--- title: Generalized Affine Springer Fibers url: https://www.emergentmind.com/topics/generalized-affine-springer-fibers type: topic --- # Generalized Affine Springer Fibers Searching arXiv for recent and foundational papers on generalized affine Springer fibers to ground the article in the literature. arXiv.search(query="generalized affine Springer fibers", max_results=10, sort_by="relevance") Searching arXiv for exact and related titles. arxiv_search("generalized affine Springer fibers") Generalized affine Springer fibers are affine-Springer-type moduli spaces obtained by enlarging the classical condition \(g^{-1}\gamma g\in \mathfrak g(O)\) or \(g^{-1}\gamma g\in \operatorname{Lie}(P)\) in several different directions. In the literature, the phrase is used for several closely related constructions: Lie-algebraic fibers attached to arbitrary \(G\)-representations and parahorics, group-theoretic analogues attached to regular semisimple elements of \(G(F)\) and dominant coweights, root-valuation-adapted variants, multiplicative and parabolic versions in affine flag varieties, and type-\(A\) realizations by Hilbert schemes of planar curve singularities [2204.00303] [1710.11243] [1003.2135] [2410.09651] [2004.15024]. Across these variants, the central theme is the replacement of a single affine Springer fiber by a larger package of geometric objects carrying actions of Weyl groups, Picard groups, Coulomb-branch algebras, or Cherednik algebras. ## 1. Classical source and principal definitions For a connected reductive group \(G\) over an algebraically closed field \(k\), with \(K=k((t))\), \(O=k[[t]]\), affine Grassmannian \(\operatorname{Gr}_G=G(K)/G(O)\), and affine flag variety \(\operatorname{Fl}_G=G(K)/I\), the classical affine Springer fiber attached to \(y\in\mathfrak g(K)\) and a standard parahoric \(P\subset G(K)\) is \[ \operatorname{Sp}^P_y = \{gP\in G(K)/P\mid \operatorname{Ad}(g^{-1})y\in \operatorname{Lie}(P)\}. \] The two basic cases are \(P=G(O)\), giving \(\operatorname{Sp}^{G(O)}_y\subset \operatorname{Gr}_G\), and \(P=I\), giving \(\operatorname{Sp}_y\subset \operatorname{Fl}_G\). When \(y\) is compact regular semisimple, \(\operatorname{Sp}_y\) is finite-dimensional and of finite type; for general semisimple \(y\) it is an ind-scheme with a free lattice action [2204.00303]. Several generalizations coexist. | Variant | Defining data | Fiber | |---|---|---| | Representation-theoretic | \(G\), a representation \(N\), a parahoric \(P\), a \(P\)-stable lattice \(N_P\subset N\otimes K\), \(v\in N\otimes K\) | \(\eta\mathcal M_v=\{gP\mid g^{-1}.v\in N_P\}\) | | Group-theoretic | \(\gamma\in G(F)^{\mathrm{rs}}\), \(\lambda\in\Lambda^+\) | \(X^\lambda_\gamma=\{gG(O)\mid g^{-1}\gamma g\in G(O)\varpi^\lambda G(O)\}\) | | Root-valuation adapted | root valuation function \(r\), building point \(x\), or lattice datum \((r,\lambda)\) | \(Y_{r,x}(u)\), \(Z_{r,\lambda}(u)\) | | Multiplicative parabolic | \(\gamma\in G^{\mathrm{rs}}(F)\), \(\lambda\in X_*(T)^+\) | \(X^{\lambda,\para}_\gamma\), \(X^{\le\lambda,\para}_\gamma\) | A recurrent misconception is that “generalized” designates a single canonical definition. The literature instead uses the term for a family of constructions sharing affine-Springer-type lattice or conjugacy conditions but adapted to different geometric and representation-theoretic problems. ## 2. Representation-theoretic generalized fibers and the fiber–sheaf correspondence A particularly flexible formulation replaces the adjoint representation by an arbitrary \(G\)-representation. Given a parahoric \(P\subset G(K)\), a \(P\)-stable lattice \(N_P\subset N\otimes K\), and \(\eta=(P,N_P)\), the generalized affine Springer fiber is \[ \eta\mathcal M_v = \{gP\in \operatorname{Fl}_P\mid g^{-1}.v\in N_P\}. \] The classical affine Springer fiber is recovered by taking \(N=\operatorname{Ad}\) and \(N_P=\operatorname{Lie}(P)\). This enlargement is the framework in which the Coulomb-branch perspective becomes natural and in which the affine Springer fiber–sheaf correspondence is formulated [2204.00303]. The basic construction organizes not one fiber but the graded family \(\{\operatorname{Sp}_{t^m y}\}_{m\ge 0}\). Taking Borel–Moore homology produces \[ F_y:=\bigoplus_{i\ge 0} H_*\big(\operatorname{Sp}_{t^i y}\big), \] and the Coulomb-branch correspondences give graded maps \[ \mathcal A_d^{h=0}\otimes H_*\big(\operatorname{Sp}_{t^m y}\big) \to H_*\big(\operatorname{Sp}_{t^{m+d}y}\big). \] Hence \(F_y\) becomes a graded module over the commutative graded algebra \[ \mathcal A^{h=0}=\bigoplus_{d\ge 0}\mathcal A_d^{h=0}. \] Passing to \(\operatorname{Proj}\), one obtains a quasi-coherent sheaf on \[ \mathcal C_{\check G}=\operatorname{Proj}\mathcal A^{h=0}, \] a canonical partial resolution of the trigonometric commuting variety of the Langlands dual group \(\check G\). The main theorem asserts that for a semisimple \(y\in\mathfrak g((t))\) there is a quasi-coherent sheaf \(F_y\in \operatorname{QCoh}^{\mathbb C^\times}(\mathcal C_{\check G})\) such that, for \(m\gg 0\), \[ H^0\big(\mathcal C_{\check G}, F_y\otimes \mathcal O(m)\big)\cong H_*\big(\operatorname{Sp}_{t^m y}\big). \] This packages the whole shifted family into a single sheaf-theoretic object on the dual side [2204.00303]. For \(G=GL_n\), the partial resolution is identified with \[ \mathcal C_{\check G}\cong \operatorname{Hilb}^n(\mathbb C^\times\times \mathbb C). \] In this type-\(A\) case, homogeneous elements admit explicit sheaf descriptions: for integral slope \(k\), \[ F_y\simeq \mathcal P\otimes\mathcal O(k), \] with \(\mathcal P\) the Procesi bundle; for slope \(\frac{kn+1}{n}\), \[ F_y\cong \mathcal O(k)\big|_{\operatorname{Hilb}^n_0(\mathbb C^\times\times\mathbb C)}. \] These formulas make the generalized theory concrete and connect it directly to Cherednik representation theory and Hilbert-scheme geometry [2204.00303]. ## 3. Group-theoretic and multiplicative variants A distinct but closely related generalization replaces Lie-algebra elements by group elements. For a split connected reductive group \(G\), a regular semisimple \(\gamma\in G(F)\), and a dominant coweight \(\lambda\in\Lambda^+\), the group-type generalized affine Springer fiber is \[ X^\lambda_\gamma(k) = \{gG(O)\in G(F)/G(O)\mid g^{-1}\gamma g\in G(O)\varpi^\lambda G(O)\}. \] This is an ind-scheme locally closed in the affine Grassmannian, but it is in fact a finite-dimensional \(k\)-scheme locally of finite type. Its non-emptiness is equivalent to each of the following conditions: \(X^\lambda_\gamma\neq\varnothing\), \(\nu_\gamma\le_{\mathbb Q}\lambda\), and \(\chi(\gamma)\in C_{\le\lambda}\), where \(\nu_\gamma\) is the Newton point and \(\chi\) is the Steinberg map [1710.11243]. The geometry of \(X^\lambda_\gamma\) is controlled by a commutative group \(P_a\) attached to the regular centralizer. The variety is equidimensional, with \[ \dim X^\lambda_\gamma = \langle \rho,\lambda\rangle+\frac{1}{2}\bigl(d(\gamma)-c(\gamma)\bigr), \] where \(d(\gamma)\) is the discriminant valuation and \(c(\gamma)=\operatorname{rank}(G)-\operatorname{rank}_F(G_\gamma)\). Unlike the classical Lie-algebra case, the regular locus \(X^{\lambda,\mathrm{reg}}_\gamma\) need not be dense, and the \(P_a\)-action on it is not globally transitive; most irreducible components can be irregular. This difference is central, not incidental, and it is one of the main ways in which the group theory departs from the classical affine Springer picture [1710.11243]. The same paper formulates a representation-theoretic conjecture: if \(\mu\) is the smallest dominant integral coweight above the rational Newton point \(\nu_\gamma\), then the number of \(G_\gamma(F)\)-orbits on \(\operatorname{Irr}(X^\lambda_\gamma)\) should equal the weight multiplicity \(m_\lambda(\mu)\) in the irreducible representation \(V(\lambda)\) of the Langlands dual group \(\widehat G\). This is proved in the unramified case, where \(\nu_\gamma=\mu\in\Lambda^+\) [1710.11243]. A multiplicative and parabolic extension replaces the affine Grassmannian by the affine flag variety and \(G(O)t^\lambda G(O)\) by admissible unions of Iwahori double cosets. For \(\gamma\in G^{\mathrm{rs}}(F)\) and \(\lambda\in X_*(T)^+\), \[ \begin{aligned} X^{\lambda,\para}_\gamma &= \{gI\in \operatorname{Fl}_G(k)\mid g^{-1}\gamma g\in \bigcup_{\mu\in W(\lambda)} It^\mu I\},\\ X^{\le \lambda,\para}_\gamma &= \{gI\in \operatorname{Fl}_G(k)\mid g^{-1}\gamma g\in \bigcup_{w\in \operatorname{Adm}(\lambda)} IwI\}. \end{aligned} \] These are non-Frobenius-twisted analogues of admissible unions of affine Deligne–Lusztig varieties. They are equidimensional, and their dimension agrees with the multiplicative affine Springer fiber: \[ \dim X^{\lambda,\para}_\gamma = \langle \rho,\lambda\rangle+\frac{d(\gamma)-c(\gamma)}{2}. \] Thus, adding parabolic structure refines the stratification but does not change the dimension formula [2410.09651]. ## 4. Root valuations, Hodge–Newton decompositions, and root-valuation lattices Another influential meaning of generalized affine Springer fiber is adapted to the root-valuation stratification of Goresky–Kottwitz–MacPherson. For a split connected reductive group \(G\) over \(\mathbb C\), maximal torus \(A\), and regular \(u\in\mathfrak a(F)\), the root valuation function is \[ r_u:R\to \mathbb Z,\qquad r_u(\alpha)=\operatorname{val}(\alpha(u)). \] A function \(r:R\to\mathbb Z\) is a root valuation function if, for every integer \(m\), the subset \(R_m=\{\alpha\in R: r(\alpha)\ge m\}\) is \(\mathbb Q\)-closed. The corresponding root-valuation stratum in \(\mathfrak g(F)\) consists of split regular semisimple elements whose root-valuation function is in the \(W\)-orbit of \(r\) [1003.2135]. The first generalized space in this setting is the Hodge–Newton-type variety \[ Y_{r,x} = \{(u,gK_x)\in \mathfrak g(F)\times G(F)/K_x\mid \mu(\operatorname{ad}(u),\mathfrak k_{gx})\le \tilde r\}, \] where \(x\) is a point in the building, \(K_x\) the associated parahoric, and \(\tilde r\) the ordered Newton-point datum extracted from \(r\). If \(r'\) is weakly equivalent to \(r\) and \(u\in\mathfrak a(F)_{r'}\), then \(Y_{r,x}(u)\) consists precisely of those \(gK_x\) for which \(gx\) lies in the apartment of \(A\); when \(x\) is in the interior of an alcove, this identifies \(Y_{r,x}(u)\) with the extended affine Weyl group. The point of the construction is that, over a fixed root-valuation stratum, the fiber becomes as simple as possible [1003.2135]. The second construction uses root-valuation lattices \[ A_{r,\lambda} = \mathfrak a(F)_{\ge r}\oplus \bigoplus_{\alpha\in R} \mathfrak p_\alpha(\lambda(\alpha)), \] with stabilizer \(K_{r,\lambda}\). The associated generalized affine Springer fiber is \[ Z_{r,\lambda}(u) = \{gK_{r,\lambda}\in G(F)/K_{r,\lambda}\mid \operatorname{Ad}(g)^{-1}u\in A_{r,\lambda}\}. \] Here \(K_{r,\lambda}\) is typically not parahoric. A detailed characterization of when \(A_{r,\lambda}\) is a root-valuation lattice is given by inequalities involving \(\kappa=\lambda-r\), and the Conjugation Theorem shows that, under a cardinality condition on level sets of root-valuation functions, any \(u\in A_{r,\lambda}\) lying in a root-valuation stratum \(\mathfrak g(F)_{r'}\) is \(K_{r,\lambda}\)-conjugate to an element of \(\mathfrak a(F)_r\). This makes \(A_{r,\lambda}\) a model neighborhood of the chosen stratum [1003.2135]. These constructions are built on linear versions of Katz’s Hodge–Newton decomposition, both for general \(F\)-linear endomorphisms and for split reductive groups, and they show that “generalized affine Springer fiber” can mean a space adapted not to a fixed \(\gamma\) alone, but to an entire root-valuation profile. ## 5. Type \(A\), Hilbert schemes, Coulomb branches, and Cherednik algebras For \(G=GL_n\), generalized affine Springer fibers acquire especially explicit realizations. A central example uses the representation \(N=\operatorname{Ad}\oplus V\), where \(V=\mathbb C^n\). If \(C=\operatorname{Spec}R\) is a planar curve singularity with \[ f(x,t)=x^n-a_{n-1}(t)x^{n-1}-\cdots-a_0(t), \] and \(\gamma\in\mathfrak{gl}_n(O)\) is the companion matrix of \(f\), then for \(v=(\gamma,e_n)\) the generalized affine Springer fiber \[ M_v^{P,N}(A) = \{g\in G(A((t)))\mid g^{-1}\cdot v\in N(A)\}/P(A) \] recovers Hilbert schemes of points on \(C\). In the spherical case one has \[ M_v \cong \operatorname{Hilb}^\bullet(C), \] and in the parahoric case \[ M_v^P \cong \operatorname{PHilb}^{[\bullet,\bullet+\vec p]}(C), \] the parabolic flag Hilbert scheme. This identifies Hilbert schemes and their parabolic variants as generalized affine Springer fibers for \(GL_n\) [2004.15024]. The same paper extends Braverman–Finkelberg–Nakajima Coulomb-branch methods to these spaces. The corresponding Coulomb-branch algebra \(A^\hbar_{G,N,P,N}\) acts on \(H_*^{L_v}(M_v^{P,N})\). For \(G=GL_n\) and \(N=\operatorname{Ad}\oplus V\), the spherical Coulomb branch is identified with the spherical rational Cherednik algebra, and the Iwahori version with the full rational Cherednik algebra. Consequently, the homology of \(\operatorname{Hilb}^\bullet(C)\) and of the parabolic Hilbert schemes carries natural rational Cherednik actions [2004.15024]. A different type-\(A\) development computes Borel–Moore homology of unramified affine Springer fibers for \(GL_n\), under an equivariant-formality assumption, in terms of generalized Haiman ideals \[ J=\bigcap_{i