Hilbert Scheme of Three Points

• The Hilbert scheme of three points is a moduli space that parametrizes length-3 zero-dimensional subschemes on a smooth algebraic variety, detailing distinct, double-plus-simple, and punctual stratifications.
• It exhibits intricate local and global geometric structures, including smooth loci, singular planar points modeled by quadratic cones, and explicit descriptions of tangent spaces and normal bundles.
• The study informs advanced topics such as derived category semi-orthogonal decompositions, intersection theory, and birational geometry, with applications in moduli problems and categorical representation.

The Hilbert scheme of three points, denoted $\operatorname{Hilb}^3(X)$ or $X^{[3]}$, parametrizes length-3 zero-dimensional closed subschemes of a smooth algebraic variety $X$. As a fine moduli space in algebraic geometry, it organizes the structure and geometry of configurations of three points—possibly colliding or supported at higher-order infinitesimal thicknesses—on $X$. The in-depth study of $X^{[3]}$ illuminates derived categories, local singularities, intersection theory, and geometric representation theory.

1. Geometric Structure and Stratification

For any smooth projective variety $X$ of dimension $d$, the Hilbert scheme $X^{[3]}$ is a smooth, irreducible variety of dimension $3d$ over an algebraically closed field of characteristic zero (Nikolov, 2024, Selvaggi, 13 Jan 2026). The scheme parametrizes closed subschemes $W\subset X$ such that $\operatorname{length}(\mathcal{O}_W)=3$. The moduli functor of $X^{[3]}$ assigns to every $k$-scheme $T$ the set of closed subschemes $Z\subset X\times T$ flat over $T$, whose fibers are length-3 subschemes.

The Hilbert scheme $X^{[3]}$ admits a natural stratification by the support and local structure of the corresponding Artin $k$-algebra:

• Open locus of three distinct reduced points: Isomorphic to $$(X\times X\times X)\setminus\{\text{big diagonals}\}/S_3$$, this open, dense stratum is smooth of dimension $3d$.
• Double point plus simple point: Subschemes supported on two points (one of multiplicity two), forming a locally closed subset fibered over $X\times X$ with fibers $\mathbb{P}^{d-1}$.
• Closed punctual (triple) locus: Subschemes with all support at a single point. These further decompose into curvilinear (supported on a smooth curve) and planar (with Zariski tangent space of dimension 2) cases. The punctual planar locus is central for derived and singularity-theoretic investigations (Nikolov, 2024, Hu, 2021).

2. Local Structure and Singularities

The local analytic and tangent structure of $X^{[3]}$ is well-characterized. At a point $[Z]\in X^{[3]}$ corresponding to the subscheme $Z\subset X$, the Zariski tangent space is $\operatorname{Ext}^1(I_Z, \mathcal{O}_Z)$. For $X=\mathbb{A}^3$, all length-3 zero-dimensional $k$-algebras (split triple point, double + simple, or curvilinear triple) are broken Gorenstein and correspond to smooth points of $X^{[3]}$, yielding smooth loci of dimension 9 (Jelisiejew et al., 2024, Hu, 2021).

On a surface ($\dim X=2$), $\operatorname{Hilb}^3(X)$ is globally smooth (Ilten et al., 8 Oct 2025). On a threefold, only the planar triple locus (Artin local algebra $k[x,y]/(x,y)^2$) yields singularities. This locus is isomorphic, locally analytically, to $\mathbb{A}^3 \times \operatorname{Cone}(G(2,6))$, the product of three affine parameters and the quadratic cone over the Grassmannian $G(2,6)$, a normal, Cohen–Macaulay, and Gorenstein singularity of codimension 3 (Hu, 2021). All other length-3 Artin algebras are smooth points, and the singular locus is precisely the closure of the planar triple type.

3. Algebraic and Intersection-Theoretic Properties

The intersection theory of $X^{[3]}$ is governed by the structure of its Chow ring and tautological bundles (Selvaggi, 13 Jan 2026, Ryan et al., 2020). For $X$ smooth projective (characteristic not 2 or 3), the Chow ring $A^*(\operatorname{Hilb}^3(X))$ is generated as a subring of the Chow ring of the nested Hilbert scheme $\operatorname{Hilb}^{[2,3]}(X)$, and is presented by pullbacks from $A^*(X\times\operatorname{Sym}^2(X))$, extra classes from the small diagonal, and classes from the universal family—explicit generators and relations are given in (Selvaggi, 13 Jan 2026).

On surfaces, $A^*(\operatorname{Hilb}^3(S))$ is the invariant subring of $$A^*(S\times\operatorname{Sym}^2(S))[e,f]/(e^2-\cdots, f^2-\cdots)$$, reflecting the minimal two-step blowup construction.

In the specific case $X=\mathbb{P}^2$, the effective and nef cones in codimensions 2 and 3 are explicitly described using the Mallavibarrena–Sols basis (Ryan et al., 2020), via the intersection pairing and Kleiman transversality. For divisors, the effective cone is generated by the boundary divisor (parametrizing non-reduced subschemes) and by the difference with the pullback of the symmetric product. Gaeta resolutions and related exceptional vector bundles connect this intersection theory with Bridgeland stability and the birational geometry of $\operatorname{Hilb}^3(\mathbb{P}^2)$ (Huizenga, 2012).

4. Planar Locus, Grassmannian Bundle, and Normal Bundle

The punctual planar locus (subschemes supported at a point with Zariski tangent plane) is realized as a Grassmannian bundle $G=G(2,T_X)$ over $X$ (Nikolov, 2024). The universal planar length-3 family yields a closed embedding $i: G \to X^{[3]}$, identifying the planar locus inside the punctual stratum.

The normal bundle $N$ to $G$ in $X^{[3]}$ is described by the exact sequence: $$0 \to S^2Q \to Q\otimes S^2Q \to N \to 0,\quad N\cong S^3Q\otimes \det Q$$ where $Q$ is the universal rank-2 quotient bundle on $G$, and the splitting holds for $d\ge 5$ over a characteristic zero field. This explicit description is central for constructing semi-orthogonal decompositions in the derived category.

5. Derived Category and Semi-Orthogonal Decomposition

For $X$ smooth projective of $\dim X\ge 5$, the bounded derived category $D^b(X^{[3]})$ admits a semi-orthogonal sequence of length $\binom{d-3}{2}$, all pieces equivalent to $D^b(X)$ (Nikolov, 2024). These subcategories are constructed as images of fully faithful Fourier–Mukai functors associated to the planar Grassmannian locus. The corresponding kernels are supported on the images of the universal planar family and involve Young diagrams $a=(a_1\ge a_2\ge0)$ parameterizing certain symmetric powers of $Q$.

For each admissible Young diagram, the functor $P_a:D^b(X)\rightarrow D^b(X^{[3]})$ is fully faithful, and the images form a semi-orthogonal sequence. The construction is validated by vanishing criteria on $\operatorname{Ext}$ groups and higher cohomology involving the normal bundle powers. In the abelian case, these sequences restrict to the third generalized Kummer variety, yielding an exceptional sequence of length $\binom{d-3}{2}\cdot 3^d$.

6. Local Equations and Universal Families

The completed local ring of $\operatorname{Hilb}^3(\mathbb{A}^2)$ at the punctual triple point $[(x,y)^2]$ is a regular power-series ring in six variables, confirming global smoothness (Ilten et al., 8 Oct 2025). The universal family is defined by quadrics in these coordinates, and the local geometry matches that of the moduli of 3-dimensional commutative based algebras.

In higher dimensions, planar (embedding-dimension 2) triple points become singularities modeled analytically by quadratic cones over Grassmannians. The explicit equations and resolutions for the punctual planar locus are tractable, enabling in-depth calculations of deformation, cohomological, and combinatorial invariants.

7. Applications and Further Directions

The geometry and category theory of the Hilbert scheme of three points underlie advances in:

• Categorical representation theory: via semi-orthogonal decompositions and explicit Fourier–Mukai transforms.
• Birational geometry: through explicit computations of nef and effective cones and correspondence with Bridgeland walls.
• Intersection theory and tautological classes: by explicit cycle and Chern class computations, especially on surfaces and low-degree cases.
• Singularity theory: by analyzing the local and global structure of the punctual planar locus, with connections to cones over Grassmannians and Cohen–Macaulay property.

These methods and results generalize to higher-point Hilbert schemes, informing the study of stability, derived categories, moduli of sheaves, and enumerative invariants on algebraic varieties.

---

References

• (Nikolov, 2024): "A Semi-orthogonal Sequence in the Derived Category of the Hilbert Scheme of Three Points"
• (Selvaggi, 13 Jan 2026): "The Chow Ring of the Hilbert Cube"
• (Ryan et al., 2020): "Higher Codimension Cycles on the Hilbert Scheme of Three Points on the Projective Plane"
• (Jelisiejew et al., 2024): "The Hilbert scheme of points on a threefold, I"
• (Hu, 2021): "On singular Hilbert schemes of points: local structures and tautological sheaves"
• (Ilten et al., 8 Oct 2025): "Local Equations for Hilbert Schemes of Points"
• (Huizenga, 2012): "Effective divisors on the Hilbert scheme of points in the plane and interpolation for stable bundles"