🔢 mathematics

Base-change of locally stable families in positive characteristic

This paper investigates the preservation of local stability for one-parameter families under finite flat base-changes in positive characteristic, demonstrating that it suffices to analyze Frobenius morphisms where the outcome is determined by discrepancies and specific differential invariants, for which upper bounds are established under $F$ -splitting assumptions.

Original authors: Marta Benozzo, Quentin Posva

Published 2026-07-27

📖 8 min read🧠 Deep dive

CC BY 4.0

Original authors: Marta Benozzo, Quentin Posva

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are an architect designing a city that exists on a strange, bumpy landscape. In this world, the ground isn't just flat; it has hidden cracks and folds that only appear when you try to stretch or shrink the city. This is the world of algebraic geometry, a branch of math where shapes are defined by equations rather than blueprints. The "city" is a collection of geometric shapes (varieties), and the "ground" is a curve that acts as a timeline or a base.

To keep this city stable, architects use a special rule called local stability. Think of this as a safety code: no matter where you look on the city, the buildings must be sturdy enough not to collapse into a chaotic mess. In the "normal" world (mathematicians call this characteristic 0), this safety code is very reliable. If you stretch the city using a standard map, the buildings stay safe. But this paper explores a very weird, "bumpy" version of the world called positive characteristic. Here, the rules of stretching are different. If you try to stretch the city in certain ways, the buildings might suddenly turn into a non-solid, non-normal mess, or the safety code might break in mysterious ways. The big question for mathematicians is: If we have a stable city, and we stretch it using these weird, bumpy rules, does it stay stable?

This paper, written by Marta Benozzo and Quentin Posva, dives deep into this mystery. They investigate what happens when you take a stable family of shapes and stretch it using a specific type of "wild" transformation called a Frobenius morphism. In this strange world, stretching doesn't just stretch; it can twist and fold the geometry in ways that create "wild" divisors—special lines or surfaces that behave erratically. The authors discover that the stability of the new, stretched city depends on two things: how much the original buildings were "off-center" (called discrepancies) and a new, tricky number they call the wild coefficient. This wild coefficient measures just how "wild" the twist is.

The team proves that if you want to know if any stretch will keep the city stable, you only need to check these specific Frobenius twists. They found a precise formula that connects the old stability to the new stability, but it's complicated. The formula shows that the wild coefficient acts like a "tax" on the stability. If the tax is too high, the city becomes unstable. While they can't predict the wild coefficient for every single case (it remains somewhat mysterious), they do manage to set an "upper limit" on how high this tax can get, but only if the city has a special property called F-splitting (a kind of built-in repair mechanism). In short, they haven't solved the whole puzzle for every possible stretch, but they've cracked the code for the most important type of stretch, revealing exactly how the geometry's "wildness" threatens its stability.

The Story of the Stretchy City

Let's break down the adventure. The authors are studying families of varieties, which you can imagine as a movie reel. Each frame of the movie is a geometric shape, and the whole reel is the family. The "local stability" is the rule that says every single frame must look decent and not have any catastrophic cracks.

In the "normal" world, if you take this movie and play it back at a different speed (a base-change), the frames usually stay decent. But in this positive characteristic world, the physics are weird. If you play the movie back using a "purely inseparable" speed (a Frobenius morphism), the frames might start to glitch. The buildings might fuse together in weird ways, or the ground might become non-normal (meaning it's not a solid, clean surface anymore).

The authors asked: If the original movie is stable, will the glitched version be stable?

They found that you don't need to check every possible glitch. You only need to check the "Frobenius glitch." If the city survives the Frobenius glitch, it survives everything. This is a huge simplification, like realizing that if your car can survive a specific type of pothole, it can survive any pothole.

The Wild Coefficient: The "Twist Tax"

So, what happens when the city gets the Frobenius glitch? The authors introduce a new character: the wild coefficient.

Imagine you are stretching a rubber sheet. Sometimes, the sheet stretches smoothly. Other times, it gets a weird knot or a twist that resists the pull. The wild coefficient is a number that measures how "knotted" or "twisted" a specific part of the sheet is.

If the twist is "tame" (smooth), the stability is preserved easily.
If the twist is "wild" (knotted), the stability gets hit with a penalty.

The paper derives a formula that calculates the new stability of the twisted sheet. It looks something like this:
New Stability = (Old Stability × p) - (Twist Penalty)

Here, p is a number related to the weirdness of the world (the characteristic). The Twist Penalty is the wild coefficient.

If the twist is "invariant" (it holds its shape perfectly under the stretch), the penalty is one thing.
If the twist is "non-invariant" (it changes shape), the penalty is different.

The scary part? The wild coefficient is hard to predict. It's like a hidden tax that depends on the specific details of the knot. The authors admit that in general, this tax is a mystery. However, they found a way to put a cap on this tax, but only if the city has a special "repair tool" called F-splitting.

The Repair Tool: F-Splitting

Think of F-splitting as a magical self-healing property. If a city has this property, it means there's a special operator (a mathematical tool) that can "split" the geometry back into a clean state.

When the authors applied this repair tool to their analysis, they found they could calculate a maximum possible value for the wild coefficient. They proved that:

If the twist is "non-invariant," the wild coefficient is bounded by a specific formula involving the old stability and the number p.
If the twist is "invariant," the bound is slightly higher (by a small amount, p-2).

This is a big deal because it gives a safety net. Even though we don't know the exact wild coefficient for every case, we know it can't be too crazy if the city has this repair tool.

The Limits of the Discovery

The authors are careful not to overpromise. They show that while they have an upper bound for the wild coefficient, this bound is often too loose to guarantee stability in every single case. It's like having a speed limit sign that says "Don't drive faster than 1000 mph," which doesn't help much if the car is already going 500 mph and the road is icy.

They also point out that for some specific, very stable cities (where the "discrepancy" is very low), their methods work perfectly, and stability is guaranteed. But for the general case, the "wild coefficient" remains a bit of a black box. The paper suggests that to fully solve the problem, we might need even more advanced tools, perhaps involving "higher foliations" (a more complex version of the twist analysis), which are currently being developed by other mathematicians.

Why This Matters

Why should a curious teenager care about twisted rubber sheets and magical repair tools? Because this isn't just about abstract shapes. These geometric rules are the foundation for building moduli spaces—which are essentially "maps of all possible shapes." If we want to classify every possible universe, every possible crystal structure, or every possible shape in the mathematical multiverse, we need to know which shapes are stable and which will collapse.

In the "normal" world, we have a good map. In this "bumpy" positive characteristic world, the map is full of holes. This paper fills in a few of those holes, showing us exactly how the geometry behaves when stretched by the Frobenius morphism. It tells us that the stability of our mathematical universe depends on a delicate balance between how much the shapes are "off-center" and how "wild" the twists get.

The authors haven't solved the entire mystery of the bumpy world, but they've handed us a better compass. They've shown that the key to understanding these wild stretches lies in the relationship between the discrepancies (the off-center-ness) and the wild coefficients (the twist tax). And while the tax is still a bit of a mystery, at least now we know there's a limit to how high it can go, provided the city has its magical repair tool.

In the end, the paper is a testament to the idea that even in the most chaotic, "wild" mathematical landscapes, there are hidden patterns and rules waiting to be discovered. It's a reminder that sometimes, to understand the stability of the whole, you have to get your hands dirty with the wild, knotted details of the parts.

Technical Summary: Base-Change of Locally Stable Families in Positive Characteristic

Problem Statement
The paper investigates the permanence of "local stability" for one-parameter families of varieties $X \to C$ under finite flat base-changes $C' \to C$ , specifically in the context of an algebraically closed field $k$ of positive characteristic $p > 0$ . In characteristic 0, local stability (requiring pairs $(X, X_c)$ to be log canonical for all closed points $c$ ) is stable under base-change. In positive characteristic, this fails due to specific pathologies: base-changes may become non-normal, and the generic fiber may fail to be geometrically log canonical. To address this, the authors introduce the notion of strong local stability, which imposes conditions on the geometric fibers (demi-normality with separable nodes for closed fibers, and geometric log canonicity for the generic fiber). The central question is whether strong local stability is preserved under arbitrary finite flat base-changes.

Methodology
The authors employ a reduction strategy combined with the theory of 1-foliations and differential invariants:

Reduction to Purely Inseparable Morphisms: Building on the work of Hu and Zong [HZ20], the authors prove that the problem reduces to checking base-changes along purely inseparable morphisms. Specifically, they show that if strong local stability is preserved under purely inseparable base-changes, it is preserved under arbitrary finite flat base-changes. This relies on decomposing the Galois closure of the base-change into tame covers, Artin–Schreier covers, and purely inseparable extensions, and establishing descent results for the latter two.
Frobenius Base-Change and 1-Foliations: Since purely inseparable morphisms are iterates of the Frobenius morphism (Lemma 2.1.8), the core analysis focuses on the base-change $X^{(1)} \to C$ along the Frobenius of $C$ . The authors utilize the theory of 1-foliations. The projection $X^{(1)} \to X$ factors the Frobenius of $X^{(1)}$ , inducing a morphism $X \to X^{(1)}$ which is the quotient by the relative tangent sheaf $T_{X/C}$ .
Wild Divisors and Coefficients: The authors introduce the concept of wild divisors: vertical divisors $E$ where the multiplicity of the fiber is divisible by $p$ . For such divisors, they define a wild coefficient $\gamma_w(E)$ , a non-negative integer measuring the differential complexity of the fibration at $E$ . This coefficient is defined via the order of the differential $du$ in the module of Kähler differentials, where $t = u s^n$ relates the base uniformizer $t$ to the fiber uniformizer $s$ .
Discrepancy Formulas: Using adjunction formulas for 1-foliations, the authors derive explicit formulas relating the discrepancies of the base-changed pair $(X^{(1)}, X^{(1)}_0)$ to the original pair $(X, X_0)$ and the wild coefficients.
F-Splitting Assumptions: To bound the wild coefficients, the authors assume the existence of a splitting operator (F-splitting) along the central fiber. They analyze the action of differential operators on the unit $u$ to derive upper bounds for $\gamma_w(E)$ .

Key Contributions and Results

Reduction Theorem (Theorem 1.0.2): The authors provide a rigorous proof that the permanence of strong local stability under arbitrary finite flat base-changes follows from its permanence under purely inseparable base-changes. This clarifies a claim made without proof in [HZ20].
Descent Results (Theorem 1.0.3): They establish that local stability descends along Frobenius morphisms and Artin–Schreier covers. This allows deducing the stability of a family from its base-changed version.
Discrepancy Formulas (Theorem 1.0.4): The paper provides precise formulas for the discrepancies $a(E^{(1)}; X^{(1)}, X^{(1)}_0)$ $a (E^{(1)}; X^{(1)}, X_{0}^{(1)})$ in terms of $a(E; X, X_0)$ $a (E; X, X_{0})$ and $\gamma_w(E)$ $γ_{w} (E)$ .
- If $E$ is tame, the discrepancy increases (or stays the same), preserving stability.
- If $E$ is wild, the discrepancy is given by:
  $a(E^{(1)}) = \begin{cases} p \cdot a(E) - (p-1)\gamma_w(E) & \text{if } E \text{ is } T_{X/C}\text{-invariant} \\ a(E) - \frac{p-1}{p}\gamma_w(E) & \text{otherwise} \end{cases}$
  This shows that stability is governed by the interplay between discrepancies and wild coefficients.
Bounds on Wild Coefficients (Theorem 1.0.5): Under the assumption that the family is 1-F-split along the central fiber, the authors establish upper bounds for $\gamma_w(E)$ $γ_{w} (E)$ .
- If $E$ is not $T_{X/C}$ -invariant: $\gamma_w(E) \leq p \lceil \frac{(p-1)a(E)}{p} \rceil$ .
- If $E$ is $T_{X/C}$ -invariant: $\gamma_w(E) \leq p \lceil \frac{(p-1)a(E)}{p} \rceil + p - 2$ .
- Note: The authors acknowledge that these bounds are generally too large by a factor of $p$ to guarantee stability in all cases, but they yield satisfactory results for specific cases (e.g., small discrepancies or $p=2$ ).

Significance and Claims
The paper claims to advance the understanding of moduli theory in positive characteristic by identifying the precise obstructions to the stability of families under base-change.

Novelty of Invariants: The authors state that wild coefficients have not been studied in the literature prior to this work. They highlight that bounding these coefficients in terms of discrepancies is a delicate task.
Limitations: The paper is modest regarding the general case. It notes that while the derived formulas explain the behavior of discrepancies, the "general" wild divisor behavior remains mysterious. The upper bounds obtained under F-splitting assumptions are often insufficient to prove stability for all discrepancies, particularly for invariant divisors.
Future Directions: The authors suggest that the precise relation between the birational Minimal Model Program (MMP) over a DVR and the differential properties of locally stable families is not yet clear. They indicate that going beyond the F-split case is challenging and that the framework of higher foliations (referencing forthcoming work [CRCFF26]) may be necessary to handle iterated Frobenius base-changes more effectively.

In summary, the paper reduces the problem of base-change stability to the study of Frobenius base-changes, introduces new differential invariants (wild coefficients) to quantify the failure of stability, and provides explicit formulas and partial bounds, while admitting that a complete characterization remains an open problem.

The Story of the Stretchy City

The Wild Coefficient: The "Twist Tax"

The Repair Tool: F-Splitting

The Limits of the Discovery

Why This Matters

More like this