Crepant Hilbert–Chow Resolution

Updated 8 February 2026

The crepant Hilbert–Chow resolution is a method in algebraic geometry that provides a projective, birational morphism resolving symmetric product singularities with zero discrepancy in canonical classes.
It utilizes Hilbert schemes of points on smooth surfaces to map subschemes to Chow cycles, enabling explicit comparisons of cohomological, motivic, and enumerative invariants.
The resolution underpins advances in enumerative geometry and derived categories by linking Donaldson–Thomas and Gromov–Witten invariants through incidence correspondences and wall-crossing formulas.

A crepant Hilbert–Chow resolution is a projective, birational morphism from a smooth variety (or stack)—typically a Hilbert scheme of points—to a singular symmetric product or quotient, distinguished by the property that the canonical bundles pull back isomorphically (i.e., with discrepancy zero). This construction plays a pivotal role in the birational geometry of surfaces and higher-dimensional varieties, resolving quotient singularities in a manner compatible with symplectic or Calabi–Yau structures and enabling comparison between the cohomological, motivic, and enumerative invariants of singular and resolved spaces. Its paradigmatic instance is the Hilbert–Chow morphism $\rho \colon \Hilb^n(S) \to \Sym^n(S)$ for a smooth surface $S$ , which resolves the singularities of the symmetric product and underpins major developments in algebraic geometry, mathematical physics, and representation theory.

1. Formal Construction and Crepancy Criteria

Let $S$ be a smooth projective surface over $\mathbb{C}$ . The Hilbert scheme $X = \Hilb^n(S)$ parametrizes length- $n$ zero-dimensional subschemes $Z \subset S$ , while the $n$ -fold symmetric product $Y = \Sym^n(S) = S^n/\mathfrak{S}_n$ parametrizes effective $0$-cycles of length $n$ . The Hilbert–Chow morphism

$\rho \colon X \longrightarrow Y$

sends a subscheme $Z$ to its Chow cycle $\sum_{p} \ell_p(Z)\cdot p$ .

Fogarty's theorem ensures smoothness of $X$ , with $\dim X = 2n$ . The symmetric product $Y$ is normal, with singular locus corresponding to cycles with multiplicities (i.e., nonreduced points), notably along the big diagonal.

A resolution $f: \widetilde X \to X$ is called crepant if the canonical divisors pull back isomorphically: $K_{\widetilde X} = f^*K_X$ . In the Hilbert–Chow scenario, both $X$ and $Y$ are Gorenstein with trivial canonical class in the K3, Enriques, abelian, and many toric or surface cases. The exceptional divisor over the locus of nonreduced cycles is irreducible and isomorphic to $\mathbb{P}(T_S)$ for $n=2$ ; the discrepancy in the canonical bundle vanishes, thus ensuring crepancy (Mukherjee et al., 1 Feb 2026, Fu et al., 2016, Kaplan et al., 2023).

2. Geometry of the Hilbert–Chow Morphism

The morphism $\rho$ is projective, birational, and an isomorphism over the open locus of reduced cycles. Its exceptional locus consists of subschemes with nontrivial multiplicity structure at points of $S$ —for $n=2$ the divisor $\Delta \subset \Hilb^2(S)$ parametrizes nonreduced length-$2$ subschemes.

The diagrammatic factorization is as follows: $S^n \longrightarrow Y = \Sym^n(S) \ \quad \downarrow \ X = \Hilb^n(S) \overset{\rho}{\longrightarrow} Y$ with the Hilbert–Chow morphism factoring via an orbifold or stacky blowup of the diagonal.

The resolution inherits crucial properties:

The Hilbert scheme $X$ is smooth (and for $S$ K3 or abelian, hyperkähler).
$Y$ has rational, finite quotient singularities and is Cohen–Macaulay.
The global relations $\rho_*\mathcal{O}_X \cong \mathcal{O}_Y$ and $R^i\rho_*\mathcal{O}_X = 0$ for $i > 0$ are satisfied (Mukherjee et al., 1 Feb 2026, Kaplan et al., 2023).

3. Crepant Hilbert–Chow Resolution: Cohomological, Motivic, and Enumerative Aspects

Cohomological and Motivic Identifications

In the context of an abelian or K3 surface, the Hilbert–Chow map induces isomorphisms at the level of motives, cohomology, and Chow rings. Specifically, for an abelian surface $A$ : $h(A^{[n]}) \cong h_{\text{orb,dt}}([A^n/\mathfrak{S}_n])$ as algebra objects in the category of motives, where “dt” indicates a discrete torsion twist to account for the Reid–Taylor age shift (Fu et al., 2016).

For cohomology rings (with $\mathbb{Q}$ -coefficients), one recovers: $H^*(\Hilb^n(S)) \cong H^*_{\text{orb,dt}}([S^n/\mathfrak{S}_n])$ with the isomorphism realized via explicit incidence correspondences and Nakajima–Grojnowski/Lehn–Sorger constructions. These results were extended to generalized Kummer varieties and are fundamental to the Cohomological HyperKähler Resolution Conjecture.

Donaldson–Thomas and Gromov–Witten Theory

For three-dimensional Calabi–Yau orbifolds and their G–Hilb crepant resolutions, Donaldson–Thomas (DT) invariants and Gromov–Witten (GW) invariants associated to $X$ and $Y$ are interrelated via generating functions, reflecting a generalization of the McKay correspondence. Explicit comparison formulas for DT invariants hold for point-class and (conditionally) for curve-class invariants, with the perverse t-structure and Fourier–Mukai equivalence providing categorical bridges (Calabrese, 2012).

For surfaces, the “SYM–HILB correspondence” identifies extremal degree Gromov–Witten invariants for the Hilbert scheme and degree-zero extended orbifold invariants for the symmetric product, with a precise change of variables $q=-e^{iu}$ required for isomorphism at the level of quantum products (Cheong, 2011, Nesterov, 6 Jan 2026, Genlik et al., 2023). This realizes the prediction of Ruan’s Cohomological Crepant Resolution Conjecture (CCRC) and its generalizations.

4. Gluing, Stratification, and Uniqueness of Resolutions

The uniqueness and global projectivity of the crepant Hilbert–Chow resolution critically depend on the singularity type of $S$ . For surfaces with only du Val singularities, $\Sym^n(S)$ admits a finite stratification by the multiplicity vectors at singular points. Along each stratum, the germ of the symmetric product is locally a product of smooth and du Val-symmetric-power germs, each with a unique crepant (Hilbert–Chow) resolution (Kaplan et al., 2023).

A formalism using constructible sheaves of local resolutions and relative Picard sheaves ensures that the Hilbert–Chow map glues uniquely to a global projective resolution in the absence of gerbe obstructions. In the symplectic context, the necessary gerbe is always trivial, and the existence of a global relatively ample line bundle on $\Hilb^n(S)$ (the tautological determinant bundle) further guarantees global projectivity. Thus, the Hilbert–Chow morphism is uniquely characterized among crepant resolutions in this setting.

5. Quiver Varieties, Partial Resolutions, and ADE Surface Singularities

Hilbert schemes on crepant partial resolutions of Kleinian (ADE) surface singularities can be constructed explicitly as Nakajima quiver varieties for suitable GIT stability conditions. For $T \subset \mathrm{SL}_2(\mathbb{C})$ a finite subgroup, $A^2/T$ is resolved by $S \to A^2/T$ , and partial resolutions contract subsets of the (-2)-curve configuration.

The Hilbert scheme $\Hilb^n(X)$ is realized as a quiver variety $M_\theta(v)$ ; the Hilbert–Chow morphism is identified with variation of GIT quotient, and the exceptional locus contracts precisely the loci in $\Hilb^n(X)$ corresponding to subschemes meeting the singular locus. For generic stability parameters, both source and target are smooth symplectic Gorenstein, ensuring crepancy of the resolution (Craw et al., 2024).

In type $A_n$ singularities, the G–Hilb resolution matches the all-genera orbifold and resolved GW theories after explicit changes of variables, confirming the Crepant Resolution Conjecture in this prototypical setting (0811.2023).

6. Applications: Sheaf Theory and Enumerative Geometry

The crepant Hilbert–Chow resolution enables transfer of geometric data between the singular and resolved sides. Given an Ulrich bundle $E$ on $S$ , one can form an $\mathfrak{S}_n$ -equivariant external tensor product, descend it to $\Sym^n(S)$, and pull back reflexively to $\Hilb^n(S)$ via the crepant resolution. The vanishing of higher direct images and preservation of cohomological amplitudes are immediate consequences of the crepancy, symplectic structure, and rational singularities of the symmetric product (Mukherjee et al., 1 Feb 2026).

Enumerative correspondences—Gromov–Witten/Donaldson–Thomas, quantum cohomology, and more—are realized through wall-crossing formulas, localization, and master space constructions relating the Hilbert scheme, symmetric product, and various compactifications (Nesterov, 6 Jan 2026, Cheong, 2011). This framework underlies the modern approach to the GW/DT correspondence and the calculation of quantum invariants on surfaces and threefolds.

7. Chow Groups, Tautological Classes, and Motivic Decompositions

The Hilbert–Chow map induces an explicit decomposition of the Chow groups and motives of $\Hilb^n(S)$ in terms of those of $\Sym^n(S)$, with summands indexed by the locus of punctuality (nonreduced length at points): $\mathrm{CH}^*(\Hilb^n(S)) \cong \bigoplus_{r=0}^{n-1} \Gamma_r \mathrm{CH}^{*-r}(\Sym^n(S))$ Here $\Gamma_r$ are explicit correspondences identifying the punctual strata. The exceptional divisor corresponds to the nonreduced locus and generates the Chow ring along with tautological pull-ups from $S$ . The canonical divisor is preserved under pullback, giving another proof of crepancy (Jiang, 2020).

These structural results align with the motivic isomorphisms in the hyperkähler and abelian surface cases, providing a bridge between classical intersection theory and modern derived/motivic structures (Fu et al., 2016).

In summary, the crepant Hilbert–Chow resolution is a central construction in modern algebraic geometry, uniquely resolving the singularities of symmetric products of surfaces and higher-dimensional orbifolds in a manner compatible with symplectic and Calabi–Yau structures. Its universality is reflected in the motivic, cohomological, and enumerative theory, as well as in applications to derived categories, wall-crossing, and moduli problems. Recent results have confirmed its pivotal role in establishing conjectural equivalences and correspondences across diverse geometric, categorical, and enumerative contexts (Mukherjee et al., 1 Feb 2026, Fu et al., 2016, Li et al., 2012, Kaplan et al., 2023, Nesterov, 6 Jan 2026, Craw et al., 2024).