🔢 mathematics

Frobenius--Tschirnhausen ampleness

This paper investigates smooth projective varieties with ample Frobenius-trace kernels, demonstrating that morphisms from such varieties are necessarily finite and proving that generalized Grassmannians of classical and $\mathrm{G}_2$ types possess this property, except in specific low-characteristic cases linked to exotic isogenies.

Original authors: Raymond Cheng, Emre Alp Özavcı

Published 2026-08-13

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Raymond Cheng, Emre Alp Özavcı

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 trying to understand the shape of a mysterious, perfectly smooth building. In the world of mathematics, specifically a field called algebraic geometry, these "buildings" are called varieties. They aren't made of brick and mortar, but of equations that define shapes in space. For a long time, mathematicians have been obsessed with a specific question: How "positive" or "bouncy" is the geometry of these shapes?

To understand this, think of a shape's "tangent bundle" as a field of tiny arrows sticking out of every point on the surface, pointing in every possible direction you could walk. If these arrows are "ample," it means the shape is incredibly rigid and curved in a very specific, positive way—so much so that the only shape that fits this description is the perfect, simple sphere (or its higher-dimensional cousin, projective space). But what if we look at a different kind of arrow field? In the world of "positive characteristic" (a specific type of math universe where numbers wrap around like a clock), there is a special transformation called the "Frobenius morphism." It's like a magical lens that squashes the shape onto itself in a very strange, non-reversible way. Associated with this squashing is a "kernel" bundle—a collection of arrows that tells us how the shape behaves under this magical lens. If this kernel is "ample," the shape is called "Frobenius–Tschirnhausen ample." This is a super-rare and special property that suggests the shape is not just a random blob, but a highly structured, beautiful object, likely a "Fano variety" (a shape that curves inward like a bowl) with a very simple, single-loop structure.

This paper, written by Raymond Cheng and Emre Alp Özavci, goes on a detective hunt to find out exactly which shapes have this special "Frobenius–Tschirnhausen ample" property. They wanted to know: If a shape has this property, what else must be true about it? And conversely, which famous shapes (like Grassmannians, which are spaces of lines and planes) actually possess this superpower?

The authors discovered a very strict rule for these special shapes: if you try to map a "Frobenius–Tschirnhausen ample" variety to any other shape, the map must be "finite." In plain English, this means you can't squish the shape down into a smaller dimension or stretch it out into a long tube; it must stay compact and map one-to-one onto its destination. This strongly suggests that these shapes are incredibly simple, likely having a "Picard rank" of 1, which is a fancy way of saying they have only one fundamental type of loop or hole.

The team then tested this theory on a family of famous shapes called "generalized Grassmannians." These are spaces that organize all the possible lines, planes, or higher-dimensional slices that fit inside a larger space while satisfying certain symmetry rules. The authors proved that most of these shapes do have the special ample property, but with some very specific, quirky exceptions.

They found that the rule holds true for almost all classical types of these shapes (like those related to symmetries of spheres and cubes) and even for the exotic "G2" type, unless the math universe they are living in has a very small "clock size" (characteristic). Specifically:

If the clock size is 2 (where 1 + 1 = 0), the symplectic Grassmannians (shapes related to symplectic forms) and certain orthogonal Grassmannians (shapes related to quadratic forms) fail the test.
If the clock size is 2 or 3, the G2-Grassmannian fails the test.

Why do they fail? The paper explains that in these tiny clock universes, there exist "exotic isogenies"—weird, secret shortcuts between different algebraic groups that shouldn't be connected. These shortcuts create "foliations" (like layers of leaves on a tree) that break the rigidity required for the shape to be "ample." For example, in characteristic 2, a specific type of Grassmannian can be squashed down in a way that reveals it's actually built from simpler, less rigid pieces, disqualifying it from being "Frobenius–Tschirnhausen ample."

The authors didn't just guess; they used a mix of "infinitesimal representation theory" (studying how tiny, almost invisible symmetries behave) and constructed special "splittings" (mathematical tools that break complex structures into manageable, positive pieces) to prove their points. They showed that for the successful cases, the "Frobenius-trace kernel" is not just positive, but so positive that even if you twist it by a negative amount, it remains globally generated (meaning it has enough "arrows" to cover the whole shape).

In summary, the paper confirms that "Frobenius–Tschirnhausen ample" varieties are a very exclusive club. They are likely the simplest, most rigid Fano shapes with a single loop. While most of the famous "Grassmannian" shapes belong to this club, the authors proved that in the weird, low-number worlds of characteristic 2 and 3, some of these shapes get kicked out of the club because of hidden, exotic connections that make them too flexible. The work provides a complete map of which classical and G2 Grassmannians make the cut, correcting a few previous inaccuracies and offering a clear picture of where the boundaries of this beautiful geometric property lie.

Problem Statement
This work investigates the geometric constraints imposed on smooth projective varieties $X$ over a field of positive characteristic $p > 0$ by the ampleness of their Frobenius–Tschirnhausen bundle, denoted $E_X$ . This bundle is defined as the dual of the Frobenius-trace kernel associated with the absolute Frobenius morphism $F: X \to X$ , specifically $E_X \coloneqq (F_* \mathcal{O}_X / \mathcal{O}_X)^\vee$ . The study is motivated by classical results in characteristic 0, such as Mori's theorem (where ampleness of the tangent bundle implies $X \cong \mathbb{P}^n$ ) and the Tschirnhausen ampleness conjecture for rational homogeneous spaces. While previous work by Carvajal-Rojas and Patakfalvi established that such varieties must be Fano and lack generically smooth fibrations or divisorial contractions, the precise classification and the existence of examples beyond projective space and quadrics remained open. The central question is: Which smooth Fano varieties of Picard rank 1 possess an ample Frobenius-trace kernel, and what are the necessary conditions on the characteristic $p$ ?

Methodology
The authors employ a combination of infinitesimal representation theory, the theory of Frobenius splittings, and explicit geometric constructions to analyze the positivity of $E_X$ .

Restrictions on Morphisms and Subschemas: The paper first establishes that if $E_X$ is ample, any non-constant morphism $f: X \to Y$ must be finite onto its image. This is proven by showing that any connected closed subscheme $Z \subset X$ satisfies the $G_1$ property (formal functions on the completion of $X$ along $Z$ are constants). This relies on the observation that the restriction of the Frobenius cokernel to a subscheme relates to the conormal bundle and infinitesimal neighborhoods.
Representation-Theoretic Criteria: For rational homogeneous spaces $X = G/P$ , the authors utilize the fact that $E_X$ is globally generated. They refine this by investigating the global generation of the twisted bundle $E_X \otimes \mathcal{O}_X(-1)$ . Using the identification of the Frobenius-trace kernel with coinvariants of the tangent bundle action, they relate ampleness to the surjectivity of restriction maps on global sections of $\omega_X^{\otimes (1-p)} \otimes \mathcal{O}_X(-p)$ to $p$ -Frobenius jets.
Explicit Constructions (Rolling Determinants and Pfaffians): To prove global generation for classical types (Grassmannians, Symplectic, Orthogonal) and Type $G_2$ , the authors construct explicit sections. They introduce "rolling determinants" for classical Grassmannians and "rolling Pfaffians" for spinor varieties. These constructions involve specific polynomials in the minors of coordinate matrices on the big Schubert cell, designed to realize specific weight vectors (derivatives of the trace element $\tau$ ) in the coordinate ring of the Frobenius kernel.
Obstruction via Exotic Isogenies: For the exceptional cases where ampleness fails, the authors demonstrate the existence of finite purely inseparable morphisms (exotic isogenies) between algebraic groups or their associated homogeneous spaces. They show that the Tschirnhausen bundle of such a morphism restricts to a non-ample bundle on specific subvarieties (e.g., lines or lower-dimensional Grassmannians), thereby obstructing the ampleness of $E_X$ .

Key Contributions and Results

Theorem A (Finiteness of Morphisms): Any non-constant morphism from a Frobenius–Tschirnhausen ample variety $X$ to another variety $Y$ is finite onto its image. Consequently, $X$ admits no nontrivial extremal contractions. Combined with the Fano property, this strongly suggests (and proves under the assumption of the contraction theorem) that such varieties must have Picard rank 1.
Theorem B (Non-ampleness of Hypersurfaces): The authors construct explicit families of smooth Fano hypersurfaces in $\mathbb{P}^n$ of degree $d$ (where $p < d \le n$ ) that are not Frobenius–Tschirnhausen ample. These examples include "q-bic" or extremal hypersurfaces, generalizing the known non-ampleness of the quadric threefold in characteristic 2.
Theorem C (Classification for Classical and $G_2$ Types): The paper provides a complete classification of Frobenius–Tschirnhausen ample generalized Grassmannians of classical type and Type $G_2$ $G_{2}$ . A projective rational homogeneous space $X$ $X$ of Picard rank 1 is Frobenius–Tschirnhausen ample if and only if it is not one of the following exceptions:
- Symplectic Grassmannian $SGr(k, 2m)$ with $2 \le k \le m$ in characteristic $p=2$ .
- Orthogonal Grassmannian $OGr(m-1, 2m+1)$ in characteristic $p=2$ .
- $G_2$ -Grassmannian in characteristics $p=2$ or $p=3$ .
Correction of Previous Results: The work corrects a slight inaccuracy in prior literature regarding quadrics in characteristic 2, showing that quadrics of dimension $\ge 4$ in $p=2$ are Frobenius–Tschirnhausen ample, whereas the quadric threefold is not.

Significance and Claims
The paper claims to provide the first comprehensive set of restrictions and examples for Frobenius–Tschirnhausen ample varieties in positive characteristic. By proving that such varieties admit no nontrivial contractions, the authors provide strong evidence supporting the conjecture that these varieties are Fano varieties of Picard rank 1.

The significance of the results lies in:

Completing the Classification: Theorem C resolves the ampleness question for the most accessible class of Fano varieties (generalized Grassmannians) of classical type and Type $G_2$ , identifying precise characteristic-dependent obstructions.
Geometric Mechanism: The paper elucidates the mechanism behind the failure of ampleness in low characteristics, attributing it to the existence of "exotic isogenies" between algebraic groups (e.g., between $SO_{2m+1}$ and $Sp_{2m}$ in $p=2$ , and self-isogenies of $G_2$ in $p=3$ ). These isogenies induce morphisms whose Tschirnhausen bundles fail to be ample.
Methodological Advancement: The authors introduce a technique that avoids the explicit decomposition of $F_* \mathcal{O}_X$ (which is often intractable) by instead constructing specific Frobenius splittings and analyzing the global generation of twisted trace kernels via representation theory.

The authors remain modest regarding the exceptional types (other than $G_2$ ), conjecturing that generalized Grassmannians of other exceptional types are Frobenius–Tschirnhausen ample provided the characteristic is "good" (e.g., $p > 5$ ), but they do not prove this. The work establishes a rigorous link between the positivity of the Frobenius-trace kernel and the global geometry of the variety, suggesting a converse to the intuition that positivity of the tangent bundle implies positivity of the Frobenius-trace kernel.

More like this