🔢 mathematics

Curves on complete intersections and measures of irrationality

This paper establishes that the degree of any curve on a general complete intersection of large multidegree is bounded below by the degree of the intersection itself, thereby resolving a specific problem regarding measures of irrationality posed by Bastianelli et al.

Original authors: Nathan Chen, Benjamin Church, Junyan Zhao

Published 2026-07-13

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Nathan Chen, Benjamin Church, Junyan Zhao

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 have a giant, multi-layered cake made of different flavors of frosting. In the world of math, this cake is a shape called a "complete intersection," sitting inside a huge, multi-dimensional space. Now, imagine you want to find the simplest possible path you can draw on the surface of this cake. Maybe you want to draw a line, or a loop, or a squiggly curve.

For a long time, mathematicians wondered: Are the simplest paths just the ones you get by slicing the cake with a flat knife? Or, could there be some sneaky, twisted, super-simple path hiding somewhere that is actually "simpler" (in terms of how many times it winds or how complex it is) than the straight knife cuts?

In this paper, Nathan Chen, Benjamin Church, and Junyan Zhao (with help from Mohan Swaminathan) act like detectives investigating this cake. They prove that for cakes made with thick enough layers of frosting (mathematicians call these "large multidegrees"), there are no sneaky shortcuts that are simpler than the standard slices. While they don't prove that the slices are the only simple paths, they prove that you cannot find a path that is "simpler" (has a lower degree) than the ones you get by slicing the cake with a flat plane.

The Big Discovery: No Sneaky Shortcuts

The authors prove a specific rule about the "degree" of any curve on these shapes. Think of the "degree" as a score for how complicated a path is. A straight line has a low score; a wild, twisting spiral has a high score.

They show that if your cake is made of layers with thicknesses $d_1, d_2, \dots, d_r$ (and these numbers are big enough—specifically, at least $2n - 1$ where $n$ is the number of dimensions of the cake), then any path you draw on the cake must have a score of at least:
$(d_1 - 2n + 2) \times (d_2 - 2n + 2) \times \dots \times (d_r - 2n + 2)$

If the layers are really thick (bigger than a specific huge number $N$ ), the rule becomes even stricter: the score must be at least the product of all the layer thicknesses ( $d_1 \times d_2 \times \dots \times d_r$ ).

What does this mean? It means you can't cheat. You can't find a path that is "simpler" than the standard slices. If you try to draw a path that looks like it should be easy, the math proves it's actually just as complicated as slicing the whole cake with a knife.

The "Irrationality" Score: How Weird is Your Shape?

The paper also tackles a concept called "measures of irrationality." Imagine you have a shape, and you want to know how hard it is to turn it into a simple, flat sheet (like a piece of paper) without tearing it.

If a shape is "rational," it's easy to flatten.
If it's "irrational," it's stubborn and hard to flatten.

The "degree of irrationality" is a number that tells you how many times you have to fold or stretch the shape to make it look like a flat sheet. The authors answer a big question: How irrational are these multi-layered cakes?

They prove that for these cakes, the "irrationality score" is huge. It's almost as big as the total volume of the cake (the product of all the $d$ 's). This confirms a guess made by other mathematicians (Bastianelli, De Poi, Ein, Lazarsfeld, and Ullery) that the complexity of these shapes multiplies together rather than just adding up.

How They Solved the Mystery: The Magic of Breaking Things

How did they prove this? They didn't just stare at the cake; they broke it.

They used a technique called "degeneration." Imagine taking your perfect, smooth cake and slowly melting it until it falls apart into two smaller cakes that are stuck together at the edge.

The Setup: They took a curve (a path) on the original cake and watched what happened as the cake broke.
The Break: When the cake split, the path had to split too. It couldn't just stay whole on one piece; it had to break into pieces that covered both sides of the split.
The Logic: By studying how the path broke, they could use a "domino effect" (mathematicians call this induction). They showed that if the rule holds for the smaller broken pieces, it must hold for the big original cake.

They also used a clever trick involving "stable maps," which is like tracking a rubber band stretched over a shape. If you shrink the shape, the rubber band has to snap or rearrange itself in a very specific way. They proved that no matter how the rubber band rearranges, it can't become "simpler" than the standard slices.

What They Ruled Out

The paper explicitly argues against the idea that there are "sneaky" simple paths hiding on these shapes that are simpler than the slices.

No Hidden Simplicity: They prove that you cannot find a curve with a degree lower than the product of the layer thicknesses (once the layers are thick enough).
No Shortcuts: They show that the "simplest" curves are at least as complicated as the ones you get by slicing with linear subspaces (flat planes).

Note: While they prove the lower bound, they do not prove that the slices are the only curves with this minimal degree. In fact, they explicitly list it as an open question whether the minimal degree curves are exclusively the linear slices.

How Sure Are They?

The authors are 100% sure about the lower bounds they established. They didn't run computer simulations or guess based on patterns. They provided a rigorous mathematical proof.

They proved that for "general" complete intersections (which means "most" of them, or the typical ones you'd pick at random), the rules hold.
They proved that for "very general" ones (an even stricter category), the rules hold even more strongly.
They even calculated a specific number $N$ (which is huge, involving powers of 227 and factorials) that guarantees the rule works if the layers are thicker than that.

However, regarding the uniqueness of these paths (i.e., are the slices the only minimal paths?), the paper leaves this as an open question for future investigation.

The Takeaway

If you have a complex, multi-layered mathematical shape, don't look for a shortcut that is simpler than the standard slices. The most complicated-looking slices (the ones that cut through all the layers) are the simplest paths you can draw, in terms of degree. The complexity of the shape is locked in, and you can't bypass it. The authors have shown that for these specific shapes, you cannot find a path with a lower degree than the direct, flat cut, though they leave open the possibility that other paths might share this same minimal complexity.

Technical Summary: Curves on Complete Intersections and Measures of Irrationality

Problem Statement
The paper investigates the minimal degrees and gonalities of curves contained within general complete intersection varieties $X \subseteq \mathbb{P}^{n+r}$ . Specifically, it addresses two central questions regarding the complexity of such varieties:

Degrees of Curves: Are the curves obtained by slicing $X$ with linear subspaces the "simplest" curves (in terms of degree) on $X$ ? The authors aim to verify a folklore conjecture that for complete intersections of sufficiently large multidegree, the degree of any curve is bounded below by the degree of the variety itself.
Measures of Irrationality: How do the degree of irrationality ($irr(X)$) and the covering gonality ($cov. gon(X)$) behave for general complete intersections? While previous work established additive lower bounds, it was conjectured that these invariants should behave multiplicatively with respect to the degrees of the defining polynomials ( $d_1, \dots, d_r$ ).

Methodology
The authors employ a combination of degeneration techniques, moduli space theory, and multiplier ideal machinery.

Degeneration and Stable Maps: The core of the proof for degree bounds (Theorem A) involves degenerating a general complete intersection $X$ into a union of two complete intersections $X_1 \cup X_2$ meeting transversally along a divisor $Z$ . By analyzing the deformation of stable maps from the generic fiber to the special fiber, the authors utilize a "Breaking Lemma" (Lemma 2.9). This lemma, derived from Jun Li's relative stable maps formalism, establishes that if a curve deforms to the generic fiber, its specialization must either cover the intersection $Z$ or break into components covering $X_1$ and $X_2$ with matching multiplicities along $Z$ . This allows for a simultaneous induction on dimension, codimension, and degree.
Grassmannian Techniques: To extend results from covering degrees to the degrees of arbitrary curves, the paper adapts techniques from Reidl and Yang [RY22]. This involves analyzing the locus of points on the universal complete intersection through which pass curves of a specific degree, utilizing the constructibility of Hilbert schemes to show that low-degree curves cannot cover the general fiber.
Multiplier Ideals and Point Separation: To establish bounds on covering gonality (Theorem B), the authors utilize the positivity of the canonical bundle. They develop a technical strengthening of the Angehrn–Siu theorem (Theorem C) regarding the separation of points by adjoint linear series $|K_X + dH|$ . This involves constructing effective $\mathbb{Q}$ -divisors with high multiplicities at general points and using Nadel vanishing theorems. A key innovation is optimizing the reduction of the dimension of the non-klt locus (LC-locus) to separate asymptotically large numbers of points, yielding a bound with an error term of order $O(\sqrt{d})$ rather than the $O(1)$ term available for surfaces.

Key Contributions and Results

Theorem A (Degree Bounds): The authors prove that for a general complete intersection $X \subseteq \mathbb{P}^{n+r}$ of dimension $n$ cut out by polynomials of degrees $d_1, \dots, d_r \geq 2n-1$ , any curve $C \subseteq X$ satisfies:
$\deg(C) \geq (d_1 - 2n + 2) \cdots (d_r - 2n + 2).$
Furthermore, they show that for sufficiently large degrees ( $d_i \geq N(n,r)$ ), the degree of any curve is bounded below by the degree of the variety itself: $\deg(C) \geq d_1 \cdots d_r$ . This confirms the folklore conjecture that linear slices provide the minimal degree curves for high-degree complete intersections.
Theorem B (Covering Gonality): The paper resolves the problem of measures of irrationality for general complete intersections by establishing asymptotically sharp multiplicative bounds. For a general complete intersection $X$ of dimension $n$ and degrees $d_i \geq n$ :
$cov. gon(X) \geq (d_1 - 2(n+1)\sqrt{d_1})(d_2 - n + 1) \cdots (d_r - n + 1) + 1.$
Consequently, for any $\epsilon > 0$ , if the degrees are sufficiently large, $cov. gon(X) \geq (1-\epsilon) d_1 \cdots d_r$ . Notably, this result holds for general complete intersections, not just very general ones, implying the bounds hold for almost all complete intersections defined over $\mathbb{Q}$ .
Theorem C (Point Separation): A technical result of independent interest, this theorem provides a lower bound on the number of points separated by the linear series $|K_X + dH|$ on a polarized normal Gorenstein variety. If any curve meeting an open set $U$ has degree at least $\alpha$ , then $|K_X + dH|$ separates at least $(d - \delta\sqrt{d})\cdot \alpha$ distinct points on $U$ . This improves upon previous reduction steps in the literature by optimizing the separation of asymptotically large numbers of points.
Applications:
- Genus Bounds: Theorem A implies a lower bound on the geometric genus of any curve in a general complete intersection, which approaches the genus of a linear slice for large degrees.
- Measures of Association: The results yield lower bounds on the correspondence degree between two general complete intersections, quantifying how distinct their birational types are.
- Generalizations: The degeneration method is shown to be independent of the specific geometry of projective space, allowing for extensions to complete intersections in arbitrary ambient spaces formed by intersecting different ample classes.

Significance
The paper claims to make a significant step toward showing that linear slices are indeed the simplest curves on general complete intersections of large multidegree. By verifying two open conjectures, the work provides a complete picture of the asymptotic behavior of the degree of irrationality and covering gonality for general complete intersections. The authors emphasize that their results resolve the multiplicative nature of these invariants, which had previously only been known to satisfy additive bounds or sharp multiplicative bounds only in specific low-dimensional or highly restricted cases. The technical refinement of the Angehrn–Siu separation argument (Theorem C) is highlighted as a crucial ingredient that may be of independent interest for studying measures of irrationality in broader contexts.