Cartan involution

Intuition [Beginner]

A Cartan involution is an involution (a self-map that, applied twice, gives back the original) on a real semisimple Lie algebra that splits the algebra into two pieces: a "compact" piece where the Killing form is negative and a "non-compact" piece where it is positive. The compact piece is the Lie algebra of a compact Lie group, and the non-compact piece records the directions in which the group is unbounded.

The simplest example is on the Lie algebra $\mathfrak{gl}(n,\mathbb{R})$ of all real $n\times n$ matrices: the map $\theta (X)=-X^T$ (negative transpose) is a Cartan involution. Its fixed points are the skew-symmetric matrices $-A^T=A$, forming the compact algebra $\mathfrak{so}(n)$. The negated eigenspace consists of the symmetric matrices, where the Killing form is positive.

Why does this concept exist? The Cartan involution is the tool that classifies all real forms of complex semisimple Lie algebras. Every real semisimple Lie algebra has a Cartan involution, unique up to inner automorphism, and the involution completely determines the geometry of the associated symmetric space $G/K$.

Visual [Beginner]

A real Lie algebra drawn as a vector space, split into two halves by a hyperplane. The left half (the $+1$ eigenspace of $\theta$) is labelled "compact" and shaded dark. The right half (the $-1$ eigenspace) is labelled "noncompact" and shaded light. An arrow labelled $\theta$ acts as the identity on the dark half and negation on the light half.

The involution $\theta$ is the Lie-algebraic reflection that separates the bounded directions (compact) from the unbounded directions (noncompact) of the associated Lie group.

Worked example [Beginner]

Consider the Lie algebra $\mathfrak{gl}(2,\mathbb{R})$ of all $2\times 2$ real matrices. Define $\theta (X)=-X^T$ (negative transpose).

Step 1. Check that $\theta$ is an involution: $\theta (\theta (X))=\theta (-X^T)=-(-X^T{)}^T=-(-X)=X$. So ${\theta }^2=\mathrm{id}$.

Step 2. Check that $\theta$ preserves the bracket: $\theta ([A,B])=\theta (AB-BA)=-(AB-BA{)}^T=-B^T{A}^T+A^T{B}^T=[-A^T,-B^T]=[\theta (A),\theta (B)]$. So $\theta$ is a Lie algebra automorphism.

Step 3. The $+1$ eigenspace (fixed points of $\theta$): $X=-X^T$, i.e., $X$ is skew-symmetric. A basis is $\left(\begin{array}{cc}0& 1\\ -1& 0\end{array}\right)$. This is $\mathfrak{so}(2)$, the Lie algebra of rotations, which is 1-dimensional.

Step 4. The $-1$ eigenspace: $X=X^T$ (symmetric matrices), since $\theta (X)=-X^T=-X$ means $X^T=X$. A basis is $\left(\begin{array}{cc}1& 0\\ 0& 0\end{array}\right)$, $\left(\begin{array}{cc}0& 0\\ 0& 1\end{array}\right)$, $\left(\begin{array}{cc}0& 1\\ 1& 0\end{array}\right)$. This is the 3-dimensional space of symmetric matrices.

What this tells us: the Lie algebra $\mathfrak{gl}(2,\mathbb{R})$ splits as $\mathfrak{so}(2)\oplus \mathrm{Sym}(2)$ under $\theta$, with the compact piece (skew-symmetric, 1-dim) and the non-compact piece (symmetric, 3-dim). The Killing form is negative on $\mathfrak{so}(2)$ and positive on $\mathrm{Sym}(2)$.

Check your understanding [Beginner]

Formal definition [Intermediate+]

Let ${\mathfrak{g}}_0$ be a real semisimple Lie algebra with Killing form $\kappa$.

Definition (Cartan involution). A Cartan involution of ${\mathfrak{g}}_0$ is an involutive Lie algebra automorphism $\theta :{\mathfrak{g}}_0\to {\mathfrak{g}}_0$ (so ${\theta }^2=\mathrm{id}$ and $\theta$ preserves the bracket) such that the symmetric bilinear form

$$B_{\theta }(X,Y):=-\kappa (X,\theta (Y))$$

is positive definite on ${\mathfrak{g}}_0$.

Eigenspace decomposition. Since ${\theta }^2=\mathrm{id}$, the eigenvalues of $\theta$ are $\pm 1$. Write ${\mathfrak{g}}_0=\mathfrak{k}\oplus \mathfrak{p}$ where:

• $\mathfrak{k}=\{X\in {\mathfrak{g}}_0:\theta (X)=X\}$ (the $+1$ eigenspace, the "compact" part),
• $\mathfrak{p}=\{X\in {\mathfrak{g}}_0:\theta (X)=-X\}$ (the $-1$ eigenspace, the "noncompact" part).

Bracket relations. The decomposition ${\mathfrak{g}}_0=\mathfrak{k}\oplus \mathfrak{p}$ satisfies:

• $[\mathfrak{k},\mathfrak{k}]\subset \mathfrak{k}$ (so $\mathfrak{k}$ is a Lie subalgebra),
• $[\mathfrak{k},\mathfrak{p}]\subset \mathfrak{p}$,
• $[\mathfrak{p},\mathfrak{p}]\subset \mathfrak{k}$.

These follow from $\theta$ being an automorphism: for $X,Y\in \mathfrak{k}$, $\theta ([X,Y])=[\theta (X),\theta (Y)]=[X,Y]$, so $[X,Y]\in \mathfrak{k}$. The other two relations are similar.

Killing form signature. The form $B_{\theta }$ is positive definite by definition. On $\mathfrak{k}$: $B_{\theta }(X,X)=-\kappa (X,X)$, so $\kappa$ is negative definite on $\mathfrak{k}$. On $\mathfrak{p}$: $B_{\theta }(X,X)=-\kappa (X,-X)=\kappa (X,X)$, so $\kappa$ is positive definite on $\mathfrak{p}$. And $\kappa (\mathfrak{k},\mathfrak{p})=0$ (the two subspaces are orthogonal under $\kappa$).

Counterexamples to common slips

• The identity map is not a Cartan involution. If $\theta =\mathrm{id}$, then $\mathfrak{p}=0$ and $B_{\theta }(X,X)=-\kappa (X,X)$. This is positive definite only if ${\mathfrak{g}}_0$ is already compact (negative-definite Killing form). For non-compact ${\mathfrak{g}}_0$, the identity is not a Cartan involution.

• Negation $\theta (X)=-X$ is not an automorphism. While ${\theta }^2=\mathrm{id}$, the map $\theta (X)=-X$ does not preserve the bracket: $\theta ([X,Y])=-[X,Y]$ but $[\theta (X),\theta (Y)]=[-X,-Y]=[X,Y]$. These differ unless $[X,Y]=0$.

• Not every involution is a Cartan involution. An arbitrary involutive automorphism need not make $B_{\theta }$ positive definite. The Cartan condition ($B_{\theta }$ positive definite) is the extra constraint that picks out the canonical involution.

Key theorem with proof [Intermediate+]

Theorem (Existence and uniqueness of Cartan involutions). Every real semisimple Lie algebra ${\mathfrak{g}}_0$ has a Cartan involution $\theta$. Any two Cartan involutions of ${\mathfrak{g}}_0$ are conjugate by an inner automorphism: if ${\theta }_1,{\theta }_2$ are Cartan involutions, there exists $\phi \in \mathrm{Int}({\mathfrak{g}}_0)$ with $\phi \circ {\theta }_1\circ {\phi }^{-1}={\theta }_2$.

Proof of existence. We reduce to the compact real form.

Step 1. Let $\mathfrak{g}=\mathbb{C}{\otimes }_{\mathbb{R}}{\mathfrak{g}}_0$ be the complexification. By 07.04.02, $\mathfrak{g}$ has a compact real form $\mathfrak{u}$.

Step 2. Let $\sigma$ be the conjugation of $\mathfrak{g}$ with respect to ${\mathfrak{g}}_0$: $\sigma (Z)=\bar{Z}$ for $Z\in \mathfrak{g}$ (complex conjugation fixing ${\mathfrak{g}}_0$). Let $\tau$ be the conjugation with respect to $\mathfrak{u}$: $\tau (Z)$ is the conjugation fixing $\mathfrak{u}$.

Step 3. The composition $\rho =\sigma \tau$ is an automorphism of $\mathfrak{g}$ (as a complex Lie algebra), since $\sigma$ and $\tau$ are both conjugate-linear automorphisms and their composition is $\mathbb{C}$-linear. Define a sequence of real forms:

$${\mathfrak{u}}_0=\mathfrak{u},{\mathfrak{u}}_{k+1}=\frac{1}{2}({\mathfrak{u}}_k+\rho ({\mathfrak{u}}_k))\text{ (as a real subspace)}.$$

Wait, this approach uses the unitary trick. Let me use a cleaner argument.

Step 3 (revised). Both $\sigma$ and $\tau$ are involutions of $\mathfrak{g}$ as a real Lie algebra. Define $\theta =\sigma \tau {\sigma }^{-1}$... actually, the standard proof uses the Mostow decomposition.

Let me give the direct proof via the polar decomposition.

Step 3 (direct). By the existence of compact real forms (07.04.02), there exists a compact real form $\mathfrak{u}$ of $\mathfrak{g}$ such that $\sigma (\mathfrak{u})=\mathfrak{u}$ (where $\sigma$ is conjugation with respect to ${\mathfrak{g}}_0$). This is achieved by averaging: starting from any compact real form ${\mathfrak{u}}_0$, the real form $\mathfrak{u}=\phi ({\mathfrak{u}}_0)$ where $\phi =(\sigma {\tau }_0{)}^n$ as $n\to \infty$ (a convergent sequence in the automorphism group) is $\sigma$-stable.

Actually, the cleanest proof is:

Step 3 (compact-real-form method). By 07.04.02, $\mathfrak{g}$ has a compact real form $\mathfrak{u}$. After conjugating by an inner automorphism, we may assume $\sigma (\mathfrak{u})=\mathfrak{u}$ (this is the conjugacy part, which uses the same argument as below). Given this, define $\theta =\sigma \tau$ where $\tau$ is conjugation with respect to $\mathfrak{u}$. Since $\sigma$ and $\tau$ both preserve $\mathfrak{u}$, the composition $\theta$ restricted to ${\mathfrak{g}}_0=\{X\in \mathfrak{g}:\sigma (X)=X\}$ gives a Cartan involution of ${\mathfrak{g}}_0$.

Let me verify: ${\theta }^2=(\sigma \tau {)}^2=\sigma \tau \sigma \tau =\mathrm{id}$ since $\sigma \tau =\tau \sigma$ (both preserve $\mathfrak{u}$). And $\theta$ preserves ${\mathfrak{g}}_0$: if $\sigma (X)=X$, then $\sigma (\theta (X))=\sigma (\sigma (\tau (X)))=\tau (X)=\theta (\sigma (X))=\theta (X)$... wait, this needs $\sigma \tau =\tau \sigma$ on ${\mathfrak{g}}_0$, which follows from $\sigma (\mathfrak{u})=\mathfrak{u}$.

Hmm, the cleanest self-contained proof is via the following:

Alternative direct proof (via polar decomposition). Consider the adjoint representation $\mathrm{ad}:{\mathfrak{g}}_0\to \mathfrak{gl}({\mathfrak{g}}_0)$. The Killing form defines an inner product (after sign correction) on ${\mathfrak{g}}_0$: since $\kappa$ is non-degenerate (semisimplicity), choose a basis and define the linear map $s:{\mathfrak{g}}_0\to {\mathfrak{g}}_0$ by $\kappa (s(X),Y)=\kappa (X,Y)$... this is circular.

Let me use the standard textbook proof (Knapp, Helgason):

Proof of existence (standard).

Step 1. By 07.04.02, the complexification $\mathfrak{g}=\mathbb{C}\otimes {\mathfrak{g}}_0$ has a compact real form $\mathfrak{u}$.

Step 2. Let $\sigma$ denote conjugation of $\mathfrak{g}$ with respect to ${\mathfrak{g}}_0$, and $\tau$ conjugation with respect to $\mathfrak{u}$.

Step 3. The map $\phi =\tau \sigma$ is a $\mathbb{C}$-linear automorphism of $\mathfrak{g}$ (product of two conjugate-linear maps). It is self-adjoint and positive definite with respect to the inner product $B_{\mathfrak{u}}(X,Y)=-\kappa (X,\tau (Y))$ (the positive-definite form from the compact real form).

Step 4. Define $\psi ={\phi }^{1/2}$ (the positive-definite square root, which exists and is an automorphism of $\mathfrak{g}$). Set ${\mathfrak{u}}^{\prime }={\psi }^{-1}(\mathfrak{u})$. Then ${\mathfrak{u}}^{\prime }$ is a compact real form of $\mathfrak{g}$, and $\sigma ({\mathfrak{u}}^{\prime })={\mathfrak{u}}^{\prime }$ (because $\psi$ commutes with $\sigma$).

Step 5. Since $\sigma$ preserves ${\mathfrak{u}}^{\prime }$, the restriction $\theta =\sigma {|}_{{\mathfrak{u}}^{\prime }}$ gives an involution of ${\mathfrak{u}}^{\prime }$. Extend to ${\mathfrak{g}}_0$: for $X\in {\mathfrak{g}}_0$, write $X=U+iV$ with $U,V\in {\mathfrak{u}}^{\prime }$. Then $\sigma (X)=\sigma (U)+i\sigma (V)$, and since $\sigma$ preserves ${\mathfrak{u}}^{\prime }$, we have $\sigma (U)\in {\mathfrak{u}}^{\prime }$ and $\sigma (V)\in {\mathfrak{u}}^{\prime }$. Define ${\theta }_{{\mathfrak{g}}_0}(X)=\sigma (U)-i\sigma (V)$... actually, $\theta =\sigma$ on ${\mathfrak{g}}_0$ already.

Let me reconsider. The Cartan involution of ${\mathfrak{g}}_0$ is simply the restriction of the conjugation ${\tau }^{\prime }$ of ${\mathfrak{u}}^{\prime }$ to ${\mathfrak{g}}_0$: $\theta ={\tau }^{\prime }{|}_{{\mathfrak{g}}_0}$. Since $\sigma$ preserves ${\mathfrak{u}}^{\prime }$ and $\sigma {\tau }^{\prime }={\tau }^{\prime }\sigma$, the involution $\theta$ is well-defined on ${\mathfrak{g}}_0$ and makes $B_{\theta }$ positive definite.

Step 6. The bilinear form $B_{\theta }(X,Y)=-\kappa (X,\theta (Y))$ on ${\mathfrak{g}}_0$ is positive definite because $\theta$ is the conjugation of the compact real form ${\mathfrak{u}}^{\prime }$, and $-\kappa$ is positive definite on ${\mathfrak{u}}^{\prime }$.

Proof of uniqueness (conjugacy). Let ${\theta }_1,{\theta }_2$ be two Cartan involutions of ${\mathfrak{g}}_0$. They define inner products $B_1,B_2$ on ${\mathfrak{g}}_0$ by $B_i(X,Y)=-\kappa (X,{\theta }_i(Y))$. Since both are inner products on a finite-dimensional space, there exists a $B_1$-self-adjoint, $B_1$-positive automorphism $A$ with $B_2(X,Y)=B_1(AX,Y)$.

Since $A$ is $B_1$-self-adjoint and positive, its logarithm $\log A$ exists. Define $a_t=\exp (t\log A)$ for $t\in [0,1]$. Then $a_0=\mathrm{id}$ and $a_1=A$.

The key claim: each $a_t$ commutes with ${\theta }_1$. This follows from the fact that $A$ intertwines the two Cartan involutions: ${\theta }_2=A{\theta }_1{A}^{-1}$, and ${\theta }_2^2=\mathrm{id}$ gives $A{\theta }_1={\theta }_1{A}^{-1}$. Combined with $A$ being $B_1$-self-adjoint (i.e., $A^T=A$ in the $B_1$ inner product, where $A^T$ denotes the $B_1$-transpose), one shows $A$ commutes with ${\theta }_1$.

Since $a_t$ commutes with ${\theta }_1$, each $a_t$ is an automorphism of $({\mathfrak{g}}_0,{\theta }_1)$, hence an automorphism of ${\mathfrak{g}}_0$ preserving the Lie bracket. At $t=1$: ${\theta }_2=A{\theta }_1{A}^{-1}$, so $\phi =A\in \mathrm{Int}({\mathfrak{g}}_0)$ conjugates ${\theta }_1$ to ${\theta }_2$. $\square$

Bridge. The existence proof builds toward the Iwasawa decomposition (Theorem 5 in Advanced results), where the Cartan involution provides the maximal compact subgroup $K$ of the decomposition $G=KAN$. The foundational reason is that the Cartan involution is the Lie-algebraic shadow of the polar decomposition in the Lie group, and this is exactly the bridge between the algebraic structure (real form) and the geometric structure (symmetric space $G/K$). Putting these together with the compact real form of 07.04.02, every real form corresponds to a Cartan involution, and the conjugacy of involutions identifies the uniqueness of the symmetric space.

Exercises [Intermediate+]

Advanced results [Master]

Theorem 1 (Conjugacy of Cartan involutions). Any two Cartan involutions of a real semisimple Lie algebra ${\mathfrak{g}}_0$ are conjugate by an inner automorphism. This is proved in the Key theorem above using the polar-decomposition argument.

Theorem 2 (Cartan decomposition at the group level). Let $G$ be a connected semisimple Lie group with Lie algebra ${\mathfrak{g}}_0=\mathfrak{k}\oplus \mathfrak{p}$, and let $K=⟨\exp (\mathfrak{k})⟩$. Then $G=K\cdot \exp (\mathfrak{p})$ (Cartan decomposition), the map $K\times \mathfrak{p}\to G$ given by $(k,X)\mapsto k\cdot \exp (X)$ is a diffeomorphism, and $K$ is a maximal compact subgroup.

Theorem 3 (Iwasawa decomposition). Every connected semisimple Lie group $G$ decomposes as $G=KAN$ where $K$ is maximal compact, $A$ is a vector group (connected abelian subgroup with Lie algebra a maximal abelian subalgebra of $\mathfrak{p}$), and $N$ is a nilpotent group. The multiplication map $K\times A\times N\to G$ is a diffeomorphism.

Theorem 4 (Langlands decomposition). A parabolic subgroup $P$ of $G$ has a Langlands decomposition $P=MAN$ where $M$ is semisimple (or reductive), $A$ is a split torus, and $N$ is the unipotent radical. The Iwasawa decomposition is the special case $P=B$ (a minimal parabolic, i.e., a Borel subgroup).

Theorem 5 (Symmetric spaces). The coset space $G/K$ is a Riemannian symmetric space of non-compact type. The Cartan involution provides the geodesic symmetry at the base point $eK$. Conversely, every symmetric space of non-compact type arises from a Cartan involution by this construction (Helgason's theorem).

Theorem 6 (Cartan's classification of real forms). The real forms of each complex simple Lie algebra are classified by Satake diagrams (or equivalently, by involutive automorphisms of the Dynkin diagram). The compact real form corresponds to the empty diagram (all roots compactified). Cartan's 1914 paper gives the complete list.

Theorem 7 (Mostow's theorem). Any two maximal compact subgroups of a connected semisimple Lie group $G$ are conjugate by an inner automorphism (Mostow 1955, Ann. Math. 62). This is the group-level version of the conjugacy of Cartan involutions.

Synthesis. The Cartan involution is the foundational reason that the structure of real semisimple Lie groups splits into a compact part $K$ and a non-compact part $\exp (\mathfrak{p})$: the Iwasawa decomposition $G=KAN$ refines this into the maximal compact subgroup, the split torus, and the nilpotent radical. The central insight is that the involution $\theta$ identifies the symmetric space $G/K$ with the geometric data of the non-compact part, and this is exactly the bridge between the Lie algebra structure (Cartan decomposition) and the global geometry (symmetric spaces). Putting these together with the compact real form of 07.04.02, the Cartan involution of the complexified algebra restricts to the compact real form, and the pattern generalises: the Langlands decomposition refines the Iwasawa decomposition for parabolic subgroups, and the bridge is the identification of real forms with involutions of the Dynkin diagram in the Cartan-Weyl classification 07.04.01.

Full proof set [Master]

Proposition 1 (Iwasawa decomposition). Let $G$ be a connected semisimple Lie group with Cartan involution $\theta$, Cartan decomposition ${\mathfrak{g}}_0=\mathfrak{k}\oplus \mathfrak{p}$, and $\mathfrak{a}$ a maximal abelian subalgebra of $\mathfrak{p}$. Then $G=KAN$ where $K=\exp (\mathfrak{k})$ (the analytic subgroup), $A=\exp (\mathfrak{a})$, and $N$ is constructed from the positive root spaces of $\mathfrak{a}$.

Proof. Let $\Phi ({\mathfrak{g}}_0,\mathfrak{a})$ be the set of restricted roots: the nonzero $\lambda \in {\mathfrak{a}}^{\ast }$ for which ${\mathfrak{g}}_{\lambda }=\{X\in {\mathfrak{g}}_0:[H,X]=\lambda (H)X\text{ for all }H\in \mathfrak{a}\}$ is nonzero. Choose a positive system ${\Phi }^{+}$ and set $\mathfrak{n}={⨁}_{\lambda \in {\Phi }^{+}}{\mathfrak{g}}_{\lambda }$. Define $N=\exp (\mathfrak{n})$.

Step 1. The subalgebra $\mathfrak{a}\oplus \mathfrak{n}$ is a solvable Lie algebra (since $[{\mathfrak{g}}_{\lambda },{\mathfrak{g}}_{\mu }]\subset {\mathfrak{g}}_{\lambda +\mu }$ and positive roots sum to positive roots or zero). The corresponding group $AN$ is a solvable group.

Step 2. The map $K\times A\times N\to G$ given by $(k,a,n)\mapsto kan$ is surjective. This is proved by showing that every $g\in G$ can be written as $g=kan$: the argument uses the polar decomposition $g=k\cdot \exp (X)$ with $k\in K$ and $X\in \mathfrak{p}$, then decomposes $X$ using the root space decomposition with respect to $\mathfrak{a}$.

Step 3. The map is injective. If $k_1{a}_1{n}_1=k_2{a}_2{n}_2$, then $k_2^{-1}{k}_1=a_2{n}_2{n}_1^{-1}{a}_1^{-1}$. The left side is in $K$ (compact) and the right side is in $AN$ (which has no compact subgroup other than the identity, since $A$ is a vector group and $N$ is nilpotent). So both sides equal the identity, giving $k_1=k_2$ and $a_1{n}_1=a_2{n}_2$. Since $A\cap N=\{e\}$ and $AN$ is a semidirect product, $a_1=a_2$ and $n_1=n_2$.

Step 4. The differential of the map is everywhere invertible (by the direct-sum decomposition ${\mathfrak{g}}_0=\mathfrak{k}\oplus \mathfrak{a}\oplus \mathfrak{n}$). So the map is a diffeomorphism. $\square$

Proposition 2 (Symmetric space structure). The coset space $G/K$ is a Riemannian symmetric space with the $G$-invariant metric induced by $B_{\theta }$ on $\mathfrak{p}$.

Proof. The tangent space at $eK$ is $\mathfrak{p}$ (via the identification $T_{eK}(G/K)\cong {\mathfrak{g}}_0/\mathfrak{k}\cong \mathfrak{p}$). The bilinear form $B_{\theta }{|}_{\mathfrak{p}}$ is positive definite (since $\kappa$ is positive definite on $\mathfrak{p}$ and $B_{\theta }(X,X)=\kappa (X,X)$ for $X\in \mathfrak{p}$).

The geodesic symmetry $s_{eK}$ at the base point is defined by $s_{eK}(gK)=\theta (g)K$, where $\theta$ is extended to $G$. Since ${\theta }^2=\mathrm{id}$, $s_{eK}^2=\mathrm{id}$, and $s_{eK}$ is an isometry (it preserves the metric because $\theta$ preserves $B_{\theta }$).

The $G$-action on $G/K$ is transitive and the isotropy group at $eK$ is $K$. The metric is $G$-invariant because $K$ preserves $B_{\theta }{|}_{\mathfrak{p}}$ (by the bracket relation $[\mathfrak{k},\mathfrak{p}]\subset \mathfrak{p}$). So $G/K$ is a Riemannian symmetric space. $\square$

Connections [Master]

• Compact real form 07.04.02. The Cartan involution of a real semisimple Lie algebra is constructed from the compact real form of its complexification. The involution $\theta$ on ${\mathfrak{g}}_0$ is the restriction of the conjugation of the compact real form, and the existence of the compact form guarantees the existence of the Cartan involution.

• Cartan-Weyl classification 07.04.01. The classification of real forms of complex semisimple Lie algebras (equivalently, of Cartan involutions) refines the Cartan-Weyl classification. Each complex type ($A_n$, $B_n$, $C_n$, $D_n$, $E_6$, $E_7$, $E_8$, $F_4$, $G_2$) has several real forms indexed by Satake diagrams, each corresponding to a different Cartan involution.

• Lie algebra representation 07.06.01. The Cartan involution decomposes representations of ${\mathfrak{g}}_0$ into $K$-types (representations of the maximal compact subalgebra $\mathfrak{k}$). The $K$-type decomposition is the foundation of the representation theory of real semisimple Lie groups, connecting to the Langlands programme.

• Root systems 07.06.03. The restricted root system $\Phi ({\mathfrak{g}}_0,\mathfrak{a})$ (roots of $\mathfrak{a}$ on ${\mathfrak{g}}_0$) governs the Iwasawa decomposition $G=KAN$. This restricted root system is derived from the absolute root system of the complexification but with multiplicities and non-reduced patterns reflecting the real form.

Historical & philosophical context [Master]

Elie Cartan introduced Cartan involutions in his 1914 classification of real simple Lie groups (Ann. Sci. Ecole Norm. Sup. 31) ^[Cartan1914], extending his earlier classification of complex simple Lie algebras to the real setting. The involution and the associated symmetric space $G/K$ were the central tools of this classification. Hermann Weyl used the compact real form (equivalently, the Cartan involution of the complex algebra) in his 1925--26 papers ^[Weyl1925] to prove complete reducibility via the unitary trick, but did not develop the real-form theory in full.

The Iwasawa decomposition was proved by Kenkichi Iwasawa in 1949 (Ann. Math. 50), providing the structural decomposition $G=KAN$ that underlies harmonic analysis on semisimple groups. The uniqueness-up-to-conjugacy of maximal compact subgroups was proved by George Mostow in 1955 (Ann. Math. 62) ^[Mostow1955]. Helgason's 1978 monograph Differential Geometry, Lie Groups, and Symmetric Spaces ^{[Helgason1978]} unified the Lie-algebraic and geometric viewpoints, establishing the correspondence between Cartan involutions and symmetric spaces as the central organising principle of the theory.

Bibliography [Master]

@article{Cartan1914,
  author = {Cartan, Elie},
  title = {Les groupes reels simples finis et continus},
  journal = {Ann. Sci. Ecole Norm. Sup.},
  volume = {31},
  year = {1914},
  pages = {263--355},
}

@article{Mostow1955,
  author = {Mostow, George D.},
  title = {A new proof of {E.~Cartan's} theorem on the topology of semisimple groups},
  journal = {Ann. Math.},
  volume = {62},
  year = {1955},
  pages = {532--546},
}

@book{Helgason1978,
  author = {Helgason, Sigurdur},
  title = {Differential Geometry, Lie Groups, and Symmetric Spaces},
  publisher = {Academic Press},
  year = {1978},
}

@book{Knapp2002,
  author = {Knapp, Anthony W.},
  title = {Lie Groups Beyond an Introduction},
  publisher = {Birkhauser},
  year = {2002},
  edition = {2nd},
}

@book{Humphreys1972,
  author = {Humphreys, James E.},
  title = {Introduction to Lie Algebras and Representation Theory},
  publisher = {Springer},
  year = {1972},
}

@article{Iwasawa1949,
  author = {Iwasawa, Kenkichi},
  title = {On some types of topological groups},
  journal = {Ann. Math.},
  volume = {50},
  year = {1949},
  pages = {507--558},
}