🔢 mathematics

Unboundedness of zero-cycles on higher dimensional Fano manifolds

This paper demonstrates that, in contrast to del Pezzo surfaces, higher-dimensional Fano manifolds generally fail to satisfy boundedness properties for their group of 0-cycles, exhibiting phenomena such as the absence of Coray-type bounds on minimal odd degrees and the unboundedness of effective 0-cycles.

Original authors: Claire Voisin

Published 2026-08-03

📖 9 min read🧠 Deep dive

CC BY 4.0

Original authors: Claire Voisin

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 a detective trying to solve a mystery about the hidden "address book" of a geometric shape. In the world of mathematics, specifically a field called algebraic geometry, shapes are defined by equations, and they live in spaces that can have many dimensions. One of the most interesting types of shapes are called Fano manifolds. Think of these as the "perfectly curved" shapes of the mathematical universe, like a sphere or a cube, but they can exist in 3, 4, or even 100 dimensions. They are special because they are "rationally connected," meaning you can draw a smooth, straight line (or a curve that looks like a line) connecting any two points on them.

Now, mathematicians are obsessed with counting "points" on these shapes. But these aren't just dots; they are "zero-cycles," which are like collections of points that might be stuck together in complex ways. A key question is: How big do these collections need to be before we can be sure they can be broken down into simple, single points? This is called "boundedness." If a shape has "bounded" zero-cycles, it means there is a magic number $N$ . If you have a collection of points with a total size bigger than $N$ , you are guaranteed that this collection is actually just a bunch of real, existing points. It's like saying, "If you have a pile of 100 coins, you can definitely find 100 real coins in your pocket." For some simple shapes, like 2D surfaces (think of a fancy, curved sheet), mathematicians already knew this magic number exists. But for higher-dimensional shapes, nobody was sure if such a limit existed, or if the "pile" could get so weird that no matter how many points you had, you might still be missing a single, real point.

This paper, written by Claire Voisin, tackles this mystery for higher-dimensional Fano shapes. The author proves that, unlike their simpler 2D cousins, these complex, high-dimensional shapes do not have a magic number limit. In fact, the "address book" of these shapes is unbounded. The paper shows that you can construct specific examples of these shapes where, no matter how large a number $N$ you pick, there is a version of the shape that has a collection of points of size $N$ that cannot be reduced to a single point of a smaller, odd size. It's as if you have a pile of 1,000,000 coins, but no matter how hard you try, you cannot find a single real coin hidden inside; the pile is "unbreakable" in a specific mathematical sense. The paper also introduces a new way of thinking about this "unboundedness" by connecting it to the behavior of "differential forms" (which are like mathematical fluids flowing over the shape), showing that if these fluids exist in certain ways, the point-counting limit cannot exist.

The Main Discovery: The Infinite Pile

The core finding of this paper is a definitive "no" to a long-standing question. Mathematicians had hoped that for any smooth Fano shape (like a quartic threefold, which is a 3D shape defined by a specific type of equation), there would be a universal bound. They wondered: "If I have a collection of points with an odd total size, is there a maximum size I need to check to find a single point?" For 2D shapes (del Pezzo surfaces), the answer was yes; there is a limit. But Voisin proves that for shapes with 3 or more dimensions, this limit does not exist.

The paper constructs a specific, "generic" example of a 3D shape (a quartic hypersurface) defined over a special field. In this example, for any odd number $N$ you choose (say, 7, 9, 11, or a million), the author shows you can create a version of this shape that has a "point" of size $N$ , but no point of any smaller odd size. This means the "minimal odd degree" of a point on this shape can be arbitrarily large. There is no ceiling. The paper explicitly rules out the idea that a "Coray-type bound" (a specific kind of limit named after a previous mathematician) exists for these higher-dimensional shapes. It is not just that we haven't found the number yet; the paper proves that no such number can exist for these specific types of shapes.

How the Detective Worked: The "Specialization" Trick

To prove this, Voisin uses a clever mathematical technique called specialization. Imagine you have a shape that is very smooth and perfect (the "generic" shape). The author imagines slowly deforming this shape, like melting a block of ice, until it turns into a slightly different shape in a different mathematical world (specifically, a world where the math works with the number 2, known as characteristic 2).

In this "melted" state, the shape becomes a "double cover" of a simpler shape, and it develops some singularities (kinks or folds). However, the author shows that if you smooth out these kinks (desingularization), the resulting shape has a very special property: it contains a non-zero "algebraic form" of a certain degree. Think of this form as a unique, non-vanishing "flow" or "vibration" that exists on the smoothed shape.

Here is the crucial link: The paper proves that if a shape has this kind of "flow" (a non-zero form of degree 2 or higher), then it cannot have a bounded set of points. The logic is that if the points were bounded, you could perform a mathematical trick (involving "traces" and "ranks" of these flows) that would force the flow to disappear. But since the flow is proven to exist and be non-zero, the assumption that the points are bounded must be false. It's like saying, "If the music is still playing, the speaker cannot be broken."

The "Unbounded" Nature of the CH0 Group

The paper also introduces a concept called the unbounded CH0-group. In simple terms, the CH0-group is a way of organizing all the possible collections of points on a shape. If a shape has a "bounded" CH0-group, it means that once your collection of points gets big enough, it's guaranteed to be a "real" collection (effective). If it's "unbounded," it means you can always find a collection of points that is "fake" or "impossible" to break down, no matter how large it is.

Voisin proves that for very general hypersurfaces of even degree $d$ in dimensions $n \ge 3$ (where $d$ is large enough, specifically $d \ge 2\lceil \frac{n+2}{3} \rceil$ ), the CH0-group is unbounded. This means there is no integer $N$ such that all zero-cycles of degree $\ge N$ are effective. The paper establishes this by combining the "specialization" method with a new, generalized version of a famous theorem by David Mumford. Mumford originally showed that if a shape has certain "flows" (forms), its point-group is infinite-dimensional. Voisin extends this to show that even if the shape is "rationally connected" (which usually makes things simple), the universal version of this group (looking at the shape over all possible field extensions) can still be unbounded.

The Quartic Threefold Example

A major part of the paper focuses on quartic threefolds (3D shapes defined by an equation of degree 4). The author constructs a specific scenario:

Start with a generic quartic threefold over a field of characteristic 0 (like the rational numbers).
Consider a field extension that adds a "generic point" of odd degree $N$ .
The paper proves that in this new setting, the shape has a point of degree $N$ , but no point of any smaller odd degree.

This result is striking because it contrasts sharply with 2D surfaces (del Pezzo surfaces), where such a bound does exist. For a 2D surface of degree 2, if you have a point of odd degree, you are guaranteed to find a point of degree 1, 3, or 7. But for the 3D quartic, the author shows you can have a point of degree 101, 1001, or 1,000,001, with no smaller odd-degree points to be found. The paper explicitly states that this holds for quartic hypersurfaces in 4D space ( $P^4$ ) and even for double covers of 4D space ramified along sextic or octic hypersurfaces.

The Role of "Tensor Rank" and Characteristic 2

One of the technical hurdles the paper overcomes is working in "characteristic 2" (a mathematical world where $1+1=0$ ). In this world, standard ways of measuring the "size" or "rank" of mathematical objects (like the flows mentioned earlier) break down. Voisin introduces the concept of tensor rank to handle this.

Think of "rank" as the number of simple building blocks needed to construct a complex object. In normal math, this is straightforward. In characteristic 2, the rules change. The paper proves that even with these tricky rules, the "tensor rank" of the flow on the specialized shape is high enough to prevent the "boundedness" from happening. Specifically, the paper shows that if you try to assume the points are bounded, you end up with a contradiction in the "tensor rank" of the flows: the rank on one side of the equation would be too small to match the rank on the other side. This contradiction proves that the assumption (that points are bounded) is false.

Conclusion: The Infinite Frontier

In summary, this paper shatters the hope that higher-dimensional Fano manifolds behave like their simpler 2D counterparts regarding the "size" of their points. It proves that for a wide class of these shapes, the "minimal odd degree" of a point is unbounded. There is no universal limit. You can always find a shape where the smallest "odd" point is as large as you want.

The paper does not just suggest this; it provides a rigorous proof using specialization, desingularization, and the properties of algebraic forms. It explicitly rules out the existence of a "Coray-type bound" for these shapes. The confidence is high, grounded in theorems that are proven to hold for "very general" hypersurfaces. The work bridges the gap between the arithmetic of points and the geometry of flows, showing that the "infinite dimensionality" of these shapes is a fundamental feature, not just a quirk of a specific example. For anyone interested in the deep structure of geometric shapes, this paper reveals that the universe of higher-dimensional points is far more chaotic and limitless than previously imagined.

=== SUMMARY ===
Technical Summary: Unboundedness of Zero-Cycles on Higher Dimensional Fano Manifolds

Problem Statement
The paper investigates the boundedness properties of the Chow group of 0-cycles ( $CH_0$ ) for smooth projective Fano manifolds defined over fields of characteristic 0. While it is a classical result (and recently refined by Colliot-Thélène and the author) that del Pezzo surfaces (Fano surfaces of degree $\le 9$ ) satisfy a "boundedness" property—meaning there exists an integer $N$ such that any 0-cycle of degree $\ge N$ is effective—the situation for higher-dimensional Fano varieties remains less understood.

Specifically, the paper addresses two related questions:

Coray-type bounds: Does there exist a uniform bound $N(n)$ such that any smooth Fano hypersurface of dimension $n$ with a 0-cycle of degree 1 possesses a closed point of degree $\le N(n)$ coprime to the degree of the hypersurface?
Universal Boundedness of $CH_0$ : Does there exist an integer $N$ (depending on the variety and field) such that for any field extension $L \supset K$ , any 0-cycle of degree $\ge N$ on $X_L$ is effective? This notion is an arithmetic analogue of Mumford's concept of finite dimensionality for $CH_0$ .

Methodology
The author employs a combination of specialization techniques, algebraic cycle theory, and the analysis of differential forms to establish unboundedness.

Extension of Mumford's Argument: The paper adapts Mumford's theorem (which links the non-vanishing of global differential forms $H^0(X, \Omega^l)$ for $l \ge 2$ to the infinite dimensionality of $CH_0$ ) to the setting of universal boundedness over arbitrary fields. The core technical tool is a rank argument involving the action of algebraic cycles on differential forms.
- The author defines a "bounded $CH_0$ " condition where 0-cycles of sufficiently high degree are effective.
- Theorem 1.12 establishes that if a variety $X$ has bounded $CH_0$ , then specific relations between the universal 0-cycle $\Gamma_N$ and effective cycles must hold. By analyzing the action of these cycles on differential forms (via pull-backs and traces), the author proves that if $X$ admits non-zero differential forms of degree $l \ge 2$ , it cannot have bounded $CH_0$ .
- Crucially, the proof handles non-zero characteristic by utilizing tensor ranks of forms rather than standard linear ranks, particularly in characteristic 2, to manage inseparability issues.
Specialization to Characteristic 2: To demonstrate the existence of Fano varieties with unbounded $CH_0$ , the paper utilizes the specialization method pioneered by Kollár, Colliot-Thélène, and Totaro.
- The author considers very general hypersurfaces of even degree $d$ in $\mathbb{P}^{n+1}$ .
- These are specialized to characteristic 2, where they become cyclic double covers of hypersurfaces ramified along a divisor.
- The desingularized special fiber in characteristic 2 is shown to possess non-zero algebraic forms of degree $n-1$ (or 2/3 in specific dimensions).
- By combining the specialization method with the rank arguments from Theorem 1.12, the author deduces that the generic fiber (over characteristic 0) must have unbounded $CH_0$ .
Construction of Counterexamples for Coray-type Bounds: For the specific case of quartic threefolds, the paper constructs a generic quartic threefold $X_\eta$ over a function field. By analyzing the rational equivalence of the generic point of degree $N$ against points of lower odd degree, and using a birational map $\sigma$ to relate these cycles, the author shows that no "Coray-type" upper bound exists for the minimal odd degree of points.

Key Contributions and Results

Theorem 1.4 (Failure of Coray-type bounds): The paper proves that for quartic threefolds (and other specific Fano 3-folds and 4-folds), there is no uniform bound $N$ on the minimal odd degree of closed points, even if the variety possesses a 0-cycle of degree 1. Specifically, for any odd integer $N \ge 7$ , there exists an extension field where the variety has a point of degree $N$ but no point of odd degree strictly smaller than $N$ .
Theorem 1.11 (Unbounded $CH_0$ for Fano Hypersurfaces): The main result states that very general hypersurfaces of even degree $d$ in $\mathbb{P}^{n+1}$ (where $n \ge 3$ and $d \ge 2\lceil \frac{n+2}{3} \rceil$ ) have unbounded $CH_0$ . This means there is no integer $N$ such that all 0-cycles of degree $\ge N$ are effective over any field extension. The result also holds for double covers of $\mathbb{P}^n$ ramified along smooth hypersurfaces.
Theorem 1.8 (Boundedness for Del Pezzo Surfaces): For completeness, the paper extends previous results to show that all smooth del Pezzo surfaces (degrees 4 through 9) have bounded $CH_0$ , with a bound $N$ independent of the surface and the ground field.
Theoretical Framework (Theorem 1.12): The paper provides a generalized criterion (valid in characteristic 0 and specific cases of non-zero characteristic) linking the existence of non-zero differential forms of degree $l \ge 2$ to the unboundedness of the universal $CH_0$ group.

Significance
The paper fundamentally alters the understanding of 0-cycles on higher-dimensional Fano varieties. While Fano varieties are geometrically rationally connected (implying $CH_0(X_{\bar{K}}) \cong \mathbb{Z}$ ), this work demonstrates that their arithmetic behavior regarding 0-cycles is far more complex than that of del Pezzo surfaces.

The significance lies in:

Negative Answer to Generalized Coray Conjectures: It shows that the "Coray-type" bounds observed for surfaces do not extend to higher dimensions, even for Fano varieties with index 1 or 2.
Distinction between Rational Connectedness and Boundedness: It highlights that rational connectedness (which implies triviality of $CH_0$ over the algebraic closure) does not imply the boundedness of $CH_0$ over arbitrary field extensions.
Methodological Advancement: The successful adaptation of Mumford's rank arguments to the universal setting and to non-zero characteristic (using tensor ranks) provides a robust tool for studying the arithmetic of algebraic cycles in positive characteristic and for varieties with non-trivial universal $CH_0$ .

The results rely on the existence of non-trivial universal $CH_0$ groups for certain Fano varieties, a phenomenon previously established by Totaro and others, but here linked explicitly to the failure of degree bounds for points.