On maximality of involutions of hyper-Kähler manifolds and punctual Hilbert schemes of surfaces

1. Problem Statement and Motivation

The paper investigates the maximality of involutions on complex algebraic varieties. An involution $\sigma$ on a topological space $X$ is called maximal if it achieves equality in the Smith inequality:
$$b_*(X^\sigma, \mathbb{F}_2) = b_*(X, \mathbb{F}_2)$$
where $b_*$ denotes the total $\mathbb{F}_2$-Betti number and $X^\sigma$ is the fixed locus. In the context of real algebraic geometry, a real structure (an anti-holomorphic involution) is maximal if the real locus $X(\mathbb{R})$ has the maximal possible topology allowed by the complex variety.

The authors focus on two main classes of varieties:

1. Hilbert schemes of points on surfaces ($S^{[n]}$): Given a surface $S$ with an involution, does the naturally induced involution on $S^{[n]}$ preserve maximality?
2. Compact Hyper-Kähler (HK) manifolds: Specifically those of $K3^{[n]}$-type (deformation equivalent to the Hilbert scheme of $n$ points on a K3 surface). The paper addresses whether these manifolds admit maximal "brane involutions" (holomorphic or anti-holomorphic involutions preserving or reversing the hyper-Kähler structure, corresponding to (AAB), (ABA), (BAA), or (BBB) branes in Kapustin-Witten theory).

Key Questions:

• Under what conditions is the induced involution on $S^{[n]}$ maximal?
• Do compact HK manifolds of dimension $\ge 4$ ($n \ge 2$) admit maximal brane involutions?

2. Methodology

The authors employ a sophisticated blend of algebraic topology, lattice theory, and hyper-Kähler geometry:

• Smith Theory and Cohomological Characterizations: They utilize the Smith inequality and its cohomological characterizations (Proposition 2.1). Maximality is equivalent to the surjectivity of the map $H^*_G(X, \mathbb{F}_2) \to H^*(X, \mathbb{F}_2)$ and the triviality of the $\mathbb{Z}/2\mathbb{Z}$-action on the cohomology with $\mathbb{F}_2$ coefficients.
• Comessatti Characteristic: They use the Comessatti characteristic $\lambda(M, \sigma)$, defined as the dimension of the image of $1+\sigma$ on $M \otimes \mathbb{F}_2$. An involution is maximal only if this characteristic vanishes for all cohomology groups.
• Integral Cohomology of Hilbert Schemes: The paper relies heavily on the integral basis for $H^*(S^{[n]}, \mathbb{Z})$ established by Li, Qin, and Wang, and the Nakajima operators. This allows for explicit computation of how involutions act on specific cohomology classes.
• Monodromy and Lattice Theory: For $K3^{[n]}$-type manifolds, the authors use the monodromy group representation on cohomology (Markman's results). They analyze the action of involutions on the Beauville–Bogomolov–Fujiki lattice $H^2(X, \mathbb{Z})$ and its discriminant group.
• Kalinin Spectral Sequence: To handle cases with free involutions (no fixed points), they employ the Kalinin spectral sequence to detect non-trivial differentials that obstruct maximality.
• Hyper-Kähler Rotation: They use the rotation of complex structures to relate anti-holomorphic involutions (real structures) to holomorphic anti-symplectic involutions.

3. Key Contributions and Results

A. Maximality on Hilbert Schemes of Surfaces ($S^{[n]}$)

The authors provide necessary and sufficient conditions for the maximality of the induced involution on $S^{[n]}$ for a smooth projective surface $S$ with $H^1(S, \mathbb{F}_2) = 0$.

• Theorem 1.8 (Anti-holomorphic case): The induced real structure on $S^{[n]}$ is maximal if and only if:
  1. The real structure on $S$ is maximal.
  2. The real locus $S(\mathbb{R})$ is connected.
  3. (Equivalently) The involution acts as $-\text{id}$ on $H^2(S, \mathbb{Z})$.
• Theorem 1.9 (Holomorphic case): The induced holomorphic involution on $S^{[n]}$ is maximal if and only if:
  1. The involution on $S$ is maximal.
  2. The involution acts trivially on $H^2(S, \mathbb{Z})$.
• Surprising Consequence (Remark 1.10): If the induced involution is maximal for one $n \ge 2$, it is maximal for all $n \ge 2$.

B. Non-existence of Maximal Brane Involutions on $K3^{[n]}$-type Manifolds

This is the paper's most significant negative result. While maximal real K3 surfaces exist, their higher-dimensional Hilbert scheme analogues do not.

• Theorem 1.4 (Anti-symplectic/Holomorphic): For $n \ge 2$, a compact HK manifold of $K3^{[n]}$-type admits no maximal holomorphic anti-symplectic involutions.
• Corollary 1.5 (Real Structures): Consequently, there are no maximal real structures (real varieties) on compact HK manifolds of $K3^{[n]}$-type for $n \ge 2$. This answers a question posed by Fu.
• Theorem 1.6 (Symplectic/Holomorphic): There are no non-trivial maximal holomorphic symplectic involutions (BBB-branes) on these manifolds.

Mechanism of Proof for Non-existence:
The proof involves a detailed lattice-theoretic analysis of the action of an involution $\sigma$ on $H^2(X, \mathbb{Z})$.

1. They classify the possible eigen-lattices $L^\sigma$ and $L^{\sigma-}$ based on the discriminant group.
2. They construct specific cohomology classes in $H^4(X, \mathbb{Z})$ using Nakajima operators (specifically involving the exceptional divisor class $\delta$).
3. They demonstrate that for any non-trivial involution, the action on a specific subspace of $H^4$ (spanned by classes $u, v, w$) yields a matrix with a non-zero Comessatti characteristic (specifically 1).
4. This non-vanishing characteristic implies the Smith inequality is strict ($b_*(X^\sigma) < b_*(X)$), proving non-maximality.

C. Free Involutions

The paper proves that if a surface $S$ (with $H^1(S, \mathbb{F}_2)=0$) admits a free involution (no fixed points), the induced involution on $S^{[n]}$ is never maximal. This is established using the Smith-Gysin sequence and the Kalinin spectral sequence for specific surfaces like quadrics and Hirzebruch surfaces.

4. Significance and Implications

• Resolution of Open Questions: The paper definitively answers Question 1.3 regarding the existence of maximal brane involutions on $K3^{[n]}$-type manifolds, providing a negative answer for $n \ge 2$.
• Generalization of Previous Work: It generalizes and completes partial results by Fu and Kharlamov–R˘asdeaconu, extending them from $n=2$ to arbitrary $n$ and from specific surfaces to general projective surfaces with vanishing $H^1$.
• Topological Obstruction: The work establishes a strong topological obstruction to maximality in hyper-Kähler geometry. It shows that the rich geometry of $K3^{[n]}$-type manifolds (specifically their monodromy and lattice structure) prevents the existence of "perfect" real forms or brane involutions in dimensions $\ge 4$.
• Connection to Physics: By framing the results in terms of Kapustin-Witten branes (A-branes, B-branes), the paper connects algebraic geometry with string theory and mirror symmetry, showing that certain "maximal" configurations expected in physics do not exist in the compact hyper-Kähler setting.
• Deformation Invariance: The results highlight that maximality is a deformation-invariant property, allowing the authors to reduce problems to specific projective models (like K3 surfaces in $\mathbb{P}^3$) or use monodromy arguments.

5. Conclusion

The paper provides a comprehensive classification of when involutions on Hilbert schemes of surfaces are maximal and proves a striking non-existence theorem for maximal involutions on compact hyper-Kähler manifolds of $K3^{[n]}$-type. The core insight is that the interplay between the monodromy action, the lattice structure of the second cohomology, and the specific integral basis of the Hilbert scheme forces the Smith inequality to be strict for $n \ge 2$, rendering these manifolds "non-maximal" in the sense of real algebraic geometry and brane theory.