DT-GV correspondence on the Mukai-Umemura variety

This paper computes Donaldson-Thomas invariants and their descendants for the local Calabi-Yau 4-fold over the Mukai-Umemura variety using localization formulas, thereby verifying the predictions of Cao, Maulik, and Toda under the assumption that genus-one Gopakumar-Vafa invariants vanish.

The Gist

Imagine you are trying to count the number of specific shapes hidden inside a very complex, multi-dimensional sculpture. In the world of advanced mathematics, this sculpture is called a Calabi-Yau 4-fold, and the shapes we are looking for are curves (like loops or lines) drawn on it.

This paper, written by Kiryong Chung and Joonyeong Won, is like a high-stakes detective story where the authors are trying to solve a mystery about how to count these curves accurately. They are testing a "Grand Theory" proposed by other mathematicians (Cao, Maulik, and Toda) that claims there are two different ways to count these curves, and both ways should give the exact same answer.

Here is the breakdown of their adventure, translated into everyday language:

1. The Two Ways to Count (The Detective's Tools)

The authors are comparing two different "counting methods" for these hidden curves:

• Method A: The "DT" Count (The Architect's View). This method is like counting the curves by looking at the blueprints and the structural stability of the sculpture. It involves complex math called "Donaldson-Thomas invariants." It's very precise but can be messy to calculate.
• Method B: The "GV" Count (The Traveler's View). This method, named after Gopakumar and Vafa, is like counting the curves by imagining them as physical travelers moving through the sculpture. It's often simpler to think about, but it relies on a specific assumption: that there are no "looping" travelers (genus-1 curves) that get stuck in a circle.

The Big Question: Do these two methods always agree? The "Grand Theory" says YES. The authors set out to prove this for a very specific, tricky sculpture.

2. The Tricky Sculpture: The Mukai-Umemura Variety

The authors didn't just pick any random shape. They chose a very special, rare sculpture called the Mukai-Umemura variety.

• Think of this as a "Golden Ticket" shape. It's a 3D object that is perfectly symmetrical and has a unique property: it's the only one of its kind that admits a specific type of rotation (an $SL_2(\mathbb{C})$ action).
• Because it's so special, it's also very hard to study. Previous mathematicians had successfully tested the "Grand Theory" on this shape for small, simple curves (like lines of length 1, 2, or 3).
• The Challenge: The authors wanted to test it for curves of length 4. This is like trying to count a tangled knot of string rather than a straight line. The math gets much harder because the "knots" (curves) can twist and overlap in complicated ways.

3. The Strategy: Freezing the Motion (Localization)

How do you count shapes in a spinning, complex sculpture? The authors used a clever trick called Localization.

Imagine the sculpture is spinning rapidly. If you try to count the curves while it's spinning, it's a blur. But, if you could magically "freeze" the sculpture at the exact moments where the spinning stops (the fixed points), the problem becomes much easier.

• At these frozen points, the complex curves break down into simple, straight lines or small loops that are easy to analyze.
• The authors spent a lot of time mapping out these "frozen points" and figuring out exactly what the curves look like there. They found that most complicated curves actually "disappear" from the count because they are too unstable (mathematically speaking, they have a "zero-weight" component that cancels them out).

4. The "Ghost" Curves and the Real Heroes

In their analysis, they discovered something fascinating:

• The Ghosts: Many potential curves looked like they should contribute to the count, but when they did the math, they turned out to be "ghosts"—they contributed nothing because of a hidden symmetry.
• The Heroes: The only curves that actually mattered for the final count were a specific type of "multiplicity-4" curve. Think of this as a single line that is wrapped around itself four times, creating a very thick, heavy rope.

The authors had to do some heavy lifting to calculate exactly how much this "thick rope" contributed to the total count. They used a computer program (Macaulay2) to help untangle the algebraic knots in the equations.

5. The Grand Finale: The Theory Holds!

After all the calculations, they compared their "Architect's Count" (DT) with the "Traveler's Count" (GV).

• The Result: The numbers matched perfectly!
• The Catch: They had to assume that the "looping travelers" (genus-1 curves) don't exist for this specific shape. Based on other research, this is a very safe assumption (like assuming there are no invisible ghosts in the room).
• The Conclusion: The "Grand Theory" is correct, even for these complex, knotted curves of length 4 on the Mukai-Umemura variety.

Summary Analogy

Imagine you are trying to verify a recipe for a perfect cake.

• The Theory says: "If you mix ingredients A and B, you get a cake that tastes exactly like mixing ingredients C and D."
• Previous tests showed this worked for small cakes (1, 2, and 3 layers).
• This paper tackles the 4-layer cake, which is known to be unstable and prone to collapsing.
• The Authors used a special technique (freezing the mixing bowl) to isolate the ingredients. They found that most of the messy ingredients cancel each other out, leaving only one specific, heavy ingredient to do the work.
• The Verdict: They baked the 4-layer cake, tasted it, and confirmed: Yes, the recipe works! The two different ways of mixing the ingredients produce the exact same delicious result.

This paper is a significant step forward because it proves that these deep mathematical connections hold up even when the shapes get complicated and "knotted," giving mathematicians more confidence in the underlying laws of the universe they are studying.

Technical Summary

Here is a detailed technical summary of the paper "DT-GV Correspondence on the Mukai-Umemura Variety" by Kiryong Chung and Joonyeong Won.

1. Problem Statement

The paper addresses the Donaldson–Thomas (DT) / Gopakumar–Vafa (GV) correspondence for Calabi–Yau 4-folds. Specifically, it investigates the relationship between DT invariants (counting stable sheaves) and GV-type invariants (counting curves) on a specific non-compact Calabi–Yau 4-fold:
$$Y = \text{Tot}(K_X)$$
where $X$ is the Mukai–Umemura variety, the unique smooth prime Fano threefold of genus 12 admitting a non-trivial $SL_2(\mathbb{C})$-action.

The core problem is to verify Conjecture 1.1 (proposed by Cao, Maulik, and Toda), which predicts a precise linear identity relating:

• Primary DT invariants $\langle \tau_0(\gamma_0) \rangle_\beta$ to genus 0 GV invariants $n_{0,\beta}(\gamma_0)$.
• Descendant DT invariants $\langle \tau_1(\gamma_1) \rangle_\beta$ to genus 0 GV invariants, meeting invariants $m_{\beta_1, \beta_2}$, and genus 1 GV invariants $n_{1,\beta}$.

The authors focus on the curve class $\beta = d[\text{line}]$ for degrees $d$ up to 4. While the conjecture was previously verified for $d \leq 3$, the case $d=4$ presents significant geometric complexity due to the structure of stable sheaves and the virtual cycle.

2. Methodology

The authors employ a rigorous torus localization approach to compute the invariants. The methodology proceeds through the following steps:

A. Geometric Setup and Fixed Points

• The Variety: $X$ is realized as a zero locus of a regular section of $(\wedge^2 S^*)^{\oplus 3}$ in the Grassmannian $Gr(3, 7)$.
• Torus Action: The $SL_2(\mathbb{C})$-action on $X$ induces a $\mathbb{C}^*$-action. The authors analyze the fixed locus of this action, identifying four fixed points ($p_{\pm 12}, p_{\pm 10}$) and classifying all $\mathbb{C}^*$-fixed irreducible rational curves of degree $d \leq 4$ (lines $L_{\pm 2}$, a conic $Q$, and a quartic $C_4$).
• Local Charts: They construct explicit local affine charts around the fixed points to determine the weights of the tangent spaces and the defining equations of the fixed curves.

B. Moduli Space Analysis

• The moduli space of stable sheaves $M_\beta(Y)$ is isomorphic to the moduli space of stable sheaves on $X$, $M_\beta(X)$.
• For $d=4$, the fixed locus of the moduli space consists of structure sheaves of specific Cohen-Macaulay (CM) curves. Crucially, the authors analyze multiple curves (curves with multiplicity) supported on the fixed lines.
• They construct a filtration of CM subcurves $D_1 \subset D_2 \subset D_3 \subset D_4$ supported on the line $L_2$, where $D_k$ has multiplicity $k$.

C. Equivariant Computations

Using Virtual Localization (Graf, Pandharipande, Thomas) and K-theoretic localization (Thomason), the authors compute:

1. Equivariant Euler Characteristics: They calculate $\chi(\mathcal{O}_C, \mathcal{O}_C)$ for all relevant fixed curves (irreducible and reducible, including the multiple curves $D_k$). This involves computing the deformation and obstruction spaces ($Ext^1$ and $Ext^2$) as representations of the torus.
   • Key Technical Insight: For any fixed curve other than the multiplicity-4 line ($D_4$), the obstruction space contains a weight-zero component. Consequently, the equivariant Euler class of the virtual tangent space vanishes for these curves, meaning they do not contribute to the localization formula.
2. Insertion Integrals: They compute the integrals of primary ($\tau_0$) and descendant ($\tau_1, \tau_2$) insertions over the virtual class $[M_4]^{vir}$. The calculation relies on the non-vanishing contributions solely from the multiplicity-4 curves $D_4$ and $D_{-4}$.

D. Verification of the Conjecture

• They compute the DT invariants $\langle \tau_i(h_{2-i}) \rangle_4$ explicitly.
• They assume the genus 1 GV invariants vanish ($n_{1,d} = 0$) for $d \leq 4$, a reasonable assumption given the non-existence of degree $d$ elliptic curves in $Y$ for these degrees.
• They calculate the meeting invariants $m_{\beta_1, \beta_2}$ recursively using the definitions provided by Klemm and Pandharipande.
• Finally, they substitute these values into the linear identity (Equation 2) of Conjecture 1.1 to verify equality.

3. Key Contributions

1. Extension to Degree 4: The paper extends the verification of the DT-GV correspondence from degree $d \leq 3$ to $d=4$ for the Mukai-Umemura variety. This is a non-trivial extension because the moduli space structure becomes significantly more complex at $d=4$.
2. Analysis of Multiple Curves: A major contribution is the detailed analysis of the multiplicity-4 curve $D_4$. The authors explicitly construct the CM filtration of $D_4$ and compute its equivariant Euler characteristic, showing it is the only fixed locus contributing to the invariants at this degree.
3. Vanishing Mechanism: The paper identifies a specific geometric phenomenon: for all fixed curves except the highest multiplicity ones, the obstruction space has a weight-zero component. This causes their contributions to the localization formula to vanish, drastically simplifying the computation.
4. Explicit Invariant Values: The authors provide explicit values for the primary and descendant DT invariants for $d=4$:
   • $\langle \tau_0(h_2) \rangle_4 = 168$
   • $\langle \tau_1(h_1) \rangle_4 = -168$
   • $\langle \tau_2(1) \rangle_4 = -28$

4. Results

• Theorem 1.2 (Theorem 3.16): Assuming $n_{1,d} = 0$ for $1 \leq d \leq 4$, the DT-GV correspondence (Conjecture 1.1) holds for the curve class$\beta = 4[\text{line}]$on$Y = \text{Tot}(K_X)$.
• Verification: The computed DT invariants satisfy the linear identity:
  $$\langle \tau_1(h_1) \rangle_4 = \frac{11 n_{0,4}(h_2)}{4} - \frac{1}{16}(6 m_{1,3} + 4 m_{2,2})$$
  where the meeting invariants $m_{1,3}$ and $m_{2,2}$ are calculated to be $-1008$.
• Ratio Consistency: The paper confirms that the ratio $\frac{\langle \tau_2(h_0) \rangle_d}{\langle \tau_0(h_2) \rangle_d} = -\frac{1}{6}$ holds for all $1 \leq d \leq 4$, verifying a prediction from Cao-Toda.

5. Significance

• Validation of Conjectures: This work provides strong evidence for the Cao-Maulik-Toda conjectures in the context of Calabi–Yau 4-folds, specifically for a variety with rich symmetry (Mukai-Umemura).
• Technical Advancement: The methods developed for handling non-reduced moduli spaces and multiple curves (specifically the filtration of $D_4$) offer a blueprint for studying higher-degree invariants on other local Fano varieties.
• Geometric Insight: The discovery that lower-degree or lower-multiplicity fixed curves vanish from the localization sum due to weight-zero obstructions highlights a deep structural property of the virtual cycle in this setting, suggesting that the "hard" geometry is concentrated in the most singular/multiple components of the moduli space.

In summary, the paper successfully bridges the gap between DT and GV theories for a challenging degree ($d=4$) by combining sophisticated localization techniques with a deep analysis of the moduli space's fixed-point structure.