🔢 mathematics

Naive atoms of blowups: examples

This paper introduces the concept of naive atomic decompositions for smooth projective varieties and demonstrates that they satisfy a naive version of Iritani's blowup formula through several computable yet non-trivial examples.

Original authors: Christian Böhning, Hans-Christian Graf von Bothmer, Zac Su'a

Published 2026-06-17

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Christian Böhning, Hans-Christian Graf von Bothmer, Zac Su'a

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 trying to understand the "DNA" of complex geometric shapes called varieties. In the world of advanced mathematics, these shapes are like intricate sculptures. Mathematicians have developed a way to break these sculptures down into their most fundamental, indivisible pieces, which they call "atoms."

This paper is a computational experiment. The authors, Bohning, von Bothmer, and Su'a, are testing a specific recipe for breaking these shapes down. They call their method "naive atomic decomposition."

Here is a simple breakdown of what they did, using everyday analogies:

1. The Goal: Finding the "Fingerprint" of a Shape

Think of a smooth, projective variety (the mathematical shape) as a complex machine. The authors want to know: What are the basic gears inside this machine?

In the most advanced version of this theory (involving "big quantum cohomology"), finding these gears is like trying to solve a puzzle with a million pieces—it's incredibly hard. So, the authors decided to use a simpler, "naive" version. They look at a specific operator (a mathematical machine that transforms the shape) called the anticanonical bundle.

The Analogy: Imagine you have a complex musical instrument. The "big" theory tries to analyze every single vibration of every string to find the notes. The "naive" theory just taps the instrument with a specific stick (the anticanonical bundle) and listens to the main tones it produces.
The Result: They list the "notes" (eigenvalues) they hear and count how many times each note repeats. This list of counts is their "naive atomic decomposition." It's like a fingerprint for the shape.

2. The Big Question: What happens when you blow up a shape?

In geometry, "blowing up" a shape is like taking a smooth balloon and inflating a tiny bubble on its surface. You are replacing a small point or line with a whole new surface (like a sphere).

The authors wanted to test a famous idea (Iritani's blowup formula) in this simpler "naive" setting. The idea is:

If you take a shape $X$ and blow it up at a smaller shape $Z$ , the new "fingerprint" of the big shape should be a simple mix of the old fingerprint of $X$ and the fingerprint of $Z$ .

It's like saying: If you take a cake ( $X$ ) and add a cherry ( $Z$ ) on top, the new cake's flavor profile should just be the cake's profile plus the cherry's profile.

3. The Experiments: Testing the Recipe

The authors didn't just guess; they did the math on several specific, tricky examples to see if the "mixing recipe" worked.

Example A: The Cubic Fourfold (A 4D Cube-like shape).
They calculated the fingerprint. It turned out to be a specific list of numbers.
Example B: Blowing up a Plane.
They took that 4D shape and "blowed up" a flat plane inside it. They checked if the new fingerprint was just the old one plus the fingerprint of the plane.
- The Surprise: In most math, if you have a "double root" (a note that repeats twice), changing the shape slightly usually breaks that repetition. But here, the repetition stayed. The "double note" survived the explosion. This is a rare and special feature they call a "miracle."
Example C: The Castelnuovo Surface (A tricky test case).
They tried blowing up a 4D space with a very specific, complex surface (a Castelnuovo surface).
- The Result: The recipe failed. The new fingerprint didn't match the simple mix of the old parts.
- The Lesson: This taught them that the recipe only works if the smaller shape fits into the big shape in a very specific, "honest" way. If the geometry is too twisted, the simple mixing rule breaks.

4. The "Mutation" Mystery

In one example (a shape made by intersecting a sphere and a cube), there are two different ways to cut the shape into pieces (called semiorthogonal decompositions). It was a mystery: Are these two ways actually the same, just viewed from different angles?

By looking at their "naive atoms," the authors found that the two ways of cutting the shape actually result in the same set of atoms.

The Analogy: Imagine you have a Lego castle. You can take it apart into "walls and towers" or "windows and doors." The authors found that, at the atomic level, these two different disassembly plans actually result in the exact same pile of bricks. This suggests the two views are indeed equivalent.

5. The Tools: Computers and "Naive" Math

The authors admit they used computers (Macaulay2, SageMath) and even AI (ChatGPT, Gemini) to do the heavy lifting.

The Analogy: Calculating these fingerprints is like trying to count the grains of sand on a beach by hand. It's possible, but you'd go crazy. They used computers to count the grains, and then they carefully checked the AI's work to make sure it didn't make a silly mistake (like counting a grain twice).

Summary

This paper is a "proof of concept" for a simplified way of understanding complex geometric shapes.

They defined a simple way to list the "atoms" of a shape.
They tested if these atoms mix predictably when you modify the shape (blow it up).
They found that sometimes it works perfectly (preserving special repetitions), sometimes it reveals hidden equivalences (solving the mutation mystery), and sometimes it fails (if the shapes don't fit together right).

They aren't claiming this solves every problem in geometry, but they have shown that this "naive" method is a powerful, computable tool that captures the most interesting features of the deeper, more complicated theories.

Technical Summary: Naive Atoms of Blowups: Examples

Problem Statement
The paper addresses the computational complexity inherent in the theory of "atoms" within birational geometry, a framework developed in foundational works such as [KKPY25] and [CKK25]. In this theory, atoms are derived from the spectrum of quantum multiplication by the Euler vector field in the big quantum cohomology. However, computations in the big quantum cohomology are notoriously difficult. The authors aim to investigate whether a "naive" variant of this theory, restricted to the small quantum cohomology and the operator of quantum multiplication by the anticanonical bundle ( $-K_X$ ), retains sufficient structural information to be computable and meaningful. Specifically, the paper seeks to test a naive version of Iritani's blowup formula [Iritani23] in concrete examples, examining whether the "naive atomic decomposition" (the sequence of eigenvalue multiplicities of $(-K_X)^\star$ ) behaves predictably under blowups.

Methodology
The authors define the naive atomic decomposition of a smooth projective Fano variety $X$ as the ordered sequence of multiplicities of the eigenvalues of the operator $(-K_X)^\star$ acting on the algebraic cohomology, evaluated at generic values of the Novikov parameters.

The methodology involves:

Explicit Computation: For a series of specific varieties, the authors construct the matrix $M(X, -K_X)$ $M (X, - K_{X})$ representing quantum multiplication by the anticanonical bundle. This involves:
- Selecting a homogeneous basis for the algebraic cycles $A^\bullet(X)_\mathbb{Q}$ .
- Computing Gromov-Witten invariants (3-point, 2-point, and 1-point) using selection rules, the divisor axiom, and the WDVV equations.
- Utilizing commutativity relations of quantum multiplication by divisor classes to solve for unknown invariants.
- In some cases (Sections 9 and 11), employing quantum periods and differential equations to determine the matrix entries without direct invariant computation.
Limit Analysis: For blowup scenarios $\tilde{X} = \text{Bl}_Z X$ , the authors compute the characteristic polynomial of $(-K_{\tilde{X}})^\star$ . They then perform a specialization procedure, taking limits as specific Novikov parameters approach zero or infinity, to recover the characteristic polynomials of the original variety $X$ and the center $Z$ .
Verification of Conjecture: They test Conjecture 1.2, which posits that if $Z \hookrightarrow X$ induces an embedding of numerical curve classes $N_1(Z, \mathbb{Z}) \hookrightarrow N_1(X, \mathbb{Z})$ , the naive atomic decomposition of the blowup $\tilde{X}$ is the concatenation and sorting of the decomposition of $X$ and $(c-1)$ copies of the decomposition of $Z$ (where $c$ is the codimension of $Z$ ).

Key Contributions and Results

Definition and Scope: The paper formalizes the "naive atomic decomposition" as a computable proxy for the more complex Hodge atoms. It acknowledges that this is a coarser invariant (Remark 2.4) but argues it captures essential numerical data regarding the spectrum of the anticanonical operator.
Computational Examples: The authors provide detailed calculations for:
- Smooth Cubic Fourfolds: The decomposition is $(1+r, 1, 1, 1)$ , where $r$ is the rank of codimension 2 cycles.
- Blowups of Cubic Fourfolds: In cases where a cubic fourfold is blown up along a plane (Section 5) or a line (Section 7), the authors demonstrate that the characteristic polynomial of the blowup center and the original variety can be recovered as limits of the blowup's polynomial. Notably, high multiplicity eigenvalues are preserved under these limits, a phenomenon the authors describe as "remarkable" given that multiple roots are typically destroyed by generic polynomial deformations.
- Blowup of $\mathbb{P}^4$ in a $(2,2,2)$ Complete Intersection: (Section 6) The authors show that the high multiplicity entry of the center is preserved and appears twice in the blowup's decomposition, consistent with the conjecture.
- Picard Rank 1 Fano Threefolds: (Section 10) The authors compute decompositions for all such varieties and find they align perfectly with known semiorthogonal decompositions (SODs) of their derived categories.
- Verra Fourfolds: (Section 11) A Picard rank 2 example that is not a blowup, demonstrating the necessity of geometric input beyond classical intersection theory to determine the quantum matrix.
- Castelnuovo Surface Counterexample: (Section 12) The authors construct a case where the conjecture fails. Blowing up $\mathbb{P}^4$ in a Castelnuovo surface (isomorphic to $\mathbb{P}^2$ blown up in 8 points) yields a decomposition that is not the concatenation of the decompositions of $\mathbb{P}^4$ and the surface. This confirms that the assumption of the embedding $N_1(Z, \mathbb{Z}) \hookrightarrow N_1(X, \mathbb{Z})$ is necessary.
Resolution of Ambiguity: In the case of the $(2,3)$ complete intersection in $\mathbb{P}^5$ (Section 9), the paper addresses the ambiguity of two distinct semiorthogonal decompositions involving spinor bundles. The naive atomic point of view resolves this by grouping the Kuznetsov component and the mutated spinor bundle into a single atom, suggesting a canonical structure.

Significance and Claims
The paper claims to provide "experimental evidence" for the mechanisms behind Iritani's blowup formula and the broader theory of atoms. The authors state that their work highlights "surprising features," specifically the preservation of eigenvalue multiplicities under the specialization from the blowup to the components, which lacks a classical explanation.

The significance is framed modestly:

The results serve as a computational check on the "basic miracle" of Iritani's theorem.
The examples are chosen to be "simple enough to be computable by elementary methods" yet "complicated enough to show most interesting features."
The authors suggest that these data points might aid in the "subsequent simplifications of proofs of foundational results" of the theory.
The paper explicitly notes that while the naive decomposition is coarser than the full Hodge atom structure, it successfully predicts the number of components in semiorthogonal decompositions for the tested Fano threefolds.

The authors do not claim to prove the general conjecture or the full Iritani formula but rather to verify its naive variant in specific, non-trivial cases and to identify the precise geometric conditions (the embedding of curve classes) under which it holds or fails.