---
title: Hikita Conjecture Overview
url: https://www.emergentmind.com/topics/hikita-conjecture
type: topic
---

# Hikita Conjecture Overview

Searching arXiv for primary sources on the Hikita conjecture and its variants.
The Hikita conjecture is a statement in symplectic duality that relates the fixed-point geometry of one conical symplectic singularity to the cohomology of a symplectic resolution of its dual. In a standard formulation, if \(X\) and \(X^!\) are symplectic dual and \(X^!\) admits a conical symplectic resolution \(Y^!\to X^!\), then for a maximal Hamiltonian torus \(T\) acting on \(X\) one expects a graded algebra isomorphism
\[
\mathcal O(X^T)\cong H^*(Y^!),
\]
where \(X^T\) is the scheme-theoretic \(T\)-fixed-point subscheme [1807.09858]. The conjecture has become a central organizing principle for the comparison of Higgs and Coulomb branches, affine Grassmannian slices, Nakajima quiver varieties, nilpotent orbit geometry, and quantum or equivariant refinements of these structures [2503.23590].

## 1. Classical formulation and geometric framework

The conjecture is usually placed in the setting of conical symplectic singularities. One standard setup considers a normal affine Poisson variety \(X\) with a contracting \(\mathbb G_m\)-action such that \(\mathcal O(X)\) is nonnegatively graded, the degree-zero part is \(\mathbb C\), the degree-one part vanishes, the Poisson bracket has degree \(-2\), and the smooth locus carries a symplectic form extending to a resolution [1807.09858]. In this framework, the Hamiltonian automorphism group has Lie algebra \(\mathcal O(X)^2\), and a maximal torus \(T\subset \operatorname{Aut}(X)\) defines the fixed-point scheme that appears on the Hikita side.

The content of the conjecture is scheme-theoretic rather than merely set-theoretic. The algebra \(\mathcal O(X^T)\) records nilpotent structure in the fixed locus, so nonreduced fixed-point schemes are part of the expected correspondence. This feature is visible in explicit examples: for \(\overline{T^*(G/U)}\), the fixed-point subscheme \(\overline{T^*(G/U)}^{T\times B/U}\) is finite-dimensional and nonreduced in general [2410.16217].

Within symplectic duality, the conjecture is compatible with a broader exchange of linear data. One expects identifications
\[
H_2(X;\mathbb C)\cong (t^!)^*, \qquad t\cong H^2(X^!;\mathbb C),
\]
together with a matching between positive equivariant roots on one side and positive Kähler roots on the other [1807.09858]. This places the Hikita conjecture alongside wall-crossing, deformation theory, and quantum connection structures rather than as an isolated isomorphism.

## 2. Equivariant, Hikita–Nakajima, and quantum forms

An equivariant refinement, often called the Hikita–Nakajima conjecture, replaces the fixed-point coordinate ring by the \(B\)-algebra of a quantization and replaces ordinary cohomology by equivariant cohomology. In the formulation recalled in the quantum Hikita literature, the proposed comparison is
\[
B(A)\cong H^*_{T^!\times G}(X^!;\mathbb C),
\]
where \(A\) is a quantization of the singularity and \(B(A)\) is the Cartan-type quotient associated with a torus grading [1807.09858]. In a characteristic-\(p\) restatement, for a \(X^\bullet(T)\)-graded algebra \(A\),
\[
B(A)=A_0\Big/\sum_{\langle \sigma,\lambda\rangle>0}A_\lambda A_{-\lambda},
\]
which makes precise how the scheme-theoretic fixed-point algebra is replaced by a quantized analogue [2503.23590].

The quantum Hikita conjecture upgrades the algebra isomorphism to an isomorphism of \(D\)-modules. On the quantization side one forms a module of graded traces, and on the geometric side one takes a specialized quantum \(D\)-module built from equivariant quantum cohomology. The conjecture predicts that, after localization away from root hyperplanes, these modules are isomorphic [1807.09858]. In that formulation, the specialization \(q=0\) recovers the \(B\)-algebra/equivariant Hikita picture, while the specialization \(q=1\) is related to degree-zero Hochschild homology and conjecturally to intersection cohomology [1807.09858].

This hierarchy of statements is important because many proofs proceed through the refined versions rather than directly through the original cohomology/fixed-point identity. In practice, equivariant cohomology, Cartan subquotients, trace modules, and quantum differential or \(q\)-difference equations often carry more structure than the classical rings and can be compared more directly.

## 3. Established cases and model examples

Several major families of symplectic dual pairs now satisfy classical, equivariant, or quantum Hikita-type statements.

| Dual pair or setting | Result | Source |
|---|---|---|
| Affine Grassmannian slices and Nakajima quiver varieties | \(\mathbb C[(\overline{\Gr^\lambda_\mu})^{\mathbb C^\times}]\cong H^*(\mathfrak M(\mathbf m,W))\) | [1511.09131] |
| Minimal nilpotent orbit closure and ADE Kleinian resolution | Classical Hikita proved; quantum version proved for ADE and extended to BCFG analogues | [1903.12205], [2302.13249] |
| Hypertoric varieties and Springer resolution | Quantum Hikita conjecture proved | [1807.09858] |
| Gieseker variety \(\mathfrak M(n,r)\) | Hikita–Nakajima conjecture proved explicitly on generators | [2202.09934] |
| Finite ADE quiver gauge theories | Equivariant and K-theoretic Hikita conjectures proved | [2509.06226] |

For affine Grassmannian slices, the proved statement
\[
\mathbb C\big[(\overline{\Gr^\lambda_\mu})^{\mathbb C^\times}\big]\cong H^*(\mathfrak M(\mathbf m,W))
\]
ties affine Grassmannian geometry directly to quiver-variety cohomology and also fits the representation-theoretic analysis of truncated shifted Yangians and product monomial crystals [1511.09131]. In type \(A\), the same work proves that highest weights of truncated shifted Yangians are exactly the relevant product monomial crystal points, so the Hikita isomorphism sits inside a larger algebra–geometry–combinatorics correspondence [1511.09131].

For the minimal nilpotent orbit closure \(\overline{\mathcal O_{\min}}\) of ADE type and the minimal resolution \(Y\to \mathbb C^2/\Gamma\), the classical theorem identifies
\[
H^*(Y)\cong \mathbb C[(\overline{\mathcal O_{\min}})^T].
\]
The proof proceeds by comparing the cohomology ring of the ADE surface resolution with the projection of the defining ideal of \(\overline{\mathcal O_{\min}}\) to the Cartan algebra [1903.12205]. This case was later quantized: the specialized quantum \(D\)-module of the equivariant quantum cohomology of the Kleinian resolution was shown to agree with the \(D\)-module of graded traces on the Joseph quantization of the minimal nilpotent orbit, verifying the quantum Hikita conjecture in the ADE case and producing analogous folded statements for BCFG types [2302.13249].

For the Gieseker variety \(\mathfrak M(n,r)\), the Hikita–Nakajima isomorphism was proved explicitly by embedding equivariant cohomology, the Coulomb-branch fixed-point algebra, and the center of a degenerate cyclotomic Hecke algebra into a common product algebra indexed by multipartitions. Under this identification, tautological Chern classes, Jucys–Murphy elements, and Dunkl–Opdam central elements become the same generators [2202.09934]. This is one of the cleanest generator-level realizations of the conjecture.

## 4. Partial results, failures, and refined formulations

A persistent misconception is that the naïve statement of the Hikita conjecture should hold uniformly for every expected symplectic dual pair. Recent work shows that this is false. For nilpotent Slodowy slices and affinizations of certain \(G\)-equivariant covers of special nilpotent orbits, the original Hikita–Nakajima statement fails: the relevant Cartan quotient can have torsion, while the cohomological algebra is free over the parameter ring, and on the classical side one encounters non-cyclicity or indecomposability mismatches [2410.20512]. The replacement proposed there is a refined Hikita conjecture comparing the images of canonical maps from larger polynomial algebras; this refined statement is proved for parabolic Slodowy varieties [2410.20512].

In the setting of classical Lie algebras, the conjecture is verified only under explicit geometric conditions. When \(\overline{\mathcal O}_e\) is normal and \(\tilde{\mathcal O}_e=\mathcal O_e\), the desired isomorphism is equivalent to a weak flatness condition for the scheme-theoretic intersection \(\overline{\mathcal O}_e\cap \mathfrak h\) together with surjectivity of the Springer pullback map. The paper proves the conjecture for several families in types \(B\), \(C\), and \(D\), especially when all \(\mathfrak{gl}\)-factors in the relevant Levi have size at most \(2\), and gives further evidence in distinguished and special spherical cases [2409.13914]. This classification-based work makes clear that the geometric fixed-point ring can deviate from Springer-fiber cohomology unless additional flatness and surjectivity mechanisms are present.

There are also cases where one direction of the conjecture is established but the full isomorphism remains open. For
\[
\mathcal N///T=\{x\in \mathcal N_{\mathfrak{sl}_n}\mid \operatorname{diag}(x)=0\}/T,
\]
the Hamiltonian reduction is an unframed Nakajima quiver variety admitting a symplectic resolution \(\widetilde{\mathcal N///T}\), while the dual BFN Coulomb branch is \(\overline{T^*(G/U)}\). In this example the expected Hikita isomorphism is
\[
H^*(\widetilde{\mathcal N///T})\cong \mathbb C[\overline{T^*(G/U)}^{T\times B/U}],
\]
but the proved result is only a surjective graded algebra map
\[
\mathbb C[\overline{T^*(G/U)}^{T\times B/U}] \twoheadrightarrow H^*(\widetilde{\mathcal N///T}).
\]
The construction uses an explicit presentation of the fixed-point ring by generators
\[
f_{S,T}=\prod_{s\in S,\;t\in T}(x_s-y_t),\qquad |S|+|T|=n,
\]
together with Kirwan surjectivity for smooth quiver varieties [2410.16217]. Since equality of graded dimensions is still open, the paper proves “Hikita surjectivity” rather than the full conjecture [2410.16217].

## 5. K-theoretic, elliptic, and arithmetic extensions

The conjectural pattern extends beyond ordinary cohomology. For hypertoric varieties, an elliptic analogue replaces ordinary cohomology and \(K\)-theory by equivariant elliptic cohomology and replaces additive or multiplicative moment maps by elliptic-valued moment maps. In this setting one obtains an elliptic Hikita statement
\[
\mathrm{Ell}_{K^\vee}\bigl(\mathfrak M_v^+(\alpha,\beta)\bigr)\cong (X_\vartheta^n/K)^A,
\]
identifying the equivariant elliptic cohomology of an additive hypertoric variety with the \(A\)-fixed-point scheme of an elliptic hypertoric quotient [2204.12233]. The construction exhibits theta functions as elliptic analogues of the linear and multiplicative classes that appear in the cohomological and \(K\)-theoretic theories.

For quiver gauge theories, a K-theoretic Hikita conjecture replaces equivariant cohomology on the Higgs side by equivariant \(K\)-theory and replaces the homological Coulomb branch by the K-theoretic Coulomb branch. A central structural theorem identifies appropriate completions of K-theoretic and homological Coulomb branches, allowing the K-theoretic statement to be deduced from the homological one for a large class of theories. This program yields proofs of the equivariant K-theoretic Hikita conjecture in finite ADE type [2509.06226].

A further arithmetic extension works in characteristic \(p\gg 0\). In that setting the mod-\(p\) quantum Hikita conjecture proposes that Frobenius-constant quantizations on the trace side correspond to quantum Steenrod operations on the quantum-cohomology side. The paper verifies this for dual Springer resolutions and for Gale-dual hypertoric varieties, showing that the extra characteristic-\(p\) operators on the two \(D\)-modules are intertwined by the Hikita comparison [2503.23590]. This adds a genuinely arithmetic layer to 3D mirror symmetry rather than merely changing coefficients.

At roots of unity, an explicitly K-theoretic quantum version replaces differential equations by \(q\)-difference equations. The proposed isomorphism
\[
M_{q,eq}(X)\cong M_{q,Kah}(X^!)
\]
compares a module built from the K-theoretic quantized Coulomb branch with a quasimap quantum \(K\)-theory module on the Higgs side. After specializing \(q\) to a primitive root of unity, the paper matches Frobenius-constant central operators with quantum Adams operations and proves the resulting arithmetic K-theoretic quantum Hikita conjecture for hypertoric varieties [2510.09335].

## 6. Mirror-symmetric, enumerative, and combinatorial directions

The quantum Hikita framework increasingly interacts with enumerative geometry. For abelian quantum Higgs branches, twisted traces on the quantum Hamiltonian reduction admit an explicit expansion in terms of twisted traces of Verma modules. The resulting formula is interpreted as having the appearance of an Atiyah–Bott localization formula under a quantum Hikita isomorphism: Verma-module traces behave like fixed-point contributions, while the sphere trace plays the role of integration over the mirror Coulomb branch [2308.15198]. This does not prove quantum Hikita in full generality, but it clarifies the trace-theoretic side expected by the conjecture.

A related enumerative direction appears in quiver gauge theories under the operation of slant sum. There, branching formulas for quasimap vertex functions on Nakajima quiver varieties lead, after \(\hbar=q\) specialization, to product structures that the authors explicitly connect to the quantum Hikita conjecture. When transported through mirror symmetry, these formulas yield conjectural expressions for graded traces of Verma modules on the Coulomb side, and in ADE type several corresponding Coulomb-side conjectures are proved [2510.02496].

The name “Hikita” also appears in adjacent combinatorial literatures that should not be conflated with the fixed-point/cohomology conjecture. The extended rational shuffle theorem identifies \(Q_{m,n}(1)\) with the extended Hikita polynomial \(H_{m,n}[X;q,t]\), and a substantial literature analyzes its Schur expansion and symmetry properties [1806.04348]. Separately, the affine \(\Delta\)-Springer fiber \(Y_{n,k}\) was introduced as a geometric generalization of the affine Springer fiber studied by Hikita for \(\nabla e_n\), producing a “Hikita-style” geometric realization of the Delta Conjecture through bigraded Borel–Moore homology [2501.00197]. A plausible implication is that “Hikita” now names a broader research lineage spanning symplectic duality, affine Springer geometry, and shuffle-theoretic combinatorics, even though the original Hikita conjecture remains the fixed-point/cohomology correspondence of symplectic duality.

The present state of the subject is therefore mixed but structurally coherent. There are complete proofs in several important families, robust quantum, elliptic, K-theoretic, and arithmetic generalizations, and explicit mechanisms—Kirwan maps, \(B\)-algebras, graded traces, localization, and Riemann–Roch comparisons—that explain why the conjecture should exist. At the same time, recent counterexamples and refined formulations show that the naïve form is not universally stable, and that in some symplectic-duality settings the correct object is an image algebra, a surjective Hikita map, or a localized or completed module rather than a direct unqualified ring isomorphism [2410.20512].

Source: https://www.emergentmind.com/topics/hikita-conjecture