🔢 mathematics

On the Codimension-1 $\mathrm{PGL}_4$ Orbit Closures in $\mathrm{Gr}(2,10)$

This paper investigates the codimension-1 orbit closures of the natural $\mathrm{PGL}_4$ action on the Grassmannian of pencils of quadrics in $\mathbb{P}^3$ , constructing a family of such orbits via the $j$ -invariant of the discriminant binary quartic and explicitly computing their Chow classes.

Original authors: Ari Krishna

Published 2026-03-31

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Ari Krishna

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 working in a vast, 16-dimensional universe called The Grassmannian. In this universe, every single point represents a specific "pencil" of shapes.

To make this concrete, let's shrink the universe down to something you can visualize: 3D space (like your living room).

A Quadric is a 3D shape defined by a quadratic equation. Think of it as a sphere, a cube, a saddle, or a cone.
A Pencil is a "family" of these shapes created by mixing two specific shapes together. Imagine taking a sphere and a cube and blending them with a slider. As you slide from 0 to 1, you get a smooth transition of shapes.
The Grassmannian is the giant library where every possible "slider" (every possible pencil) has its own shelf.

The Main Character: The PGL Group

Now, imagine a giant, invisible hand (mathematicians call this PGL(V)) that can stretch, shrink, rotate, and shear the entire 3D room.

If you take a pencil of shapes and stretch the room, the pencil changes, but it's still "essentially" the same family of shapes, just viewed from a different angle.
The mathematicians want to know: If we let this hand shake the room as much as possible, how many unique "families" of pencils are left?

The Problem: A Missing Dimension

The library has 16 dimensions. The hand has 15 dimensions of movement.

If you have 15 ways to move things around in a 16-dimensional room, you can't reach every single point. You will always be stuck on a "wall" or a "sheet" that is one dimension smaller than the whole room.
This paper is about mapping out that 15-dimensional wall (which is called a "codimension-one orbit closure").

The Secret Code: The Discriminant Quartic

How do we tell one family of pencils apart from another?
Every pencil has a "fingerprint" hidden inside it. If you look at the shapes in the pencil, some of them will be "broken" or singular (like a cone that has collapsed into a point).

The paper shows that the pattern of these broken shapes forms a quartic equation (a polynomial with four roots).
Think of this like a DNA strand with four specific markers.
The "hand" (PGL) can rotate the room, but it cannot change the relative distances between these four markers. It can only stretch the whole DNA strand.

The J-Invariant: The Universal ID Card

Since the hand can't change the relative distances, mathematicians use a special number called the j-invariant to label these families.

Imagine the j-invariant is a serial number or a zip code.
If two pencils have the same j-invariant, they are "twins" (they belong to the same orbit).
The paper proves that if you pick a random j-invariant, the set of all pencils with that number forms a perfect, smooth 15-dimensional sheet.

The Journey: Counting the Sheets

The author, Ari Krishna, does two main things:

Counting the Sheets (The General Case):
He asks: "If I pick a random j-invariant, how 'thick' is that sheet?"
Using a clever trick involving Schubert calculus (which is like a high-level game of counting intersections in geometry), he calculates that these sheets have a specific "weight" or "class."
- Analogy: Imagine the library is a giant cake. The author is slicing it. He proves that for most slices, the slice has a specific size (mathematically, $12\sigma_1$ ).
The Special Slices (The Edge Cases):
Not all slices are the same. Sometimes the "DNA" (the quartic) gets weird.
- The "Nodal" Slice: There is one special slice where the family of shapes degenerates into a shape with a single "knot" or "node" (like a figure-8). This is the boundary divisor. The author proves this knot is the only major boundary of the library.
- The "Super-Symmetric" Slices: There are two special numbers (0 and 1728) where the DNA has extra symmetry.
  - At 1728, the slice is "folded" twice. It looks like one sheet, but mathematically it counts as two layers.
  - At 0, the slice is "folded" three times. It counts as three layers.

The Big Reveal

The paper concludes with a map of this 16-dimensional universe:

Most families are smooth sheets with a standard weight of 12.
The "Knot" family (the boundary) is also a sheet with a weight of 12.
The "Super-Symmetric" families are special:
- The one at 1728 is actually a double-layered sheet (weight 6 per layer).
- The one at 0 is a triple-layered sheet (weight 4 per layer).

Why Does This Matter?

This isn't just about counting shapes. It's about understanding the geometry of the universe of shapes.

It tells us exactly how these families of shapes fit together.
It connects the abstract world of high-dimensional algebra to the concrete world of 3D geometry (like the intersection of two spheres).
It solves a puzzle that was similar to one solved for 2D shapes (conics), but much harder because we are now dealing with 3D shapes (quadrics).

In a nutshell: The author built a map of a 16-dimensional room filled with families of 3D shapes. He proved that almost all families are grouped by a single number (the j-invariant), calculated the "size" of these groups, and identified the one special "knot" shape that acts as the edge of the room.

1. Problem Statement

The paper investigates the natural action of the projective linear group $PGL(V)$ on the Grassmannian $G = \text{Gr}(2, \text{Sym}^2 V^\vee)$ , where $V$ is a 4-dimensional vector space over an algebraically closed field of characteristic 0.

Geometric Context: Points in $G$ correspond to 2-dimensional linear systems of quadrics in $\mathbb{P}^3$ (i.e., pencils of quadrics).
Dimensional Analysis: $\dim(G) = 16$ and $\dim(PGL(V)) = 15$ . Consequently, the generic orbit has codimension 1 in $G$ .
Objective: The author aims to explicitly construct the one-parameter family of generic orbits, classify the codimension-1 orbit closures, and compute their Chow classes (divisor classes) in the Chow ring $A_1(G)$ .

2. Methodology

The author employs a combination of classical invariant theory, Schubert calculus, and intersection theory to solve the problem.

A. The Discriminant and the $j$ -Map

Discriminant Quartic: For a pencil $W = \langle Q_0, Q_1 \rangle$ , the discriminant is defined as the binary quartic $\Delta_W(s, t) = \det(sQ_0 + tQ_1)$ . The roots of $\Delta_W$ correspond to the singular members of the pencil.
Orbit Classification: On the open locus $U_{sm}$ where $\Delta_W$ has four distinct roots (smooth pencils), the $PGL(V)$-orbit is determined entirely by the projective equivalence class of the root divisor on $\mathbb{P}^1$ . This class is parameterized by the $j$ -invariant of the binary quartic.
The Rational Map: This defines a rational map $p: G \dashrightarrow \mathbb{P}^1$ (the $j$ -map), where the fiber over a value $a$ corresponds to the set of pencils with a specific $j$ -invariant.

B. Intersection Theory and Schubert Calculus

To compute the Chow class of a fiber $F_a = p^{-1}(a)$ , the author uses the following strategy:

Schubert Basis: The Chow ring $A_1(G)$ is generated by the special Schubert class $\sigma_1$ (the hyperplane class under the Plücker embedding). Thus, any divisor class is of the form $m\sigma_1$ .
Complementary Cycle: To find the coefficient $m$ , the author intersects a general fiber with a complementary Schubert cycle $\sigma_{8,7}$ (representing lines in a fixed plane $\Pi \subset \mathbb{P}^9$ passing through a fixed point $q$ ).
Reduction to Plane Geometry: The intersection of the determinant hypersurface $D$ (singular quadrics) with the plane $\Pi$ yields a smooth plane quartic curve $C$ . The problem reduces to counting the number of lines through $q$ tangent to $C$ .
Degree Calculation: Using the Riemann-Hurwitz formula and the geometry of the $j$ -map restricted to this slice, the degree of the map is computed.

C. Analysis of Special Fibers

The author analyzes the boundary behavior where the discriminant has repeated roots:

Generic Tangency: The locus where the pencil is tangent to the determinant hypersurface $D$ at a smooth point.
Higher Codimension: Proving that other degenerate cases (e.g., tangency at singular points of $D$ , or higher-order contact) form loci of codimension $\ge 2$ and thus do not contribute to the divisor class.

3. Key Contributions and Results

A. Classification of Orbit Closures

The paper establishes that the codimension-1 orbit closures are the reduced fibers of the $j$ -map:

Generic Fibers ( $a \notin \{0, 1728, \infty\}$ ): The reduced fiber $O_a$ is a single irreducible orbit closure.
Special Fibers (CM Values):
- At $j=1728$ , the scheme-theoretic fiber is non-reduced: $F_{1728} = 2 O_{1728}$ .
- At $j=0$ , the scheme-theoretic fiber is non-reduced: $F_0 = 3 O_0$ .
The Boundary Divisor ( $j=\infty$ ): The fiber over $\infty$ $\infty$ corresponds to pencils with repeated roots. The unique divisorial component is the closure of the orbit of a specific nodal pencil ( $W_{node}$ $W_{n o d e}$ ).
- Geometric Description: The complete intersection defined by $W_{node}$ is a quartic curve in $\mathbb{P}^3$ with arithmetic genus 1 and exactly one ordinary node (geometric genus 0).
- Normal Form: The generic element of this orbit is equivalent to the pencil $\langle x_0^2+x_1^2+x_2^2, x_0x_3 + x_1^2 + 2x_2^2 \rangle$ .

B. Computation of Chow Classes

The main theorem provides the explicit classes of these orbit closures in $A_1(G)$ :

General Fiber Class: Every divisorial fiber of the rational map $p$ has the class $12\sigma_1$ .
Reduced Orbit Classes:
- For generic $a$ : $[O_a] = 12\sigma_1$ .
- For $j=1728$ : $[O_{1728}] = 6\sigma_1$ (since the fiber has multiplicity 2).
- For $j=0$ : $[O_0] = 4\sigma_1$ (since the fiber has multiplicity 3).
- For the nodal boundary $T$ : $[T] = 12\sigma_1$ .

C. Technical Proofs

Degree of the $j$ -map: The author proves that the restriction of the $j$ -map to a Schubert slice has degree 12. This is derived from the fact that a general point in the plane of a smooth quartic curve has exactly 12 tangent lines (via the Plücker formula or Riemann-Hurwitz).
Multiplicity Analysis: By analyzing the ramification of the Legendre $j$ -function ( $j(\lambda)$ ), the author confirms the multiplicities of 2 and 3 at the CM values.
Uniqueness of the Boundary Component: It is proven that the locus of "bad" repeated roots (e.g., tangency at singular points of the discriminant) has codimension $\ge 2$ , ensuring that the nodal divisor $T$ is the unique divisorial component of the special fiber.

4. Significance

Extension of Previous Work: This work parallels and extends Choudhary's computations for $PGL_3$ acting on nets of conics ( $\text{Gr}(3, V_2)$ ). However, it highlights a crucial geometric difference: in the $PGL_4$ case, the special fiber has a unique divisorial component, whereas in the $PGL_3$ case, it splits into two components with multiplicities.
Explicit Geometry: The paper provides a concrete geometric realization of the "bad" locus (the nodal complete intersection) and calculates its precise contribution to the Chow ring.
Moduli Interpretation: The results clarify the relationship between the geometry of pencils of quadrics in $\mathbb{P}^3$ and the moduli space of elliptic curves (via the $j$ -invariant), specifically detailing how the boundary of the moduli space corresponds to specific degenerations of the quadric intersection.
Foundation for Higher Codimension: The paper sets the stage for studying higher-codimension orbit closures (e.g., root types $2+2$ , $3+1$ , $4$) and their stratification within the repeated-root locus.

In summary, the paper successfully constructs the moduli of generic pencils of quadrics, identifies the specific geometric nature of the boundary divisors, and computes their precise intersection-theoretic classes, resolving the structure of codimension-1 orbits in this Grassmannian.