Unified Theory of Cartan Subgroups

• Unified Theory of Cartan Subgroups is a framework classifying maximal nilpotent subgroups across Lie, locally compact, and algebraic groups by integrating Chevalley, Lie-theoretic, and pro-Lie methods.
• It employs structural decompositions like the Levi and Wüstner theorems to delineate the interplay between semisimple, solvable, and nilpotent components.
• The theory establishes cohomological and density criteria that govern power map behavior, exponentiality, and quotient structures, enhancing our understanding of group representations.

A unified theory of Cartan subgroups encompasses their classification, existence, and structural properties across locally compact groups, Lie groups, and affine algebraic groups. The theory synthesizes the Chevalley, Lie-theoretic, and pro-Lie approaches, including the correspondence with maximal toroids in the algebraic context, and is fundamental for understanding the internal geometry, quotient behavior, and generation questions in group theory. Additionally, it establishes cohomological frameworks and density criteria pivotal for exponentiality.

1. Foundational Definitions and Decompositions

For a connected Lie group $G$, let $R$ be the maximal connected solvable normal subgroup and $N \subset R$ the maximal connected nilpotent normal subgroup. The Levi decomposition is given by

$$G = S R, \qquad \mathfrak{g} = \mathfrak{s} \oplus \mathfrak{r}$$

where $S$ is a maximal connected semisimple subgroup. The root-space decomposition with respect to a Cartan subalgebra $\mathfrak{c} \subset \mathfrak{g}$ writes

$$\mathfrak{g}_{\mathbb{C}} = \mathfrak{c}_{\mathbb{C}} \oplus \bigoplus_{\alpha\in\Delta}\mathfrak{g}_\alpha$$

where $$\mathfrak{g}_\alpha = \{ X : [H, X] = \alpha(H) X \}$$ for roots $\alpha$.

Cartan subgroup (Chevalley): A closed subgroup $C \leq G$ is Cartan if (i) $C$ is maximal among nilpotent subgroups, and (ii) whenever $L \leq C$ is a normal subgroup of finite index, $[N_G(L):L]<\infty$ (Mandal et al., 2020, Mandal et al., 2023). Equivalently, $C$ is the normalizer of a Cartan subalgebra with $N_G(C)/C$ finite.

For affine algebraic groups $G$ over a field $k$, a Cartan subgroup is a closed, connected, maximal nilpotent subgroup. The toroid–Cartan correspondence states that Cartan subgroups are precisely the identity component of the centralizer of a maximal toroid (Sercombe, 22 Jan 2026).

2. Structural Theorems and Levi-Type Decompositions

The Levi decomposition extends to Cartan subgroups, encapsulated in the Wüstner theorem and its generalizations.

Wüstner decomposition: Given $G = S R$ with $S$ semisimple, every Cartan subgroup $C$ admits a unique factorization

$C = (C \cap S)(C \cap R)$

where $C_S := C \cap S$ is a Cartan subgroup of $S$ and $C_R := C \cap R$ is connected, nilpotent, and centralizes $C_S$ (Mandal et al., 2020, Mandal et al., 2023). More generally [Mandal–Shah theorem], given any Cartan subgroup $C_S$ of $S$, the centralizer $Z_R(C_S)$ is connected, and any Cartan $C_{Z_R(C_S)}$ yields a Cartan subgroup $C = C_S\,C_{Z_R(C_S)}$ of $G$.

In locally compact groups, the decomposition $G = S \ltimes R$ holds, and Cartan subgroups satisfy $C = (C \cap S)(C \cap R)$ with analogous properties. The centralizer $Z_R(C_S)$ is always connected and absorbs the radical (solvable part) (Mandal et al., 2023).

3. Construction and Classification in Varied Contexts

Solvable Lie groups: Cartan subgroups can be constructed from any nilpotent complement $L$ of $N$ via the iterative process

$L_1 = N_G(L),\quad L_2 = N_G(L_1),\ldots$

stabilizing to a maximal connected nilpotent subgroup $L_m$ that is Cartan (Mandal et al., 2020).

Affine algebraic groups: Every toroid is contained in a maximal toroid, and hence every $G$ admits a Cartan subgroup (Sercombe, 22 Jan 2026). Base-change invariance ensures the correspondence holds under extension of scalars.

Pro-Lie and locally compact groups: The existence of Cartan subgroups extends via projective limits and maximal compact normal subgroups (Mandal et al., 2023). If $K \triangleleft G$ is the maximal compact normal subgroup with $G/K$ a Lie group, Cartans are constructed from Cartans of $Z^*(K)$ and those of simple components $K_\alpha$.

4. Cohomological, Quotient, and Generation Properties

Quotient behavior: For any closed normal $H \triangleleft G$, Cartan subgroups descend to quotients: If $C \leq G$ is Cartan, then $\pi(C) = CH/H$ is Cartan in $G/H$, and every Cartan subgroup of $G/H$ is of this form (Mandal et al., 2020, Mandal et al., 2023).

Toroid–Cartan correspondence:

$$\{\text{maximal toroids }T \subset G\} \longleftrightarrow \{\text{Cartan subgroups }C \subset G\}$$

via $T \mapsto Z_G(T)^\circ$ and $C \mapsto Z(C)^\circ_s$, where $Z(-)$ denotes the schematic centralizer, and maximal toroids yield Cartans and vice versa (Sercombe, 22 Jan 2026).

Generation questions: In classical cases, Cartans generate the group ($G = G_c$), equivalently the group has a unique maximal toroid if and only if $G$ is nilpotent (Sercombe, 22 Jan 2026). In non-smooth or non-nilpotent cases, generation by Cartans can fail.

5. Power Maps, Exponentiality, and Density Criteria

Power map density: For $P_k : G \to G$, $P_k(g) = g^k$, the image $P_k(G)$ is dense if and only if $P_k(C) = C$ for every Cartan $C$ (Mandal et al., 2020, Mandal et al., 2023). In locally compact groups, density of $P_k$ on all Cartan subgroups is equivalent to density in $G$; several equivalent criteria involve reductions to quotients, the radical, and semisimple or compact factors.

Weak exponentiality: $G$ is weakly exponential (density of $\exp \mathfrak{g}$ in $G$) if and only if every Cartan subgroup of $G$ is connected (Mandal et al., 2023). For connected nilpotent Cartans, exponentiality follows.

6. Galois Cohomology, Cartan Cohomology, and Real Forms

For a complex reductive group $G$, real forms are classified by nonabelian Galois cohomology $H^1(\Gamma, G)$, where $\Gamma = \operatorname{Gal}(\mathbb{C}/\mathbb{R})$. Cartan's classification via maximal compact subgroups translates to Cartan cohomology $H^1(\mathbb{Z}/2Z, G)$ with Cartan involution $\theta$ (Adams et al., 2016). The canonical isomorphism

$$H^1\bigl(\operatorname{Gal}(\mathbb{C}/\mathbb{R}), G\bigr) \simeq H^1(\mathbb{Z}/2Z, G)$$

unifies approaches to real group classification, Matsuki duality, and conjugacy of Cartan subgroups.

Borovoi’s theorem: For a fundamental $\sigma$–stable Cartan $H_f$, $H^1(\sigma, G) \simeq H^1(\sigma, H_f)/W_i$, with $W_i$ the Weyl group of imaginary roots. In equal-rank cases, $H^1(\theta, G)$ corresponds to 2-torsion in a single Cartan modulo Weyl group.

7. Synthesis and Conceptual Roadmap

The unified theory asserts that Cartan subgroups in a connected group are classified by:

• Cartans in chosen Levi factors (semisimple parts),
• Cartans in the centralizer within the radical (solvable part),
• Explicit normalizer constructions from nilpotent seeds in solvable groups,
• Compatibility with all closed normal quotients,
• Power map density criteria ($P_k(C) = C$ for all $k$).

This scheme provides a transparent paradigm applicable to Lie groups, locally compact groups, and algebraic group schemes. The foundational principle is: Cartan subgroups are maximal nilpotent subgroups preserved by centralization in the radical and under quotients, encoding the fine structure that governs exponentiality, representation theory, and real forms.

---

Key References:

• Mandal–Shah, "The Structure of Cartan Subgroups in Lie Groups" (Mandal et al., 2020)
• Adams–Taïbi, "Galois and Cartan Cohomology of Real Groups" (Adams et al., 2016)
• "Cartan subgroups in connected locally compact groups" (Mandal et al., 2023)
• "Maximal toroids and Cartan subgroups of algebraic groups" (Sercombe, 22 Jan 2026)