analytic maps are inherently algebraic, provided the target has the appropriate arithmetic structure.
• Foundation for Future Work: The techniques developed (particularly the weight filtration on log crystals and the constancy of F-crystals) provide a toolkit for studying other arithmetic geometric problems involving p-adic uniformization and boundary behavior.

In summary, the paper resolves a major open problem in arithmetic geometry by proving that the "hyperbolicity" properties of Shimura varieties (extension of maps from punctured disks) hold in the p-adic setting for the most general cases, using a novel approach based on crystalline structures and prismatic cohomology rather than classical uniformization.