🔢 mathematics

Ulrich sheaves and determinantal representations for higher secant varieties of curves

This paper establishes that higher secant varieties of sufficiently ample smooth projective curves admit symmetric admissible determinantal representations with rank-one Ulrich sheaves, enabling the spectrahedral representation of their convex hulls and providing explicit formulas for rational normal curves via higher Szegő kernels and Scorza correspondences.

Original authors: Daniele Agostini, Mario Kummer, Jinhyung Park

Published 2026-07-27

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Daniele Agostini, Mario Kummer, Jinhyung Park

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 a world where shapes aren't just drawn on paper but are built from invisible mathematical rules. In the branch of math called algebraic geometry, scientists study these shapes, which are often defined by equations. One of the most fascinating puzzles in this field involves "determinantal representations." Think of a complex shape as a giant, multi-layered cake. A determinantal representation is like finding a special recipe—a specific matrix (a grid of numbers) made of simple linear ingredients—that, when you mix them together and take a specific calculation (the determinant), recreates the exact shape of the cake. If you can find this recipe, you can understand the shape's hidden structure.

Why does this matter? Because these recipes are the keys to understanding "convex hulls"—the tightest possible rubber band you could wrap around a shape. In the real world, this connects to optimization, which is the science of finding the best solution to a problem, like the most efficient route for a delivery truck or the safest design for a bridge. If a shape is "hyperbolic," it behaves nicely, and finding its recipe allows computers to solve problems inside it incredibly fast. For decades, mathematicians have wondered: Can we find these special recipes for every kind of shape? And if the shape is made of real-world numbers (not just imaginary ones), can we make the recipe "positive definite," meaning it acts like a perfect, stable bowl that holds everything securely?

This paper, titled "Ulrich Sheaves and Determinantal Representations for Higher Secant Varieties of Curves," dives deep into a specific family of shapes called "secant varieties" of curves. Imagine a smooth, winding curve (like a snake or a ribbon) floating in space. The "secant variety" is the cloud of all the straight lines that connect any two, three, or more points on that curve. As you add more points to connect, the cloud gets bigger and more complex. The authors, Daniele Agostini, Mario Kummer, and Jinhyung Park, tackle a big question: Can we find these special, symmetric recipes for these complex clouds, especially when the original curve is embedded in a high-dimensional space?

The paper proves that yes, we can! Under certain conditions (specifically, when the curve is "sufficiently ample," meaning it's stretched out enough in space), these complex clouds of lines always have a "rank one" recipe. In math-speak, this means the recipe is as simple as it can possibly be. Even better, if the curve is a "rational normal curve" (a very specific, clean type of curve), the authors found a way to write down the recipe explicitly using "Littlewood–Richardson coefficients." These are numbers from a different branch of math (combinatorics) that count how to combine shapes, acting like a secret code that unlocks the recipe for the cloud.

But the story gets even cooler when we look at real-world curves. The authors show that if the original curve is "real" (made of real numbers) and has a specific property called being "vastly real" (meaning its lines don't wander off into imaginary territory too much), then the recipe for the cloud is not just a recipe, but a "definite" one. This means the matrix in the recipe acts like a perfect, positive bowl. This is a huge deal because it proves that the "convex hull" of these shapes (the tightest rubber band around them) is a "spectrahedron." In plain English, this means the shape can be described by a simple set of rules that computers love, allowing for rapid and efficient optimization. The paper doesn't just suggest this; it provides a rigorous proof and explicit formulas, turning a theoretical possibility into a concrete tool for understanding the geometry of the universe.

Technical Summary: Ulrich Sheaves and Determinantal Representations for Higher Secant Varieties of Curves

Problem Statement
The paper addresses the existence and construction of rank-one Ulrich sheaves and their corresponding symmetric admissible determinantal representations for higher secant varieties of smooth projective curves. While it is known that every projective variety supports an Ulrich sheaf (under mild conditions), the minimal rank of such a sheaf is a central question in algebraic geometry (Eisenbud–Schreyer). Specifically, the authors investigate whether the $k$ -th secant variety $\Sigma_k$ of a curve $C \subseteq \mathbb{P}^r$ admits a rank-one Ulrich sheaf, which would imply the existence of a determinantal representation of size equal to the degree of $\Sigma_k$ . Furthermore, the paper explores the real algebraic geometry of these varieties, seeking conditions under which these representations are definite, thereby characterizing the convex hulls of real curves as spectrahedra.

Methodology
The authors employ a synthesis of classical algebraic geometry, representation theory, and real algebraic geometry. The core methodology involves:

Higher Szegő Kernels and Scorza Correspondences: The construction relies on the $(k+1)$ -Szegő kernel associated with a line bundle $\alpha$ on $C$ having no cohomology ( $h^0(\alpha) = h^1(\alpha) = 0$ ). This kernel defines a section on the product of symmetric products $C_{k+1} \times C_{k+1}$ , whose zero locus is the higher Scorza correspondence.
Gaussian Maps and Tautological Bundles: The authors utilize higher-order Gaussian maps (generalizing the Wahl map) and first-order tautological bundles on symmetric products to relate the geometry of the curve to the secant variety.
Admissible Determinantal Representations: Building on the work of Shamovich and Vinnikov, the paper constructs linear maps $\gamma: \wedge^{n+1} V \to \text{Hom}(U, U)$ that serve as determinantal representations. Theorem A provides a technical criterion for a linear map to yield an admissible determinantal representation of rank one based on the behavior of fibers of linear projections.
Cohomological Vanishing: To prove the existence of Ulrich sheaves for general curves, the authors use cohomological vanishing theorems on symmetric products of curves, specifically leveraging the properties of $k$ -very ample embeddings.
Littlewood–Richardson Coefficients: For rational normal curves, the construction is made explicit using the coproduct of symmetric functions and Littlewood–Richardson coefficients.

Key Contributions and Results

Theorem A (Construction of Representations): The paper establishes a general criterion (Theorem A) for constructing rank-one admissible determinantal representations and their associated Ulrich sheaves. It links the isometry class of the resulting quadratic form to the $\mathbb{A}^1$ -degree of the linear projection, provided the projection has fibers consisting of rational points.
Theorem B (Existence for Secant Varieties): The main result proves that for a smooth projective curve $C$ $C$ embedded via a $k$ $k$ -very ample line bundle $L$ $L$ (with $r \ge 2k+2$ $r \geq 2 k + 2$ ), the $k$ $k$ -th secant variety $\Sigma_k$ $Σ_{k}$ admits an admissible determinantal representation of rank one.
- The associated Ulrich sheaf is of the form $A_{k, L \otimes \alpha}$ , where $\alpha$ is a line bundle with no cohomology.
- If $\alpha$ is a theta-characteristic, the representation is symmetric.
- This generalizes previous results known for $g=0$ or $k=1$ to arbitrary genus and higher secant varieties.
Theorem C (Explicit Representations for Rational Normal Curves): For rational normal curves, the authors derive explicit symmetric determinantal representations. The entries of the representing matrix are linear forms indexed by partitions, with coefficients given by Littlewood–Richardson coefficients. This provides a concrete combinatorial description of the Chow form of the secant variety.
Theorem D (Real Hyperbolicity and Spectrahedra): In the context of real algebraic curves, the paper establishes an equivalence between three conditions for a linear projection $\pi: \Sigma_k \to \mathbb{P}^{2k+1}$ $π : Σ_{k} \to P^{2 k + 1}$ :
1. The projection is real-fibered (the preimage of real points consists only of real points).
2. The defining linear system is "vastly real" (every divisor in the system has at most $2k$ non-real points).
3. There exists an admissible determinantal representation that is definite (positive or negative) at the center of the projection.
- Consequently, if the convex hull of a real curve is a hyperbolicity cone, it can be represented as a spectrahedron (the intersection of the cone of positive semidefinite matrices with an affine subspace).

Significance and Claims
The paper claims to resolve the question of the existence of rank-one Ulrich sheaves for higher secant varieties of curves under $k$ -very ample embeddings. By providing explicit constructions (via Szegő kernels) and cohomological proofs, it extends the theory of determinantal representations from curves and their first secants to higher secant varieties.

The significance of the real algebraic geometry results lies in answering Question (4) from prior work [20] affirmatively: the convex hull of a real curve (when it is a hyperbolicity cone) is a spectrahedron. This is achieved by showing that the defining polynomial of the secant variety admits a definite determinantal representation. The authors note that their constructions are explicit, allowing for the computation of minimal free resolutions and the representing matrices, as demonstrated in examples involving rational normal curves and real curves of genus two.

The paper remains modest regarding arithmetic geometry, noting that while the assumption of fully rational fibers in Theorem A could likely be removed, the current focus is on the geometric and real-algebraic applications where this assumption holds. The work also connects the geometry of secant varieties to the theory of hyperbolic polynomials and semidefinite programming, offering a theoretical foundation for optimization over convex hulls of algebraic curves.

More like this