Generic flatness of the cohomology of thickenings

This paper establishes a generic flatness result for the cohomology of thickenings of smooth projective schemes over characteristic zero Noetherian domains, while simultaneously demonstrating that for nine points in the projective plane, the associated local cohomology module fails to be generically free and possesses infinitely many associated prime ideals, thereby addressing open questions regarding the constancy of the least degree of hypersurfaces with prescribed multiplicities.

The Gist

Here is an explanation of the paper "Generic Flatness of the Cohomology of Thickenings" using simple language and creative analogies.

The Big Picture: Painting with Layers

Imagine you are an artist working on a canvas (which represents a geometric shape, like a curve or a surface). You have a specific design you want to draw, say a circle.

In this paper, the authors are studying what happens when you don't just draw the circle once, but you draw it over and over again in the exact same spot.

• Layer 1: You draw the circle once.
• Layer 2: You draw it again right on top of the first one. Now the paint is "thicker."
• Layer 3: You draw it a third time. The paint is even thicker.

In math, these "thickened" layers are called thickenings. The paper asks a very specific question: If we change the "recipe" for our paint (the underlying field or ring), does the behavior of these thick layers stay predictable?

The Two Main Characters

The paper has two main goals, which are like two sides of the same coin:

1. The Good News: Smooth Shapes are Predictable

The Scenario: Imagine your shape is a perfectly smooth circle (mathematically, a "smooth scheme").
The Discovery: The authors prove that if your shape is smooth, there is a "safe zone" of recipes where the behavior of the thick layers is stable and predictable.

• The Analogy: Think of a well-oiled machine. No matter how many times you add a layer of paint (as long as you are in the "safe zone"), the machine hums the same tune. The mathematical properties (called cohomology) don't suddenly glitch or change their nature.
• The Result: They found a specific "magic ingredient" (a non-zero number) that, if added to your recipe, guarantees that the math works smoothly for every thickness level, from 1 to infinity.

2. The Bad News: Nine Points in a Plane are Chaotic

The Scenario: Now, imagine instead of a smooth circle, you have nine distinct dots scattered on a piece of paper (the projective plane).
The Mystery: Mathematicians have long wondered: If you pick 9 random dots, is there a "typical" behavior for the thickest possible lines that pass through them?

• The Question: If I pick 9 dots at random, is the "hardest" line I can draw through them always the same difficulty?
• The Discovery: The authors say NO.
• The Analogy: Imagine you are trying to build a bridge through 9 specific islands.
  • If the islands are arranged in a "normal" way, the bridge is easy to build.
  • But, the authors found that for 9 points, the "bridge difficulty" changes wildly depending on the exact position of the points. Sometimes the points align perfectly to make a hard problem; other times they don't.
  • Crucially, there is no single "safe zone" where you can say, "For all thicknesses, the difficulty is constant." The behavior is "erratic."

The "Associated Primes" Mystery

The paper also solves a famous riddle about Associated Primes.

• The Riddle: In algebra, every mathematical object has a list of "prime numbers" (or prime ideals) that describe its structure. A big open question was: Is this list always finite? (i.e., does the object have a limited number of "flavors"?)
• The Old Answer: For many years, people thought yes, the list was always finite.
• The New Answer: The authors constructed a specific example using those nine points where the list of "flavors" is infinite.
• The Metaphor: Imagine a box of crayons. Usually, a box has a fixed number of colors (Red, Blue, Green...). The authors built a "magic box" (based on the nine points) where, as you keep adding layers, new and new colors keep appearing forever. You never run out of new colors; the list is endless.

Why Does This Matter?

1. It connects to a 100-year-old puzzle: The problem of the nine points is related to Hilbert's 14th Problem, a famous challenge in mathematics from the early 1900s. The authors show that the "thickening" of these points is much more complicated than anyone expected.
2. It sets limits: It tells mathematicians, "Don't try to find a simple, universal rule for all points in a plane. Sometimes, the math is just too messy and unpredictable."
3. It uses Elliptic Curves: The "chaos" of the nine points comes from a deep connection to elliptic curves (shapes that look like donuts). The nine points act like "torsion points" on this donut. If the points line up just right (mathematically speaking), they create a ripple effect that makes the math explode into infinite complexity.

Summary in One Sentence

The authors proved that while smooth shapes behave nicely and predictably when you "thicken" them, a specific arrangement of nine points behaves like a chaotic storm, creating an infinite number of mathematical complexities that defy simple rules.

Technical Summary

Here is a detailed technical summary of the paper "Generic Flatness of the Cohomology of Thickenings" by Edoardo Ballico, Yairon Cid-Ruiz, and Anurag K. Singh.

1. Problem Statement and Motivation

The paper addresses two interconnected problems in algebraic geometry and commutative algebra:

1. Generic Flatness of Thickening Cohomology: Given a projective scheme $X$ over a Noetherian domain $A$ (containing a field of characteristic zero), one considers the $t$-th thickening $X_t = V(\mathcal{I}_X^t)$. A classical result (Hartshorne, 1966) establishes that for a smooth $X$ over a field of characteristic zero, the cohomology groups $H^i(X_t, \mathcal{O}_{X_t} \otimes \mathcal{F})$ stabilize for large $t$. The authors investigate the relative version: Does there exist a single dense open subset of $\text{Spec}(A)$ over which the cohomology groups $H^i(X_t, \mathcal{O}_{X_t} \otimes \mathcal{F})$ are flat for all $t \geq 1$ simultaneously? While generic flatness holds for a fixed $t$, the uniformity across all $t$ is non-trivial.

2. The Symbolic Power Problem (Nagata's Question): Related to the above is a classical question regarding points in projective space. Given $m$ distinct points in $\mathbb{P}^n_k$, what is the minimum degree $\alpha_t(X)$ of a hypersurface passing through each point with multiplicity at least $t$?

   • Question 1.2: Does there exist a dense open set of $m$-tuples of points such that $\alpha_t(X)$ is constant for all $t \geq 1$?
   • The authors investigate whether the generic freeness of local cohomology modules of Rees algebras (which would imply a positive answer to Question 1.2) holds universally.

2. Methodology

The authors employ a blend of relative algebraic geometry, homological algebra, and the geometry of elliptic curves.

• Relative Ample Vector Bundles: To handle the base scheme $S$ (where $S = \text{Spec}(A)$), the authors develop a theory of $f$-ample vector bundles for a projective morphism $f: X \to S$. They establish a fiberwise criterion (Theorem 2.2) and prove that the normal bundle $N_X$ of a smooth subscheme $X \subset \mathbb{P}^n_S$ is $f$-ample (Lemma 2.3).
• Uniform Vanishing Theorem: They prove a relative version of Hartshorne's vanishing theorem (Theorem 2.4). Specifically, for a smooth morphism $f: X \to S$ over a field of characteristic zero, there exists a uniform integer $t_0$ such that $H^i(X_s, \text{Sym}^t(N_{X_s}) \otimes \mathcal{F}_s) = 0$ for all $i \neq \dim X$ and $t \geq t_0$, uniformly across all fibers $s \in S$.
• Local Cohomology and Ext Modules: The proof connects the cohomology of thickenings to Ext modules via graded local duality. They utilize the identification:
  $$\varinjlim_t \text{Ext}^i_R(R/I^t, R) \cong H^i_I(R)$$
  They combine the uniform vanishing result with a generic freeness theorem for local cohomology modules (from [CRS24]) to deduce the flatness of the Ext modules, and subsequently the cohomology of the thickenings.
• Elliptic Curve Torsion (Counterexample Construction): To disprove the general case for 9 points in $\mathbb{P}^2$, the authors construct a specific Rees algebra associated with 9 generic points. They analyze the cohomology of the thickenings by restricting to a smooth cubic curve (an elliptic curve) passing through these points. The behavior of the cohomology is linked to the torsion points in the Picard group of the elliptic curve.

3. Key Contributions and Results

A. Positive Result: Smooth Schemes (Theorem A)

The authors prove that if $X \subset \mathbb{P}^n_A$ is smooth over a Noetherian domain $A$ (containing a field of char 0), then the cohomology of all thickenings is generically flat.

• Statement: There exists a non-zero element $a \in A$ such that for all $i \geq 0$ and all $t \geq 1$, the module $H^i(X_t, \mathcal{O}_{X_t} \otimes \mathcal{F}) \otimes_A A_a$ is a flat $A_a$-module.
• Significance: This confirms that for smooth schemes, the "bad" behavior of cohomology (non-flatness) is confined to a closed subset of the base, and this subset is independent of the thickness $t$.

B. Negative Result: Nine Points in $\mathbb{P}^2$ (Theorem B)

The authors show that the generic freeness fails for the specific case of 9 points in the projective plane.

• Construction: Let $A = \mathbb{C}[a_{ij}]$ be the ring of parameters for 9 points in $\mathbb{P}^2$. Let $I$ be the ideal defining these points, and $R(I)$ the Rees algebra.
• Result: The local cohomology module $H^2_{(x_0, x_1, x_2)}(R(I))$ is not generically free over $A$.
• Consequence: The set of associated prime ideals of this module is infinite. Specifically, the union of associated primes over all $t$ is infinite.
• Mechanism: The non-flatness arises because the condition for the cohomology to vanish depends on whether the 9 points form a specific torsion configuration on the unique cubic curve passing through them. Since the torsion points of an elliptic curve are dense but countable, no single dense open set can avoid all "bad" torsion configurations for all $t$ simultaneously.

C. Implications for Symbolic Powers (Corollary 5.6)

The failure of generic freeness for 9 points implies a negative answer to Question 1.2 for this case.

• Conclusion: There is no dense open subset $U \subset (\mathbb{P}^2_{\mathbb{C}})^9$ such that the function $S \mapsto \alpha_t(X_S)$ is constant for all $t \geq 1$.
• Interpretation: The symbolic powers of the ideal of 9 points exhibit "erratic" behavior; the minimal degree of hypersurfaces with multiplicity $t$ fluctuates depending on the specific geometric configuration of the points, even within a general family.

4. Significance

1. Resolution of a Relative Flatness Question: The paper settles the question of whether cohomology of thickenings is uniformly flat over a base for smooth schemes, providing a robust affirmative answer (Theorem A).
2. Counterexample to Generic Constancy: It provides a definitive counterexample to the hope that invariants of point configurations (like the Waldschmidt constant or minimal degree $\alpha_t$) are constant on a dense open set for all $t$. The case of 9 points in $\mathbb{P}^2$ is shown to be inherently unstable due to the arithmetic of elliptic curves.
3. Infinite Associated Primes: The construction yields a local cohomology module with infinitely many associated primes. This contributes to the long-standing problem (posed by Huneke) regarding the finiteness of associated primes of local cohomology modules. While finiteness is known for regular rings in many cases, this example shows that for Rees algebras of point ideals, the set of associated primes can be infinite.
4. Geometric Insight: The paper elegantly bridges commutative algebra (flatness, Rees algebras) with the geometry of elliptic curves and torsion points, demonstrating how arithmetic properties of the base field and the geometry of the underlying curve dictate the asymptotic behavior of algebraic invariants.

In summary, the paper establishes a strong positive result for smooth schemes while revealing a deep, intrinsic obstruction to uniformity in the case of 9 points in the plane, driven by the torsion structure of elliptic curves.