🔢 mathematics

Disconnected multigraded Hilbert schemes on $\mathbb{P}^2\times\mathbb{P}^1$

This paper establishes the existence of an infinite family of disconnected multigraded Hilbert schemes on the biprojective space $\mathbb{P}^2 \times \mathbb{P}^1$ , demonstrating that even standard bigradings on polynomial rings in five variables can yield disconnected Haiman-Sturmfels multigraded Hilbert schemes.

Original authors: Yairon Cid-Ruiz

Published 2026-08-11

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Yairon Cid-Ruiz

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 the mathematical world as a giant, infinite playground where shapes and spaces are built from invisible Lego bricks. In this playground, mathematicians have a special tool called a "Hilbert scheme." Think of a Hilbert scheme not as a single building, but as a massive, magical map. On this map, every single dot represents a different possible shape you can build using a specific set of rules and a specific number of bricks. If you have a map where all the dots are connected by roads, it means you can smoothly transform one shape into another without the shape ever breaking apart or vanishing. For a long time, mathematicians believed that for a certain type of space (called projective space), this map was always one big, connected island. You could start with a simple cube and, by slowly twisting and stretching the bricks, turn it into a complex sphere without ever losing your place on the map.

However, this paper explores a slightly more complicated version of that playground: a space made by combining two different projective worlds, like stacking a 3D grid on top of a 2D grid. The question the author asks is simple but profound: Is the map for these combined spaces still one big, connected island? Or are there hidden cliffs and chasms that split the map into separate, unreachable islands? This matters because if the map is disconnected, it means there are shapes that look similar on paper but are fundamentally impossible to turn into one another, no matter how hard you try to stretch them. It reveals a hidden "fracture" in the geometry of the universe.

The Discovery: A Map with a Chasm

In this paper, the author, Yairon Cid-Ruiz, proves that the answer is a resounding "no." The map is not always one big island. In fact, the author has found an infinite family of these maps that are broken into at least two separate pieces.

To understand how this works, imagine you are trying to build a specific type of sculpture using a recipe that tells you exactly how many bricks of different colors you need. The "recipe" in this math world is called a "Hilbert polynomial." It's like a nutritional label for a shape, telling you its size and complexity. The author focuses on a specific recipe, which we can call the "pa-recipe" (defined by the formula $2az_1 + az_2 + 3a - 2a^2$ , where $a$ is a number at least 2).

The author shows that when you use this recipe to build shapes on the combined space ( $\mathbb{P}^2 \times \mathbb{P}^1$ ), you end up with two very different kinds of sculptures that cannot be connected to each other.

The Two Islands

The first island on the map is filled with sculptures that are "complete intersections." Imagine these as sculptures built by taking two giant, transparent sheets of glass (defined by equations) and looking only at where they cross each other. These shapes are very orderly and predictable. The author proves that all these orderly shapes form a safe, connected neighborhood on the map. You can walk from any one of these glass-crossing shapes to another without trouble.

The second island is filled with a very different kind of sculpture. These are built using a "glitchy" recipe where the bricks are arranged in a way that doesn't look like a simple crossing of two sheets. The author constructs a specific example of this (using a set of equations like $x_0^{2a}, x_0y_0, x_1^ay_0$ ) that acts like a stranger in town.

Here is the kicker: The author proves that this "glitchy" sculpture has a specific property—it has no "hidden lines" of a certain type (mathematically, the space of sections of its ideal sheaf is zero). The orderly, glass-crossing sculptures, on the other hand, always have at least one such hidden line. Because of this fundamental difference, the "glitchy" sculpture cannot be smoothly transformed into any of the "orderly" ones. If you tried to morph one into the other, the shape would have to snap or break, which is not allowed in this mathematical world.

The Proof and the Result

The paper doesn't just guess this; it proves it rigorously. By using a tool called "cohomology" (which is like a mathematical X-ray that checks for hidden structural features), the author shows that the set of orderly shapes is both "open" and "closed." In map terms, this means the neighborhood of orderly shapes is a self-contained bubble. You can't walk out of it without falling off the edge of the map. Since the "glitchy" sculpture exists on the map but isn't inside that bubble, the map must be split into at least two separate pieces.

This discovery is significant because it shatters the hope that these maps are always connected. The author shows that for any integer $a \ge 2$ , you can create a disconnected map. For example, if you pick $a=2$ , you get a map for a specific recipe that is definitely broken.

Why It Matters (Even for a Teenager)

You might wonder, "So what? It's just math about shapes." But this is like discovering that the rules of physics change depending on where you are in the universe. It turns out that the "smoothness" of geometry isn't guaranteed everywhere. The paper also mentions that this result has a side effect: it proves that there are infinitely many disconnected maps even for a system with just five variables (like a polynomial ring with five letters). This means the "fracture" is a common feature, not a rare accident.

The author even used a computer (specifically, AI tools like OpenAI Codex) to help find the first example of this broken map (when $a=2$ ), but the heavy lifting of the proof—the logic that explains why it's broken and how to prove it for all other numbers—was done by the human mathematician. The paper concludes that the world of these multigraded Hilbert schemes is full of surprises, with hidden chasms separating shapes that might look similar at first glance. The map is not a single island; it's an archipelago, and some of the islands can never be reached from others.

Technical Summary: Disconnected Multigraded Hilbert Schemes on $\mathbb{P}^2 \times \mathbb{P}^1$

Problem Statement
The paper addresses the fundamental question of the connectedness of multigraded Hilbert schemes. While Hartshorne's classical result establishes that the standard Hilbert scheme $\text{Hilb}_p(\mathbb{P}^n_k)$ is always connected for a numerical polynomial $p$ , the situation for multigraded Hilbert schemes $\text{Hilb}_p(P)$ , where $P$ is a product of projective spaces, has remained open. Previous work by Haiman–Sturmfels, Peeva–Stillman, and Santos provided examples of disconnected Hilbert schemes for affine semigroup gradings or fixed Hilbert functions, but it was unknown whether disconnectedness could occur for the multigraded Hilbert scheme of a product of projective spaces $P = \mathbb{P}^{n_1}_k \times \cdots \times \mathbb{P}^{n_m}_k$ with a multigraded Hilbert polynomial $p \in \mathbb{Q}[z_1, \dots, z_m]$ .

Methodology
The author constructs an infinite family of counterexamples on the biprojective space $P = \mathbb{P}^2_k \times_k \mathbb{P}^1_k$ . The methodology relies on the following steps:

Polynomial Selection: For an integer $a \geqslant 2$ , the author defines a specific multigraded Hilbert polynomial:
$p_a(z_1, z_2) = 2az_1 + az_2 + 3a - 2a^2$
This polynomial corresponds to the Hilbert polynomial of complete intersections of bidegrees $(1,0)$ and $(a, 2a)$ .
Cohomological Characterization: The core of the proof involves analyzing the locus $Z_a \subset \text{Hilb}_{p_a}(P)$ consisting of subschemes $Y$ that are complete intersections of the specified bidegrees. Using Proposition 2.2, the author establishes that for any closed subscheme $Y$ with Hilbert polynomial $p_a$ , the condition that $Y$ is such a complete intersection is equivalent to the cohomological condition $h^0(Y, \mathcal{O}_Y(1,0)) = 2$ (or equivalently, $H^0(P, \mathcal{I}_Y(1,0)) \neq 0$ ).
Topological Analysis of the Locus: By applying the upper semicontinuity theorem to the universal family over the Hilbert scheme, the author demonstrates that the locus $Z_a$ is both open and closed in $\text{Hilb}_{p_a}(P)$ .
- Openness: Follows from the condition $h^0(Y, \mathcal{O}_Y(1,0)) \leqslant 2$ being an open condition.
- Closedness: Follows from the condition $H^0(P, \mathcal{I}_Y(1,0)) \neq 0$ being a closed condition.
Construction of Disconnected Components:
- Component 1: The author identifies a point $[Y_{1,a}]$ in $Z_a$ corresponding to the complete intersection $V(x_0, x_1^a y_0^{2a})$ .
- Component 2: The author constructs a distinct point $[Y_{2,a}]$ defined by the ideal $I_{2,a} = (x_0^{2a}, x_0y_0, x_1^a y_0)$ . Through an exact sequence calculation, it is shown that $Y_{2,a}$ possesses the correct Hilbert polynomial $p_a$ . However, $H^0(P, \mathcal{I}_{Y_{2,a}}(1,0)) = 0$ , implying $[Y_{2,a}] \notin Z_a$ .

Key Contributions and Results

Theorem A: For any field $k$ and integer $a \geqslant 2$ , the multigraded Hilbert scheme $\text{Hilb}_{p_a}(\mathbb{P}^2_k \times_k \mathbb{P}^1_k)$ is disconnected. Specifically, the locus of complete intersections $Z_a$ forms a non-empty, proper, open, and closed subset, separating the scheme into at least two connected components.
Corollary 2.8: As a consequence, there exist infinitely many disconnected Haiman–Sturmfels multigraded Hilbert schemes for a standard bigrading on a polynomial ring in five variables ( $k[x_0, x_1, x_2, y_0, y_1]$ ).
Distinction of Points: The paper identifies two specific "bilex" points (in the sense of [ACDN00]) corresponding to ideals $I_{1,a}$ and $I_{2,a}$ that lie in distinct connected components. Furthermore, Remark 2.7 notes that these points have different cohomological properties regarding the Segre embedding $\mathcal{O}_P(1,1)$ , specifically $h^0(Y_{1,a}, \mathcal{O}_{Y_{1,a}}(1,1)) = 4$ versus $h^0(Y_{2,a}, \mathcal{O}_{Y_{2,a}}(1,1)) = a+3$ , confirming they do not belong to the same fiber-full scheme.

Significance
The paper resolves the open question regarding the connectedness of multigraded Hilbert schemes on products of projective spaces in the negative. It demonstrates that, unlike the classical projective case, the geometry of multigraded Hilbert schemes can be inherently disconnected even for smooth projective toric varieties like $\mathbb{P}^2 \times \mathbb{P}^1$ . The construction relies on controlling cohomological loci rather than combinatorial arguments, highlighting a distinct feature of the multiprojective setting where the complete intersection locus is both open and closed, a phenomenon not observed in the classical projective setting.

More like this