🔢 mathematics

Nonnegative Polynomials and Moment Problems on Algebraic Curves

This paper establishes a duality between the faces of the cone of nonnegative polynomials on a smooth real projective curve and totally real effective divisors, enabling a complete description of the face lattice for genus one curves, the computation of Carathéodory numbers for elliptic normal curves, and the derivation of sharp bounds for truncated moment problems on affine cubic curves.

Original authors: Lorenzo Baldi, Grigoriy Blekherman, Rainer Sinn

Published 2026-03-02

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Lorenzo Baldi, Grigoriy Blekherman, Rainer Sinn

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 chef trying to bake the perfect cake. In the world of mathematics, specifically Real Algebraic Geometry, the "cake" is a polynomial (a fancy math equation), and the "perfect" version is one that is nonnegative. This means that no matter what numbers you plug into the equation, the result is never negative—it's either zero or positive, like a cake that never tastes bitter.

For a long time, mathematicians knew how to prove a polynomial was nonnegative (using "sums of squares," like proving a cake is sweet by showing it's made of sugar). But they didn't really understand the shape of the collection of all these nonnegative polynomials. Is it a smooth sphere? A jagged star? A weird, twisted blob?

This paper, by Baldi, Blekherman, and Sinn, is like a team of cartographers finally mapping out the hidden terrain of these mathematical shapes. They focus on a specific, beautiful landscape: curves (think of them as lines drawn on a piece of paper, but in higher dimensions).

Here is the breakdown of their discovery using everyday analogies:

1. The Map and the Landmarks (Faces and Divisors)

Imagine the collection of all nonnegative polynomials as a giant, multi-dimensional mountain range.

The Mountain Range: This is the "cone" of nonnegative polynomials.
The Faces: The flat sides, ridges, and peaks of this mountain are called "faces."
The Landmarks: The authors discovered a secret code. Every flat side (face) of this mountain corresponds perfectly to a specific arrangement of points on the curve, which they call a divisor.

The Analogy: Think of the mountain range as a giant sculpture. The authors found that if you look at a specific flat side of the sculpture, you can instantly tell exactly which "dots" on the ground (the curve) are touching it. It's like having a map where every cliff face points directly to a specific village. This connection allows them to describe the entire shape of the mountain just by looking at the villages (the points) on the ground.

2. The Special Case: The Elliptic Curve (The Donut)

The paper focuses heavily on curves with genus one. In simple terms, these are curves that look like a donut (or a torus).

The Shape Matters: A donut can be drawn in two ways on a real piece of paper:
1. One Loop: A single, connected ring (like a standard donut).
2. Two Loops: Two separate rings floating near each other (like a figure-eight or two distinct circles).

The authors found that the "shape" of the nonnegative polynomial mountain depends entirely on whether the underlying donut has one loop or two.

3. The "Carathéodory Number" (The Minimum Ingredients)

This is one of the most practical parts of the paper. Imagine you want to recreate a specific flavor (a mathematical "moment") using a mix of ingredients (points on the curve).

The Question: What is the minimum number of ingredients you need to mix to get any possible flavor?
The Discovery:
- If your curve is one connected loop, you need exactly $n+1$ ingredients.
- If your curve is two disconnected loops, you suddenly need $n+2$ ingredients.

The Analogy: It's like baking. If your kitchen is one big open room (connected), you can make any cake with a standard set of 5 tools. But if your kitchen is split into two separate rooms (disconnected), you suddenly need a 6th tool to reach the ingredients in the second room. The "topology" (the layout) of the curve changes the complexity of the math required to describe it.

4. The Affine Moment Problem (The Real-World Application)

The paper then moves from the perfect, abstract "projective" world to the messy "affine" world (the real world we live in, where things can go off to infinity).

The Problem: Can we tell if a set of data comes from a real physical distribution of mass on a curve?
The Old Way: Mathematicians used a test called "flat extension." It's like checking if a puzzle piece fits perfectly.
The New Discovery: For curves with two loops, the "perfect fit" test sometimes fails, even if the data is real. The authors introduced a new test called "almost flat extension."
The Metaphor: Imagine trying to close a suitcase.
- Flat Extension: The lid closes perfectly flush with the suitcase.
- Almost Flat Extension: The lid is slightly ajar (by just a tiny bit), but you know for a fact that if you push it one more inch, it will close.
- The authors proved that for certain complex curves (the two-loop ones), you must use the "almost flat" test to know if your data is valid. If you only look for the "perfectly flat" closure, you might throw away valid data.

Summary of the "Big Picture"

This paper is a breakthrough because it moves from asking "Is this polynomial positive?" to "What does the shape of all positive polynomials look like?"

They mapped the shape: They showed that the geometry of these shapes is dictated by the points where the polynomials hit zero.
They found a surprise: The number of "ingredients" needed to describe these shapes changes based on whether the curve is one piece or two pieces.
They fixed a tool: They updated the tools used by engineers and statisticians to analyze data on curves, showing that for complex shapes, you need a slightly more flexible test ("almost flat") to avoid missing the truth.

In short, they took a rigid, abstract mathematical object and showed that its behavior is surprisingly sensitive to the "shape" of the world it lives in, providing a new, more accurate way to measure and understand it.

1. Problem Statement

The paper addresses two fundamental and interconnected problems in real algebraic geometry and functional analysis:

Facial Structure of Nonnegative Polynomials: While the cone of nonnegative polynomials ( $P_{X,2d}$ ) on a variety $X$ is central to the field, its facial structure (specifically the extreme rays and the lattice of faces) is poorly understood outside of classical cases (e.g., univariate polynomials, ternary quartics). The authors aim to systematically describe this structure for smooth real projective curves.
The Truncated Moment Problem: Given a linear functional $L$ on polynomials of degree $\le 2d$ , determine if $L$ can be represented as an integration against a measure supported on the real locus $X(\mathbb{R})$ . This is equivalent to determining if $L$ lies in the dual moment cone $P^\vee_{X,2d}$ . The authors seek to characterize membership in this cone, specifically focusing on the Carathéodory number (the minimal number of point evaluations required to represent any functional in the cone) and the conditions for flat extensions (algebraic certificates for the moment problem).

2. Methodology

The authors employ a synthesis of convex geometry, algebraic geometry (specifically the theory of divisors and Jacobians), and real algebraic geometry.

Duality between Faces and Divisors: The core methodological innovation is establishing a correspondence between the faces of the cone of nonnegative quadratic forms ( $P_{X,2}$ $P_{X, 2}$ ) and totally real effective divisors on the curve $X$ $X$ .
- They define a map $\Phi$ from a face $F$ to a divisor $\text{div}(F)$ (derived from the common zeroes of forms in the relative interior of $F$ ).
- They define a map $\Psi$ from a divisor $D$ to a face $F_D$ (forms vanishing with multiplicity at least $2D$ ).
- They prove these maps form a Galois connection, reducing the study of the face lattice to the study of which divisors arise as "face divisors."
Jacobian and Torsion Points: To characterize the extreme rays (1-dimensional faces), the authors utilize the Jacobian variety $J(X)$ of the curve. They map extreme rays to 2-torsion points in the real part of the Jacobian, $J(\mathbb{R})_2$ . Specifically, they distinguish between "positive" 2-torsion points (those admitting nonnegative rational functions) and non-positive ones.
Projective vs. Affine Transition: The authors first solve the problem in the projective setting (using Veronese embeddings to reduce degree $2d$ to quadratic forms). They then apply these projective results to the affine setting (plane cubic curves) by analyzing the behavior of points at infinity and the topology of the real locus.
Flat and Almost Flat Extensions: For the moment problem, they utilize the flat extension criterion (where the rank of the moment matrix stabilizes). They introduce a novel concept, the almost flat extension, where the rank is allowed to increase by exactly one, to handle cases where standard flat extensions fail due to topological constraints.

3. Key Contributions and Results

A. Facial Structure of Nonnegative Cones

Theorem A (Galois Connection): Establishes a duality between the poset of positive-dimensional faces of $P_{X,2}$ and the set of totally real effective divisors. A face $F$ is uniquely determined by its associated divisor $\text{div}(F)$ .
Theorem B (Extreme Rays and Jacobian): For a smooth totally real curve of genus $g$ $g$ , the set of extreme rays spanned by forms with only real zeroes corresponds bijectively to the positive 2-torsion points of the Jacobian, denoted $J(\mathbb{R})^+_2 \cong (\mathbb{Z}/2\mathbb{Z})^g$ $J (R)_{2}^{+} ≅ (Z /2 Z)^{g}$ .
- Significance: This proves the existence of nonnegative polynomials that are not sums of squares (SOS) for curves of genus $g \ge 1$ and sufficiently high degree, as SOS extreme rays correspond only to the zero element of the Jacobian.
Theorem 5.1.2 (Genus One Case): Provides a complete description of the face lattice for elliptic normal curves (genus 1). The faces are classified by the degree of the associated divisor and whether the corresponding 2-torsion point is the identity ( $O$ $O$ ) or the unique non-trivial positive 2-torsion point ( $T$ $T$ ).
- If $[D-H] = O$ , the extreme ray is a sum of squares.
- If $[D-H] = T$ , the extreme ray is nonnegative but not a sum of squares.

B. Carathéodory Numbers

Theorem C (Topology Dependence): The Carathéodory number $C_{X,2}$ $C_{X, 2}$ of the dual moment cone for an elliptic normal curve depends strictly on the topology of the real locus $X(\mathbb{R})$ $X (R)$ :
- If $X(\mathbb{R})$ is connected: $C_{X,2} = \deg(X) = n+1$ .
- If $X(\mathbb{R})$ is disconnected: $C_{X,2} = \deg(X) + 1 = n+2$ .
- Significance: This is the first complete description of Carathéodory numbers for non-rational curves and demonstrates that the topology of the support set fundamentally alters the complexity of quadrature rules.

C. Truncated Moment Problem on Affine Cubics

Theorem D (Almost Flat Extension): For smooth affine plane cubic curves, a linear functional $L$ is a moment functional if and only if it admits an almost flat extension (an extension to degree $2d+4$ where the rank increases by at most 1).
Theorem 7.2.2 & 7.2.4 (Refined Bounds):
- If $X(\mathbb{R})$ is connected and has one point at infinity, the standard flat extension (degree $2d+2$ ) suffices, and the affine Carathéodory number equals the projective one ( $3d$ ).
- If $X(\mathbb{R})$ is disconnected (or has multiple points at infinity), the standard flat extension is insufficient. The affine Carathéodory number jumps to $3d+1$ , necessitating the almost flat extension condition.
- This resolves the moment problem for irreducible, non-rational curves, a case previously unsolved.

4. Significance and Impact

Resolution of Open Problems: The paper provides the first complete description of the facial structure of nonnegative cones for non-rational curves (specifically genus 1), moving beyond Hilbert's classical cases.
Topological Sensitivity: It rigorously demonstrates that the Carathéodory number is not merely a function of the degree or dimension but is sensitive to the connectedness of the real locus. This has implications for numerical optimization and quadrature design on algebraic curves.
New Certificates for Positivity: By linking extreme rays to 2-torsion points in the Jacobian, the authors provide a geometric mechanism to construct nonnegative polynomials that are not sums of squares, offering explicit certificates (via Artin-type representations) for these forms.
Advancement of Moment Problems: The introduction of "almost flat extensions" provides a necessary and sufficient condition for the truncated moment problem on affine cubics, correcting previous assumptions that flat extensions were always sufficient. This bridges the gap between projective theory and affine analysis.
Methodological Bridge: The work successfully bridges convex geometry (faces of cones) with complex algebraic geometry (divisors, Jacobians, Riemann-Roch), offering a template for studying positivity on higher-dimensional varieties.

In summary, this paper fundamentally advances the understanding of nonnegative polynomials on curves by revealing a deep connection between the geometry of the curve (genus, topology, Jacobian torsion) and the convex geometry of the associated polynomial cones, with direct applications to the solvability of moment problems.