🔢 mathematics

Local stability of Frobenius and Artin--Schreier base-change: a comparison

This paper investigates how discrepancies of divisors behave under Artin--Schreier and Frobenius base-changes for families of pairs in positive characteristic, utilizing ramification data to recover a result on the permanence of local stability.

Original authors: Quentin Posva

Published 2026-08-26

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: 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

In the vast landscape of mathematics, there is a branch called algebraic geometry that studies shapes defined by equations. Imagine these shapes not as static drawings, but as families of objects that change smoothly as you move along a path, like a train of cars where each car is a slightly different version of the one before. For decades, mathematicians have been trying to understand how these families behave when they are stretched or twisted, a process known as base-change. In the world of ordinary numbers, which we call characteristic zero, we have a very clear set of rules for how these families hold together. However, when we switch to a different kind of number system based on a specific prime number, known as positive characteristic, the rules become much more slippery. A key concept in this field is "local stability," which essentially asks whether the family remains well-behaved and free of sudden, chaotic breaks when we perform these transformations. While we know this works in the ordinary world, it has been a mystery whether it holds true in these more exotic number systems.

A recent paper by Quentin Posva tackles this mystery by comparing two specific ways of twisting these families: one called an Artin–Schreier cover and another called a Frobenius cover. You can think of these as two different methods for creating a new, larger family from an existing one. The Artin–Schreier method is a type of twist that behaves like a standard rotation, while the Frobenius method is a more aggressive, purely mathematical operation that only exists in these specific number systems. The central question was whether the stability of the family is preserved when we apply these twists. Previous researchers had suggested that the difficult case of the Artin–Schreier twist could be reduced to the simpler Frobenius case, but the explanation for why this worked was buried in complex algebraic manipulations that were hard to follow.

Posva's work clarifies this relationship by looking at the specific points where the family might break or become unstable. He focuses on the "discrepancies," which are measurements of how much a shape deviates from being perfectly smooth at a specific spot. The paper proves that if a family is stable when twisted by the Frobenius method, it will also be stable when twisted by the Artin–Schreier method. The key to this discovery lies in understanding how the "wild" parts of the twist behave. In these number systems, some twists are "tame," meaning they are gentle and predictable, while others are "wild," meaning they are chaotic and difficult to control. The author demonstrates that even if a twist starts out as wild and chaotic, if you apply the Frobenius twist enough times in a row, the wildness eventually disappears, and the situation becomes tame and manageable.

This finding is significant because it confirms a long-held suspicion that the stability of these mathematical families is robust, even in the most difficult number systems. The paper does not just assume this is true; it provides a rigorous proof by tracking exactly how the chaotic parts of the twist evolve. The author shows that the chaotic behavior is temporary and that, given enough iterations of the Frobenius operation, the system settles into a stable state. This allows mathematicians to use the simpler, better-understood Frobenius case to predict the behavior of the more complex Artin–Schreier case. The result is a clearer, more transparent path to understanding these families, replacing a maze of complicated algebraic equations with a logical sequence of steps that shows exactly why the stability holds.

Ultimately, the paper resolves a technical gap in the work of previous researchers who had reached the same conclusion but relied on a method that was difficult to verify. By focusing on the underlying structure of the number systems rather than just manipulating the equations, Posva provides a more direct and convincing argument. The work confirms that the property of local stability is indeed preserved under these base-changes, provided the family is already stable under the Frobenius twist. This gives mathematicians a powerful new tool for constructing and analyzing these families, ensuring that the complex shapes they study remain coherent and well-defined, even when pushed to the limits of these strange number systems.

Technical Summary: Local Stability of Frobenius and Artin–Schreier Base-Change

Problem Statement
The paper addresses a fundamental question in the birational geometry of varieties in positive characteristic $p > 0$ : whether the property of "local stability" is preserved under base-change along a smooth curve. While local stability, introduced by Kollár and Shepherd-Barron, is a crucial technical condition for constructing compact moduli spaces in characteristic zero, its behavior in positive characteristic remains less understood. Specifically, it was an open question whether a locally stable family $(X, \Delta) \to C$ remains locally stable after base-change along an Artin–Schreier cover $C' \to C$ . Previous work by Hu and Zong [HZ20] suggested a reduction to the case of Frobenius base-change, but their technical arguments were complex and, as the author notes, contained a gap regarding the use of Frobenius descent.

Methodology
The author, Quentin Posva, provides a streamlined and transparent proof of the preservation of local stability under Artin–Schreier base-change. The approach relies on a detailed comparison of ramification data between Artin–Schreier extensions and purely inseparable (Frobenius) extensions of Discrete Valuation Rings (DVRs).

Ramification Analysis: The core technical machinery involves analyzing the extension of local rings $\mathcal{O}_{Y,E} \hookrightarrow \mathcal{O}_{Y',D}$ $O_{Y, E} ↪ O_{Y^{'}, D}$ , where $Y \to X$ $Y \to X$ is a birational morphism and $D$ $D$ is a divisor on the base-changed space dominating a divisor $E$ $E$ on $Y$ $Y$ . The author introduces and utilizes specific invariants to classify ramification:
- For Artin–Schreier extensions defined by $X^p - X - f$ , the author uses the quantity $\lambda(f)$ , defined via the valuation of $f + \wp(x)$ , to determine the ramification index $e$ and the ramification break $r$ .
- For degree $p$ purely inseparable extensions defined by $X^p - a$ , the author introduces the invariant $e\lambda(a)$ , defined via the valuation of $a + x^p$ .
Discrepancy Formulas: Using these invariants, the paper derives explicit formulas (Proposition 5.3) relating the discrepancy of a divisor $D$ in the base-changed family to the discrepancy of the original divisor $E$ . These formulas distinguish between "tame" divisors (where the coefficient of the fiber is prime to $p$ ) and "wild" divisors (where the coefficient is divisible by $p$ ).
Reduction to the Tame Case: The primary difficulty arises in the wild case. The author proves (Proposition 5.5) that for any wild divisor $E$ , there exists an integer $n \gg 1$ such that after $n$ iterations of Frobenius base-change, the corresponding divisor $E(n)$ becomes tame. This reduction allows the application of known results for the tame case.
Frobenius Descent: The proof concludes by invoking Frobenius descent for local stability (established in [BP26]), which allows lifting the stability property from the Frobenius base-changed family back to the original family.

Key Contributions and Results

Theorem 1.1 (Theorem 5.1): The main result establishes that if $(X, \Delta) \to C$ is a strongly locally stable family over a smooth curve in characteristic $p$ , and $C' \to C$ is an Artin–Schreier cover, then the base-changed family $(X', \Delta') \to C'$ is strongly locally stable, provided that the family remains strongly locally stable under all iterates of the Frobenius base-change.
Clarification of Hu–Zong's Argument: The paper recovers the main result of Hu and Zong [HZ20] but replaces their algebraic manipulations and partial normalizations via blow-ups with a more direct module-theoretic approach using ramification invariants ( $\lambda$ , $e\lambda$ , $r$ ).
Gap Filling: The author explicitly notes that Hu and Zong's original argument lacked a mention of Frobenius descent, which is strictly necessary for the reduction step. This paper integrates that descent step formally.
Comparison of Discrepancies: Section 5.1 provides a detailed comparison between the discrepancies arising from Artin–Schreier base-change and those from Frobenius base-change. The author demonstrates that one cannot generally restrict the stability condition to just the first Frobenius iterate ( $n=1$ ); higher iterates may be required to tame wild divisors.
Torsor Structures: The paper clarifies the structure of the effective models of group actions on the local rings, distinguishing between cases where the extension forms a torsor for $\mathbb{Z}/p$ (Artin–Schreier) versus $\alpha_p$ (purely inseparable).

Significance
The paper claims to offer a "more transparent" proof of the permanence of local stability under Artin–Schreier base-change. By avoiding the approximation of overring presentations and instead relying on precise ramification data, the author simplifies the technical landscape of the problem. The work confirms that the reduction strategy proposed by Hu and Zong is valid, provided one accounts for the necessity of Frobenius descent and the eventual tameness of wild divisors under iterated Frobenius base-change. This result is a step toward establishing the robustness of local stability as a condition for moduli problems in positive characteristic.

Limitations and Scope
The author remains modest regarding the scope of the results. The paper focuses specifically on the case where the base is a smooth curve and the base-change is Artin–Schreier. While it recovers the Hu–Zong result, it does not claim to solve the general problem of local stability for arbitrary base-changes without the intermediate Frobenius reduction. Furthermore, the author notes that the relationship between the invariants $\lambda$ and $e\lambda$ is complex, and the reduction to the tame case may require multiple Frobenius iterations, not just one.

More like this