Relative Springer Theory: Geometric Extensions

• Relative Springer Theory is a framework that generalizes classical Springer theory by incorporating additional geometric data like parabolic, symmetric, and affine enhancements.
• It employs projective G-equivariant maps, Steinberg varieties, and convolution-based homology to construct representation-theoretic correspondences.
• Applications span partial Springer sheaves, symmetric spaces, and affine Hitchin-type geometries, showcasing its versatile impact on modern representation theory.

Relative Springer Theory designates a family of extensions of classical Springer theory in which the basic Springer package is performed relative to additional geometric data: parabolic or Levi subgroups, symmetric pairs, affine and global families, moduli stacks, moment maps, deformation spaces, or motivic and $K$-theoretic enhancements. In the survey framework, Springer theory starts from a tuple $(G,P_i,V,F_i)_{i\in I}$, with homogeneous bundles $E_i:=G\times^{P_i}F_i$, disjoint union $E:=\bigsqcup_{i\in I}E_i$, a projective $G$-equivariant map $\pi:E\to V$, and Steinberg variety $Z:=E\times_VE$; the associated equivariant Borel–Moore homology algebra $H_*^A(Z)$ and the Springer fiber modules organize both classical and generalized correspondences (Sauter, 2013). Within that broadened usage, the literature treats partial Springer sheaves and relative Weyl groups (Chatterjee et al., 2024), quasi-split symmetric pairs (Leslie, 2019), affine and global Hitchin-type settings (Yun, 2011), elliptic and motivic versions (Ben-Zvi et al., 2013, Eberhardt, 2018), and $K$-motivic or spherical-variety realizations of Hecke-module phenomena (Eberhardt, 2024, Ma et al., 15 Dec 2025).

1. General framework beyond the classical nilpotent cone

The survey “A Survey on Springer Theory” defines a Springer map as a collapsing of homogeneous bundles. Starting from $(G,P_i,V,F_i)_{i\in I}$, one forms

$(G,P_i,V,F_i)_{i\in I}$0

and a projective $(G,P_i,V,F_i)_{i\in I}$1-equivariant map

$(G,P_i,V,F_i)_{i\in I}$2

The fibers $(G,P_i,V,F_i)_{i\in I}$3 are the Springer fibers, and the associated Steinberg variety is

$(G,P_i,V,F_i)_{i\in I}$4

Its equivariant Borel–Moore homology $(G,P_i,V,F_i)_{i\in I}$5, with convolution product, is the Steinberg algebra (Sauter, 2013).

A central organizing statement is the Ext-algebra description

$(G,P_i,V,F_i)_{i\in I}$6

together with the BBD decomposition theorem for $(G,P_i,V,F_i)_{i\in I}$7. In this form, the decomposition theorem parametrizes indecomposable projective graded modules and simple graded modules over the Steinberg algebra, while in the semi-small case the projective graded modules are equivalent to a category of shifts of perverse sheaves (Sauter, 2013).

The survey states explicitly that it does not develop a separate formal theory called “relative Springer theory,” but it does treat the broader framework in which $(G,P_i,V,F_i)_{i\in I}$8 need not be Borel, the image can be any $(G,P_i,V,F_i)_{i\in I}$9-stable subvariety of $E_i:=G\times^{P_i}F_i$0, and the same machinery produces Steinberg algebras, projective modules, and Springer fiber modules. This suggests that, at the structural level, relative Springer theory is best viewed not as a single definition but as a common pattern: a Springer-type map, a Steinberg-type correspondence, and a representation-theoretic package extracted from convolution, decomposition, and fiber homology (Sauter, 2013).

2. Parabolic and Levi versions: partial Springer sheaves and relative Weyl groups

A direct relative extension replaces the full flag variety by a partial flag variety attached to a parabolic subgroup $E_i:=G\times^{P_i}F_i$1 with Levi factor $E_i:=G\times^{P_i}F_i$2. In this setting one defines

$E_i:=G\times^{P_i}F_i$3

and the partial Springer sheaf

$E_i:=G\times^{P_i}F_i$4

Likewise, one has the partial Grothendieck resolution

$E_i:=G\times^{P_i}F_i$5

and

$E_i:=G\times^{P_i}F_i$6

The relevant symmetry group is the relative Weyl group

$E_i:=G\times^{P_i}F_i$7

(Chatterjee et al., 2024).

Over the regular semisimple locus, $E_i:=G\times^{P_i}F_i$8 is a Galois cover with Galois group $E_i:=G\times^{P_i}F_i$9, yielding

$E:=\bigsqcup_{i\in I}E_i$0

A particularly clean statement is

$E:=\bigsqcup_{i\in I}E_i$1

so the partial objects are the $E:=\bigsqcup_{i\in I}E_i$2-fixed parts of the full Grothendieck and Springer sheaves (Chatterjee et al., 2024).

The central comparison problem concerns two natural $E:=\bigsqcup_{i\in I}E_i$3-actions on the partial Springer sheaf: one obtained by restriction from the partial Grothendieck sheaf and one obtained through the Fourier transform. Since $E:=\bigsqcup_{i\in I}E_i$4 need not be a Coxeter group, the classical sign character is replaced by the top cohomology character

$E:=\bigsqcup_{i\in I}E_i$5

coming from the one-dimensional $E:=\bigsqcup_{i\in I}E_i$6-representation $E:=\bigsqcup_{i\in I}E_i$7. The resulting algebra automorphism is

$E:=\bigsqcup_{i\in I}E_i$8

and the main comparison theorem is

$E:=\bigsqcup_{i\in I}E_i$9

For $G$0, this recovers the usual sign-character statement. The same paper also proves that if the nilpotent cone $G$1 is $G$2-smooth, then $G$3 is a subobject of $G$4, which places the partial theory inside parabolic induction from the Levi (Chatterjee et al., 2024).

This parabolic–Levi form is one of the clearest meanings of the adjective “relative”: the full Weyl group is replaced by a relative Weyl group, the full Springer sheaf by a partial one, and the sign comparison by a canonical character twist determined by $G$5.

3. Symmetric spaces and spherical varieties

For symmetric spaces, the relative geometry starts from a quasi-split symmetric pair $G$6, where $G$7 is connected reductive, $G$8 is an involution, and $G$9. On Lie algebras one has

$\pi:E\to V$0

and the action of $\pi:E\to V$1 on $\pi:E\to V$2 is the symmetric-space analogue of the adjoint action. The quotient is written

$\pi:E\to V$3

where $\pi:E\to V$4 is the little Weyl group (Leslie, 2019).

The paper “An analogue of the Grothendieck-Springer resolution for symmetric spaces” constructs a closed subscheme

$\pi:E\to V$5

and proves a proper surjective morphism

$\pi:E\to V$6

such that over the regular locus

$\pi:E\to V$7

The same work introduces the regular stabilizer group scheme $\pi:E\to V$8 and proves the relative Donagi–Gaitsgory isomorphism

$\pi:E\to V$9

as smooth commutative group schemes over $Z:=E\times_VE$0. It also proposes a larger space $Z:=E\times_VE$1 whose fibers decompose into components indexed by the nilpotent cone of the centralizer of the semisimple part, and each piece maps as a resolution of singularities onto the corresponding irreducible component of a fiber of $Z:=E\times_VE$2 (Leslie, 2019).

The same paper is explicit about the present boundary of the subject. What remains open is a full sheaf-theoretic Springer correspondence for quasi-split symmetric spaces, especially an action of the little Weyl group $Z:=E\times_VE$3 on the cohomology of the relevant fibers analogous to the classical Weyl group action on Springer fibers. It proves some semi-smallness results, conjectures that $Z:=E\times_VE$4 is semi-small for all quasi-split involutions, and verifies this in some cases, notably type $Z:=E\times_VE$5 (Leslie, 2019).

A second moment-map form of relative Springer theory arises for spherical varieties. For a $Z:=E\times_VE$6-scheme $Z:=E\times_VE$7 with $Z:=E\times_VE$8-equivariant moment map $Z:=E\times_VE$9, one defines the relative Steinberg scheme

$H_*^A(Z)$0

and the relative Springer sheaf

$H_*^A(Z)$1

The key decomposition formula is

$H_*^A(Z)$2

where the multiplicity spaces are top-homology multiplicity spaces of fibers of the moment map. In the spherical-variety setting this yields

$H_*^A(Z)$3

and in the theta-correspondence setting it produces analogous decompositions for oscillator bimodules in terms of Springer representations (Ma et al., 15 Dec 2025).

4. Affine and global forms: Hitchin-type geometry, component groups, and microlocal control

Affine Springer theory replaces the flag variety by the affine flag variety and the Springer fibers by affine Springer fibers. For a regular semisimple element $H_*^A(Z)$4, with $H_*^A(Z)$5, the affine Springer fiber $H_*^A(Z)$6 consists of Iwahori subgroups whose Lie algebras contain $H_*^A(Z)$7. Lusztig constructed an action of the affine Weyl group, and Yun extends this to the extended affine Weyl group

$H_*^A(Z)$8

acting on both homology and cohomology. There is also an action of the component group $H_*^A(Z)$9, and these actions commute (Yun, 2011).

The paper “The spherical part of the local and global Springer actions” studies the center

$K$0

called the spherical part of $K$1. Its main theorem states that, for affine Springer fibers, the action of the spherical algebra factors through the component group of the local centralizer. The global form proves first that the spherical part of the $K$2-action on the cohomology sheaves of the parabolic Hitchin complex factors through the sheaf of component groups $K$3, and the local statement is then deduced from the global one by deforming points on a curve and using Ngô’s product formula (Yun, 2011).

This local–global passage is one of the strongest relative features in the subject. The geometry is not attached to a single fiber alone: the global Picard stack, the parabolic Hitchin fibration, and deformation along a curve control the local affine Springer problem. The paper states explicitly that this gives an example of using global Springer theory to solve more classical problems, and that the result fits into a broader framework of families, parabolic structures, and global moduli spaces (Yun, 2011).

A microlocal refinement appears in “Support singulier et homologie des fibres de Springer affines”. That work develops a theory of singular support for certain infinite-dimensional stacks and applies it to the affine Grothendieck–Springer sheaf

$K$4

associated to

$K$5

Its main support statement identifies

$K$6

as a microsupport for $K$7. The same paper relates this to the root valuation stratification

$K$8

and to the Goresky–Kottwitz–MacPherson local constancy conjecture asserting that $K$9 is locally constant on the strata. The final implication from singular-support containment to local constancy is not yet fully established in the affine setting, because the required Whitney-type microlocal formalism is not available in full generality for the strata $(G,P_i,V,F_i)_{i\in I}$0 (Bouthier, 2022).

5. Elliptic, motivic, and $(G,P_i,V,F_i)_{i\in I}$1-motivic extensions

An elliptic version replaces Lie algebras and nilpotent cones by moduli stacks of bundles on an elliptic curve $(G,P_i,V,F_i)_{i\in I}$2. The basic Eisenstein diagram is

$(G,P_i,V,F_i)_{i\in I}$3

and the key map is the restriction to degree zero

$(G,P_i,V,F_i)_{i\in I}$4

called the elliptic Grothendieck–Springer resolution. The paper proves that $(G,P_i,V,F_i)_{i\in I}$5 is proper and small, and that over the regular semisimple locus it becomes a $(G,P_i,V,F_i)_{i\in I}$6-cover. For a finite-rank $(G,P_i,V,F_i)_{i\in I}$7-equivariant local system $(G,P_i,V,F_i)_{i\in I}$8 on $(G,P_i,V,F_i)_{i\in I}$9, the elliptic Grothendieck–Springer sheaf is

$(G,P_i,V,F_i)_{i\in I}$00

The resulting functor defines a fully faithful embedding

$(G,P_i,V,F_i)_{i\in I}$01

where $(G,P_i,V,F_i)_{i\in I}$02, and the image consists of perverse sheaves with nilpotent singular support. The paper explicitly describes this as a relative version of Grothendieck–Springer theory because the base is the moduli of semistable bundles rather than a Lie algebra quotient, and because the construction is a geometric Eisenstein series specialized to degree-zero semistable bundles (Ben-Zvi et al., 2013).

A motivic extension is developed in “Springer Motives”. In the classical setting

$(G,P_i,V,F_i)_{i\in I}$03

is the Springer resolution, and for a nilpotent element $(G,P_i,V,F_i)_{i\in I}$04 the Springer fiber is $(G,P_i,V,F_i)_{i\in I}$05. The motivic theorem asserts that the motive of a Springer fiber is pure Tate: $(G,P_i,V,F_i)_{i\in I}$06 The paper then defines a category of equivariant Springer motives on the nilpotent cone generated by the Springer motive $(G,P_i,V,F_i)_{i\in I}$07 and shows that its dg endomorphism algebra is formal, with cohomology the graded affine Hecke algebra. Derived Morita theory yields an equivalence with the bounded derived category of finitely generated graded modules over the graded affine Hecke algebra (Eberhardt, 2018).

That same paper places its construction explicitly in the direction of generalized Springer correspondences and relative constructions. Its equivariant motivic formalism is described as flexible enough to accommodate equivariant settings, to replace the full flag variety by partial flag varieties, and to interpret the resulting motives as analogues of Springer motives for Levi or parabolic data. In this sense the motivic package is presented as a geometric enhancement of classical and generalized Springer correspondences (Eberhardt, 2018).

A $(G,P_i,V,F_i)_{i\in I}$08-theoretic categorical refinement appears in “K-motives, Springer Theory and the Local Langlands Correspondence”. There the main objects are reduced $(G,P_i,V,F_i)_{i\in I}$09-motives on linearly reductive stacks, equipped with a six-functor formalism

$(G,P_i,V,F_i)_{i\in I}$10

Within that framework the paper defines Springer $(G,P_i,V,F_i)_{i\in I}$11-motives, proves formality for categories of Springer $(G,P_i,V,F_i)_{i\in I}$12-motives, and uses Steinberg-type correspondences to realize representation categories of affine Hecke algebras and split reductive $(G,P_i,V,F_i)_{i\in I}$13-adic groups. The work describes this as a relative Springer theory because the construction is performed in stacky and family-based settings rather than only on an absolute nilpotent cone, and because it is designed to interface with coherent Springer theory through a categorical Chern character (Eberhardt, 2024).

6. Generalized geometries, induced representations, and limitations

A broad generalization replaces the nilpotent cone by an arbitrary conical symplectic singularity admitting a symplectic resolution

$(G,P_i,V,F_i)_{i\in I}$14

Using Namikawa’s universal Poisson deformations, one obtains a finite Galois covering $(G,P_i,V,F_i)_{i\in I}$15 with Galois group $(G,P_i,V,F_i)_{i\in I}$16, called the symplectic Galois group. The associated sheaves

$(G,P_i,V,F_i)_{i\in I}$17

are the symplectic Harish-Chandra and symplectic Springer sheaves. The theory produces a faithful $(G,P_i,V,F_i)_{i\in I}$18-action on $(G,P_i,V,F_i)_{i\in I}$19 and hence on every fiber cohomology $(G,P_i,V,F_i)_{i\in I}$20, with monodromy supplied by nearby cycles of the deformation family. The paper presents this explicitly as relative and generalized Springer theory because the representation arises from a family over a base and from monodromy of that family, not only from a fixed nilpotent cone (McGerty et al., 2019).

The same paper also stresses limitations. Unlike the classical case, no perfect Springer correspondence is guaranteed: $(G,P_i,V,F_i)_{i\in I}$21 can be smaller or larger than $(G,P_i,V,F_i)_{i\in I}$22, and multiplicity spaces in the Springer decomposition need not be irreducible. The framework is Springer-like rather than a one-to-one correspondence between local systems and irreducible $(G,P_i,V,F_i)_{i\in I}$23-representations (McGerty et al., 2019).

A combinatorial and geometric variant appears in the $(G,P_i,V,F_i)_{i\in I}$24-Springer varieties $(G,P_i,V,F_i)_{i\in I}$25, which generalize type $(G,P_i,V,F_i)_{i\in I}$26 Springer fibers. Their cohomology ring has an explicit presentation

$(G,P_i,V,F_i)_{i\in I}$27

and because the ideal $(G,P_i,V,F_i)_{i\in I}$28 is stable under permuting the variables, the quotient $(G,P_i,V,F_i)_{i\in I}$29 carries an $(G,P_i,V,F_i)_{i\in I}$30-action that transfers to cohomology. The paper identifies this action as the natural analogue of the Springer action and proves that, for $(G,P_i,V,F_i)_{i\in I}$31,

$(G,P_i,V,F_i)_{i\in I}$32

while in the extremal case the top cohomology is a skew Specht module. The work describes the construction as a genuine “relative Springer theory” for the Delta Conjecture because top cohomology yields induced modules built from Specht modules and trivial factors, reflecting the extra relative geometry (Griffin et al., 2021).

A further refinement of the classical geometry comes from stable bases of the Springer resolution $(G,P_i,V,F_i)_{i\in I}$33. Those chamber-dependent bases are described as a kind of relative Springer theory because they enrich classical Springer theory with canonical chamber-by-chamber bases, Hecke algebra actions, wall-crossing, localization formulas, characteristic cycles of Verma modules, and links to modular and $(G,P_i,V,F_i)_{i\in I}$34-adic representation theory (Su et al., 2019).

Taken together, these developments show that relative Springer theory is a program rather than a single theorem. Across the literature, the recurring features are a Springer-type resolution or correspondence, a symmetry algebra or group that is relative to the geometry at hand, and a passage from fiber geometry to representations. The main open fronts are equally recurrent: a full sheaf-theoretic correspondence for symmetric spaces remains incomplete (Leslie, 2019), the affine local constancy conjecture still awaits the full microlocal machinery needed to deduce it from singular-support bounds (Bouthier, 2022), and in generalized symplectic settings the multiplicity spaces need not satisfy the irreducibility properties familiar from the classical correspondence (McGerty et al., 2019).