Satake Groupoid

• Satake groupoid is a Lie groupoid arising from the Satake compactification, encapsulating the global geometry and tempered representations of real reductive Lie groups.
• It admits three equivalent constructions—topological, Lie-theoretic, and geometric b-groupoid—that clarify different facets of the compactification and group action gluing.
• Its C*-algebraic framework underpins Harish-Chandra’s parabolic induction, bridging discrete series and parabolic-induced representations.

The Satake groupoid is a Lie groupoid intrinsically associated with the Satake compactification of a Riemannian symmetric space of noncompact type, providing a unifying and functorial framework for the structural analysis of tempered representations of real reductive Lie groups. It admits three equivalent constructions—topological, Lie-theoretic, and geometric (“b-groupoid”)—and its $C^*$-algebraic structure yields a continuous field over the corner stratification of the compactification, directly underpinning Harish-Chandra’s parabolic induction principle for the tempered dual. The Satake groupoid encapsulates the gluing of group actions, compactifications, and boundary strata, thus encoding both the global geometry and representation theory of real reductive groups in a groupoid-theoretic language (Bradd et al., 27 Nov 2025, Bradd et al., 27 Nov 2025).

1. Satake Compactification and Underlying Structure

Let $G$ be a real reductive Lie group with maximal compact subgroup $K$, and $G/K$ the associated Riemannian symmetric space. The maximal Satake compactification $X$ is constructed as the closure in the Chabauty–Fell topology of the $G$-orbit of $K$ in $S(G)$, the compact space of closed subgroups of $G$: $$X = \overline{X_\Sigma} \subset S(G), \qquad X_\Sigma = \{ K \subset G : K \text{ maximal compact} \}$$ Alternatively, following Oshima’s model, $X$ is realized as a quotient

$$X = \bigl( G \times \mathbb{R}^\Sigma_{\geq 0} \bigr) \big/ \sim \big/ A$$

where $\Sigma$ is the set of simple restricted roots (relative to a fixed Cartan decomposition), $$t = (t_\alpha)_{\alpha\in\Sigma} \in \mathbb{R}^\Sigma_{\geq 0}$$ indexes boundary faces, and the equivalence is by the action of closed subgroups $H_t = K_{I(t)} N_{I(t)}$ (with $I(t) = \{ \alpha\in\Sigma : t_\alpha = 0 \}$ for the associated standard parabolic). $X$ is a compact $G$-space, stratified with interior $G/K$ and faces modeled on double cosets $G/K_I A_I N_I$.

2. Topological Construction via Mohsen’s Coset Groupoid

The topological construction employs the coset groupoid framework due to Omar Mohsen, specializing the following data:

• Objects: Points of $X$, i.e., closed subgroups $S \in X$.
• Arrows: Cosets $gS$ for $g\in G$, $S\in X$.

The groupoid structure is given by

$s(gS) = S,\quad t(gS) = gSg^{-1},$

$$(g_1S_1)\circ (g_2S_2) = g_1S_1g_2S_2 \quad\text{when}\quad S_1 = g_2S_2g_2^{-1}.$$

Thus,

$G_X \cong \bigsqcup_{S\in X} G/S.$

Equipped with the Fell topology, $(G_X\rightrightarrows X)$ becomes a locally compact Hausdorff groupoid with open source and target maps (Bradd et al., 27 Nov 2025).

3. Lie-Theoretic (Oshima) Model

Adopting Cartan and Iwasawa splittings ($G=KAN$) and restricting to $R^\Sigma = \mathbb{R}^\Sigma$, Oshima defines:

• Subgroups $H_t$: $H_t = \text{Ad}_{a_t}(K)$ for $t \in (R^\Sigma)_{\times}$, extended to general $t$ by closure under $|t|$.
• Lie Algebra:

$$\mathfrak{h}_t = \mathfrak{m} \oplus \bigoplus_{\gamma\in\Delta^+(\mathfrak{g},\mathfrak{a})} \{ t^{2\gamma} X + \theta(X) : X \in \mathfrak{g}_\gamma \}$$

The compact manifold with corners $M$ is

$$M = (G/H)/A,\quad \text{where } G/H = \{ [g,t] : (g_1,t_1)\sim (g_2,t_2)\iff t_1 = t_2,\,g_2^{-1}g_1\in H_{t_1} \}.$$

A transformation groupoid $G \ltimes M$ is then factored by the normal subgroupoid $H_M$ to define the Oshima groupoid $G_M = (G \ltimes M)/H_M \rightrightarrows M$. Restriction to the submanifold $\overline{M_+} \cong X$ gives an identification with $G_X$ of the topological model (Bradd et al., 27 Nov 2025).

4. Geometric b-Groupoid and Global Correspondence

The Oshima manifold $M$ carries a simple normal crossing divisor $S_\alpha = \{ [g,t] : t_\alpha=0 \}$ for each $\alpha\in\Sigma$. The b-tangent bundle $T^b M$ comprises vector fields tangent to all $S_\alpha$.

Following Monthubert and Nistor–Weinstein–Xu, the b-groupoid $\Gamma(M)\rightrightarrows M$ integrates $T^b M$. Local charts are modeled on groupoids

$$\left\{ ((n_2, t_2), a, (n_1, t_1))\mid t_2 = a t_1 \right\} \subset N \times \mathbb{R}^\Sigma \times A \times N \times \mathbb{R}^\Sigma.$$

There exists an isomorphism of Lie groupoids $G_M \cong \Gamma(M)$ via appropriate restriction maps on normal bundles. Consequently, the reduction to $X$ (i.e., $\overline{M_+}$) yields a canonical identification among the topological, Oshima, and b-groupoid constructions (Bradd et al., 27 Nov 2025).

5. Local Structure, Haar Systems, and Groupoid $C^*$-Algebra

On an open cell $U \subset X$ diffeomorphic to $N \times \mathbb{R}^\Sigma_{\ge 0}$, the groupoid $G_X$ takes the local form

$$(N \times \mathbb{R}^\Sigma_{\ge 0}) \times G / \sim \rightrightarrows N \times \mathbb{R}^\Sigma_{\ge 0},$$

with gluing data inherited from $A$-equivariance and parabolic factorization. For each $x\in X$, a Haar measure $\mu^x$ is specified by integrating over $H_x \backslash G$, modified by a Radon–Nikodym cocycle.

The convolution *-algebra $C_c(G_X)$ of compactly supported continuous functions is completed in the reduced norm arising from the regular representations $\lambda_x$ on $L^2$-spaces over groupoid source fibers, giving the reduced groupoid $C^*$-algebra $C^*_r(G_X)$ (Bradd et al., 27 Nov 2025).

6. Applications: Tempered Dual and Harish-Chandra’s Principle

The $C^*$-algebra $C^*_r(G_X)$ forms a continuous field over the strata of $X$, with fibers at boundary faces naturally isomorphic to crossed-product $C^*$-algebras of parabolic subgroups. There exists an integration $*$-homomorphism

$\iota\colon C^*_r(G) \to C^*_r(G_X)$

that intertwines regular representations. This morphism relates to two distinguished ideals in $C^*_r(G)$:

• The compact-mod-center ideal $C^*_{cmc}(G)$ (corresponding to discrete series modulo center).
• The cuspidal ideal $C^*_{cusp}(G)$ (complement of all proper parabolic inductions).

The essential property is the coincidence of these two ideals via the Satake groupoid, realizing Harish-Chandra's induction principle: every irreducible tempered representation of $G$ arises either as discrete series modulo center or through parabolic induction from such representations (Bradd et al., 27 Nov 2025).

7. Example: Rank-One Case and Structure Decomposition

For $G = \mathrm{SL}(2, \mathbb{R})$, the Satake compactification of the upper half-plane is the closed disk $\overline{\mathbb{H}} = \mathbb{H} \cup S^1$. The Satake groupoid $G_X$ is a blow-up of the pair groupoid $S^1 \times S^1$ at the preimages of the two cusps, and its $C^*$-algebra decomposes into subalgebras corresponding to discrete series and principal series, reflecting the direct sum decomposition of the tempered dual in this setting (Bradd et al., 27 Nov 2025).

---

Summary Table: Constructions of the Satake Groupoid

Construction	Underlying Data	Key References
Topological (Mohsen/Fell)	Cosets $gS$ for $S \in X$	(Bradd et al., 27 Nov 2025)
Lie-theoretic (Oshima)	$M = (G/H)/A$, $H_t$	(Bradd et al., 27 Nov 2025)
Geometric (b-groupoid)	$T^b M$, normal crossings	(Bradd et al., 27 Nov 2025)

All three models are canonically isomorphic via explicit identifications, providing a robust platform for geometric representation theory of real reductive groups.