🔢 mathematics

Automorphisms of Smooth Hypersurfaces with Fixed Loci of Codimension at Most Two

This paper investigates smooth hypersurfaces in projective space by demonstrating how the condition that an automorphism's fixed locus has codimension at most two restricts the automorphism's possible orders and influences the rationality of the associated quotient spaces.

Original authors: Taro Hayashi, Ryoichi Suzuki

Published 2026-03-03

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Taro Hayashi, Ryoichi Suzuki

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 looking at a perfectly smooth, multi-dimensional soap bubble floating in a vast, colorful space. In the world of mathematics, this is called a smooth hypersurface. Now, imagine you have a magical wand (an automorphism) that can spin, flip, or stretch this bubble.

Usually, when you wave this wand, the bubble changes shape. But sometimes, parts of the bubble stay exactly where they are. These unmoving parts are called the fixed locus.

This paper is like a detective story. The authors, Taro Hayashi and Ryoichi Suzuki, are trying to figure out: "If we know exactly how much of the bubble stays still, what kind of magic tricks (orders of automorphisms) are possible?"

Here is the breakdown of their findings using simple analogies:

1. The Setting: The Bubble and the Magic Wand

Think of the bubble as a shape in a high-dimensional room (Projective Space).

The Shape: It's defined by a specific "recipe" (degree $d$ ). A degree 3 shape is like a twisted torus; a degree 4 shape is more complex, like a kaleidoscope.
The Magic: The wand performs a rotation. If you rotate a shape 360 degrees, it looks the same. If you rotate it 120 degrees three times, it also looks the same. The number of times you need to wave the wand to get back to the start is called the order.
The Clue: The authors focus on cases where the "still parts" (the fixed locus) are very large. Specifically, they look at cases where the still parts are so big that they only miss the rest of the bubble by a tiny margin (codimension 1 or 2).

2. The Big Discovery: The "Stillness" Limits the "Spin"

In the past, mathematicians knew the maximum speed a bubble could spin based on its recipe. But they didn't know how the size of the still parts restricted the spin.

The authors found a strict rule: If a large chunk of the bubble stays still, the magic wand can only spin at very specific speeds.

Scenario A: The "Equator" is Still (Codimension 1)
Imagine the bubble has a giant equator that doesn't move at all.
- The Rule: The wand can only spin at speeds related to the recipe number ( $d$ ), or one less ( $d-1$ ), or two less ( $d-2$ ).
- The Catch: If the speed is $d-2$ , the bubble must be very small (specifically, a 2D surface in 3D space). If the bubble is huge, this speed is impossible.
Scenario B: The "Pole" is Still (Codimension 2)
Imagine only a small circle (like a pole) stays still.
- The Rule: The allowed speeds get more complicated. They are combinations like $(d-1) \times d$ , or $(d-1)^2$ , or even stranger numbers like $d^2 - 3d + 3$ .
- The Catch: The size of the bubble (how many dimensions it has) changes the list of allowed speeds. A 3D bubble has different rules than a 4D bubble.

3. The Second Discovery: When the Bubble Becomes "Rational"

This is the most exciting part for the "architects" of math.

The Question: If you take the bubble and fold it up according to the magic wand's pattern (creating a quotient space), does the result look like a simple, empty room (a rational variety)? Or is it a twisted, knotted mess that can't be untangled?
The Finding: The authors proved that if the wand spins at certain "special" speeds (multiples of $d$ , $d-1$ , or $d-2$ ) and the still parts are large enough, the resulting folded shape is always simple and rational.
The Analogy: Imagine taking a crumpled piece of paper (the complex shape) and folding it perfectly along the lines where the magic wand didn't move. The authors proved that if you follow these specific rules, the final folded paper is actually just a flat, smooth sheet. It's not a knot; it's a clean, simple surface.

4. Why This Matters

Before this paper, mathematicians had a list of "possible speeds" for the wand, but it was like a list of all possible speeds a car could theoretically reach.
Hayashi and Suzuki added a new rule: "If the car is carrying a heavy, unmoving cargo (the fixed locus), it can only reach these specific speeds."

They also fixed a small error in a previous study (Theorem 3.17), showing that sometimes, even if the "cargo" is heavy, the car behaves differently than previously thought.

Summary in One Sentence

This paper proves that if a high-dimensional shape has a large "frozen" section, the symmetries (rotations) it can undergo are strictly limited to a few specific numbers, and when these rotations happen, the resulting shape is surprisingly simple and easy to understand.

1. Problem Statement

The paper investigates the automorphism groups of smooth hypersurfaces $X$ of degree $d$ in the projective space $\mathbb{P}^{n+1}$ (where $n \ge 2$ ) over an algebraically closed field of characteristic zero.

While the possible orders of automorphisms for such hypersurfaces have been classified numerically in previous works (notably by Zheng), these classifications are based solely on the dimension and degree of the variety. The authors aim to refine this classification by incorporating geometric constraints, specifically the dimension of the fixed locus ( $\text{Fix}(g)$ ) of an automorphism $g$ .

The central problem is to determine how the assumption that the fixed locus has codimension at most two in $X$ restricts the possible orders of the automorphism and influences the geometric properties of the quotient space $X/\langle g \rangle$ .

2. Methodology

The authors employ a combination of linear algebra, algebraic geometry, and combinatorial analysis of polynomial equations. Their methodology proceeds as follows:

Reduction to Diagonal Forms: Since the automorphism group of a smooth hypersurface (under mild conditions) is a subgroup of $\text{PGL}(n+2, k)$ , any automorphism $g$ can be represented by a matrix $A$ . By analyzing the eigenspaces of $A$ , the authors reduce the study to diagonal matrices.
Classification by Fixed Locus Codimension: The condition that $\text{codim}(\text{Fix}(g)) \le 2$ $codim (Fix (g)) \leq 2$ implies that the number of distinct eigenvalues of the representing matrix is small ( $r \le 4$ $r \leq 4$ ). This allows the authors to classify the automorphisms into six specific normal forms (Types I–VI) based on the multiplicity of eigenvalues and the structure of the fixed locus:
- Type I: One non-trivial eigenvalue (Codimension 1).
- Types II & III: Two non-trivial eigenvalues (Codimension 1 or 2).
- Types IV, V, VI: Three non-trivial eigenvalues (Codimension 2).
Polynomial Analysis: For each type, the authors analyze the defining homogeneous polynomial $F$ of the hypersurface. They impose the condition that $g$ is an automorphism, which implies $A^*F = tF$ for some scalar $t$ . By expanding $F$ in terms of monomials involving the coordinates corresponding to the eigenvalues, they derive strict algebraic constraints on the exponents and the scalar $t$ .
Smoothness Constraints: A crucial step involves using the smoothness of $X$ . The authors show that if the fixed locus contains certain linear subspaces or if the polynomial lacks specific monomials, the hypersurface would possess singular points, leading to contradictions. This eliminates many potential cases and restricts the dimension $n$ in specific scenarios.
Rationality via Galois Points: To address the rationality of the quotient space, the authors utilize the theory of Galois points and Galois skew lines. They relate the automorphism $g$ to the Galois group of a projection map from a specific point or subspace. If the projection is Galois with group $\langle g \rangle$ , the quotient is rational.

3. Key Contributions and Results

A. Classification of Automorphism Orders (Theorem 1.1)

The paper provides an explicit list of possible orders for automorphisms $g$ where $\text{codim}(\text{Fix}(g)) \le 2$ . The results depend on the codimension of the fixed locus and the dimension $n$ of the hypersurface:

Codimension 1:
- The order of $g$ must divide $d$ , $d-1$ , or $d-2$ .
- Constraint: If the order divides $d-2$ and is $\ge 3$ , then necessarily $n=2$ .
Codimension 2:
- If $n \ge 4$ : The order divides $(d-1)d$ , $(d-1)^2$ , or $(d-2)d$ .
- If $n = 3$ : The order divides $(d-1)d$ , $(d-1)^2$ , $(d-2)d$ , $d^2-3d+3$ , or $(d-2)(d-1)$ .
- If $n = 2$ : The order divides a more complex set of integers including $(d-1)^2d$ , $(d-1)^3$ , $(d^2-3d+3)d$ , $(d^2-3d+3)(d-1)$ , $(d-2)(d-1)d$ , and $(d-2)(d-1)^2$ .

This refines previous numerical classifications by showing that high-order automorphisms with small fixed loci are only possible in low dimensions or specific algebraic configurations.

B. Rationality of Quotient Spaces (Theorem 1.2)

The authors investigate when the quotient variety $X/\langle g \rangle$ is a rational variety. They establish sufficient conditions based on the order of $g$ and the codimension of the fixed locus:

Case 1: If $\text{ord}(g) \in \{d-2, d-1, d\}$ and $\text{codim}(\text{Fix}(g)) = 1$ , then $X/\langle g \rangle$ is rational.
Case 2: If $\text{ord}(g) = k(d-1)$ or $kd$ (for $k \ge 2$ ) and $\text{codim}(\text{Fix}(g)) = 1$ , then $X/\langle g \rangle$ is rational.

These results generalize previous findings on Galois points (where the fixed locus is a single point) to cases where the fixed locus is a higher-dimensional subvariety.

C. Correction of Previous Literature

The paper identifies and corrects an oversight in a previous result by K. Fujita (Theorem 3.17 in the paper, referencing [8]).

The Error: The previous theorem claimed that if $\text{ord}(g) = k(d-1)$ and $\text{codim}(\text{Fix}(g)) = 2$ , then $X$ has a Galois point.
The Correction: The authors provide a counterexample (Example 3.18) showing that for $n=3$ , an automorphism can have a fixed locus of codimension 2 containing a projective line, yet the hypersurface does not necessarily have a Galois point. They provide a corrected theorem (Theorem 3.19) adding the condition that the fixed locus must not contain a projective line for the Galois point conclusion to hold.

4. Significance

Geometric Refinement: The paper moves beyond purely numerical constraints on automorphism orders, demonstrating that the geometry of the fixed locus is a powerful invariant that drastically narrows the possibilities for automorphism groups.
Dimensional Constraints: It reveals a strong interdependence between the dimension of the ambient space ( $n$ ), the degree of the hypersurface ( $d$ ), and the existence of automorphisms with specific fixed loci. For instance, automorphisms of order dividing $d-2$ with codimension 1 fixed loci are impossible for $n \ge 3$ .
Rationality Criteria: The results provide new, robust criteria for determining the rationality of quotient spaces, which is a fundamental question in birational geometry. This connects the abstract group theory of automorphisms to the concrete geometric property of rationality.
Correction of Theory: By correcting the theorem regarding Galois points in codimension 2, the paper ensures the accuracy of the theoretical framework used to study automorphisms of hypersurfaces in $\mathbb{P}^4$ and higher.

In summary, Hayashi and Suzuki successfully bridge the gap between abstract group classification and geometric reality, showing that the structure of the fixed locus is the key to understanding the symmetries of smooth hypersurfaces.