🔢 mathematics

Towards a characterization of toric hyperkähler varieties among symplectic singularities II

This paper confirms that any conical symplectic variety of dimension $2n$ with a projective symplectic resolution and an effective Hamiltonian action of an $n$ -dimensional torus is equivariantly isomorphic to a toric hyperkähler variety, thereby affirmatively answering a question posed in the authors' previous work.

Original authors: Yoshinori Namikawa

Published 2026-03-31

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Yoshinori Namikawa

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 an architect trying to understand a very strange, complex building. This building is a Toric Hyperkähler Variety. In the world of mathematics, this is a special kind of geometric shape that has two very important features:

It has a "symplectic" structure (think of this as a perfect, frictionless dance floor where every move has a specific energy).
It has a "torus" symmetry (think of it as a building that can be spun or rotated by a group of $n$ different rings, like a complex carousel).

The Problem: Two Different Maps

In the first part of this research (Part I), the mathematician Yoshinori Namikawa proved that if you have this special building, you can draw a map to a "standard model" building (called $Y(A, 0)$ ). This map is like a translation guide that says, "This weird room in your building is exactly the same as this standard room in the model."

However, there was a catch. The original building had a "scaling" feature (a $C^*$ -action). Imagine this as a magical zoom button that can make the whole building grow or shrink.

The Standard Model has a very clean, algebraic zoom button (it works like a precise mathematical formula).
The Original Building had a zoom button that worked, but it was "complex analytic." In math-speak, this means it was a bit messy, like a zoom that works perfectly in a dream but is hard to write down with a pen and paper.

The big question (Question 5.10) was: Can we fix the Original Building so that its zoom button matches the Standard Model's clean, algebraic zoom button?

The Solution: The "Renovation" Strategy

In this paper (Part II), Namikawa says, "Yes, we can!" But we can't just force the building to change. We have to do a clever renovation.

Here is the step-by-step analogy of how he does it:

1. The Blueprint and the Open Floor Plan

First, he looks at a "resolution" of the building. Imagine the original building has some sharp, jagged corners. He smooths them out to get a perfect, clean version called $\tilde{X}$ .
He then looks at a specific open area of this smooth building where the "carousel" (the torus action) works perfectly. In this area, the building looks like a stack of identical rooms arranged neatly.

2. The Two Zoom Buttons

On this open area, he identifies two different ways to zoom:

Zoom A (The Original): The way the building was originally designed.
Zoom B (The Imported Way): The way the Standard Model zooms, which he "imports" into the building using his map from Part I.

He notices that both Zoom A and Zoom B do the same thing to the "energy" of the building (they both scale the symplectic form by the same amount). However, they move the rooms slightly differently.

3. The "Ghost" Shift

He realizes that the difference between Zoom A and Zoom B is actually just a shift caused by the carousel itself.

Analogy: Imagine you are walking on a moving walkway at an airport (Zoom A). Someone else is walking on a parallel moving walkway that moves at the same speed but starts slightly ahead (Zoom B). The difference between the two isn't a change in speed; it's just that one walkway is shifted relative to the other by a specific amount.

Namikawa proves that this "shift" is actually a smooth, continuous movement generated by the carousel's own rotation.

4. The Renovation (Moser's Trick)

Now comes the magic trick (called Moser's Trick in math).
He constructs a "renovation crew" (a mathematical transformation called $\phi$ ). This crew doesn't change the shape of the rooms or the energy of the dance floor. Instead, they gently slide the rooms around while the carousel is spinning.

The Goal: They slide the rooms just enough so that when you press "Zoom A" (the original), it feels exactly like pressing "Zoom B" (the imported standard).
The Result: After the renovation, the building's internal zoom button is now perfectly aligned with the Standard Model's clean, algebraic zoom button.

The Final Outcome

Once this renovation is done, he combines the original map with the renovation crew's work.

Before: The map was a "dream map" (complex analytic).
After: The new map is a "blueprint map" (algebraic).

This means the strange building is not just similar to the standard model; it is algebraically identical to it. You can now describe the entire building using simple, precise equations, just like the standard model.

Why Does This Matter?

In the world of math, "algebraic" is the gold standard. It means the object is rigid, well-defined, and easy to compute with. "Complex analytic" is more flexible but harder to pin down.

By proving that these special varieties are always algebraic (once you pick the right zoom setting), Namikawa has shown that these complex, high-dimensional shapes are actually much more orderly and "tame" than we thought. He has essentially taken a mysterious, abstract sculpture and revealed that it is built from the same Lego bricks as the standard models we already know.

In short: He found a way to "tune" a complex mathematical instrument so that it plays the exact same song as a standard reference, proving that the two are fundamentally the same thing.

Based on the provided text, here is a detailed technical summary of the paper "Towards a characterization of toric hyperkähler varieties among symplectic singularities II" by Yoshinori Namikawa.

1. Problem Statement

The paper addresses a specific question regarding the classification of conical symplectic varieties that admit a projective symplectic resolution and an effective Hamiltonian action by an $n$ -dimensional algebraic torus $T^n$ .

Context: In the author's previous work ([Part I]), it was proven that such a variety $(X, \omega)$ is isomorphic to a toric hyperkähler variety $(Y(A, 0), \omega_{Y(A,0)})$ via a complex analytic isomorphism $\phi$ .
The Gap: The isomorphism $\phi$ established in [Part I] was only known to be complex analytic, not necessarily algebraic. Furthermore, the $C^*$ -action (conical action) on $X$ compatible with the original structure was not guaranteed to be preserved by $\phi$ in an algebraic sense.
Specific Question: The paper aims to answer Question 5.10 from [Part I]: Can one find a $C^*$ -equivariant algebraic isomorphism between $(X, \omega)$ and $(Y(A, 0), \omega_{Y(A,0)})$ ? Specifically, can the original conical $C^*$ -action on $X$ be replaced by a suitable new algebraic $C^*$ -action such that the resulting isomorphism is both algebraic and equivariant?

2. Methodology

The proof strategy involves comparing two different $C^*$ -actions on the variety and constructing an automorphism to align them. The core steps are:

Setup and Lifting:
- Let $\pi: \tilde{X} \to X$ be a projective symplectic resolution. The Hamiltonian $T^n$ -action and the original algebraic $C^*$ -action on $X$ lift to $\tilde{X}$ .
- Using Sumihiro's theorem, $\tilde{X}$ is covered by $C^* \times T^n$ -invariant affine open sets.
Construction of a Principal Bundle:
- The author identifies a $T^n$ -invariant Zariski open subset $U \subset \tilde{X}$ where the moment map $\tilde{\mu}$ restricts to a principal $T^n$ -bundle over $(\mathfrak{t}^n)^*$ .
- Crucially, because the weight of the symplectic form is 2 ( $wt(\omega)=2$ ), the scaling action on the base $(\mathfrak{t}^n)^*$ is compatible with the $C^*$ -action.
Comparison of Two $C^*$ -Actions:
- Action 1 ( $\cdot$ ): The original algebraic $C^*$ -action on $X$ (lifted to $\tilde{X}$ ).
- Action 2 ( $\bullet$ ): A new $C^*$ -action induced on $X$ via the isomorphism $\phi$ from the standard $C^*$ -action on the model space $Y(A, 0)$ . This is initially only complex analytic.
- On the open set $U$ , both actions commute with the $T^n$ -action. Since $U$ is a principal $T^n$ -bundle, the difference between these two actions can be described by a holomorphic map $h: C^* \times U \to T^n$ .
Correction via Moser's Trick:
- The author constructs a new algebraic $C^*$ -action ( $*$ ) on $U$ by modifying the original action using a homomorphism derived from the difference map $h$ .
- The goal is to find a complex analytic $T^n$ -equivariant automorphism $\phi: (U, \omega_U) \to (U, \omega_U)$ that transforms the vector field $v'$ (generating the $\bullet$ action) into the vector field $v$ (generating the $*$ action).
- Key Technical Tool: The author employs Moser's trick. By analyzing the contraction of the symplectic form with the vector fields ( $\eta = v \lrcorner \omega$ ), the problem reduces to finding a function $g$ that solves a specific differential equation involving the Euler vector field $\zeta$ on the base.
- The condition $wt(\omega)=2$ ensures that the relevant cohomological obstruction vanishes (specifically, the term corresponding to degree 1 in the power series expansion is zero), allowing the equation to be solved.
Extension to the Global Variety:
- The local automorphism $\phi$ constructed on $U$ extends to the open set $X_0 \subset X$ (where the moment map is a submersion).
- Since $X$ is a Stein normal variety and the complement $X \setminus X_0$ has codimension $\geq 2$ , the automorphism extends uniquely to the entire variety $X$ .

3. Key Contributions and Results

Main Theorem: The paper proves that for a conical symplectic variety $(X, \omega)$ of dimension $2n$ with $wt(\omega)=2$ , admitting a projective symplectic resolution and an effective Hamiltonian $T^n$ -action, there exists a $T^n$ -equivariant algebraic isomorphism $\phi: (X, \omega) \to (Y(A, 0), \omega_{Y(A,0)})$ .
Equivariance: This isomorphism is $C^*$ -equivariant with respect to a suitably chosen new algebraic $C^*$ -action on $X$ .
Diagram Commutativity: The isomorphism makes the moment map diagram commute, identifying the moment map of $X$ with that of the toric hyperkähler variety $Y(A, 0)$ .
Algebraicity: The result upgrades the previous analytic isomorphism to an algebraic isomorphism, confirming that the complex analytic structure of such symplectic singularities is rigidly determined by their algebraic toric hyperkähler counterparts.

4. Significance

Characterization of Singularities: This work provides a definitive characterization of a broad class of symplectic singularities (those with a projective resolution and torus action) as being isomorphic to toric hyperkähler varieties in the algebraic category.
Resolution of Open Questions: It directly answers Question 5.10 from the author's previous paper, closing the gap between analytic and algebraic classifications in this context.
Methodological Innovation: The proof introduces a sophisticated technique of comparing two $C^*$ -actions via the geometry of principal bundles and utilizing Moser's trick to construct the necessary automorphism. This approach differs significantly from A. Hubbard's proof (mentioned in the introduction), which uses Cox rings.
Implications for Symplectic Geometry: The result reinforces the deep connection between symplectic singularities and toric geometry, suggesting that the "toric" nature is intrinsic to the symplectic structure when a projective resolution exists.

5. Conclusion

Namikawa successfully demonstrates that the complex analytic isomorphism established in Part I can be refined into an algebraic, $C^*$ -equivariant isomorphism by adjusting the conical action on the source variety. This solidifies the classification of these varieties as algebraic toric hyperkähler varieties, providing a complete structural description within the realm of algebraic geometry.