Orthogonal symmetric Lie algebra

Intuition [Beginner]

An orthogonal symmetric Lie algebra is a pair $(\mathfrak{g},s)$: a Lie algebra $\mathfrak{g}$ equipped with an involution $s$ (a self-map satisfying $s^2=\mathrm{id}$) that preserves the bracket. The involution splits $\mathfrak{g}$ into two halves --- the fixed points $\mathfrak{k}$ where $s$ acts as the identity, and the negated part $\mathfrak{p}$ where $s$ sends each element to its negative. The fixed-point set $\mathfrak{k}$ must be compact: it generates a bounded (compact) group of transformations.

This structure encodes the infinitesimal geometry of a symmetric space. On a symmetric space, reflecting through any point is a symmetry. At the Lie algebra level, this reflection becomes the involution $s$, and the decomposition $\mathfrak{g}=\mathfrak{k}\oplus \mathfrak{p}$ records how infinitesimal symmetries split into directions preserved by the reflection (the isotropy algebra $\mathfrak{k}$) and directions reversed by it (the tangent directions $\mathfrak{p}$).

The simplest concrete example is the Cartan involution on $\mathfrak{sl}(2,\mathbb{R})$ from 07.04.03. The map $\theta (X)=-X^T$ splits $\mathfrak{sl}(2,\mathbb{R})$ into $\mathfrak{so}(2)$ (skew-symmetric matrices, dimension 1) and the symmetric traceless matrices (dimension 2). The pair $(\mathfrak{sl}(2,\mathbb{R}),\theta )$ is an orthogonal symmetric Lie algebra.

Why does this concept exist? It provides the algebraic framework that classifies all symmetric spaces. Each symmetric space corresponds to an orthogonal symmetric Lie algebra, and the algebraic classification maps directly to the geometric one.

Visual [Beginner]

A diagram showing a Lie algebra $\mathfrak{g}$ drawn as an ellipse, split by a vertical line labelled $s$. The left half is $\mathfrak{k}$ (the $+1$ eigenspace, shaded dark, labelled "compact isotropy"). The right half is $\mathfrak{p}$ (the $-1$ eigenspace, shaded light, labelled "tangent directions"). An arrow labelled $s$ maps each element of $\mathfrak{k}$ to itself and each element of $\mathfrak{p}$ to its negative.

The bracket structure respects the split: taking the bracket of two elements of $\mathfrak{k}$ stays in $\mathfrak{k}$, mixing one from each goes to $\mathfrak{p}$, and bracketing two elements of $\mathfrak{p}$ lands back in $\mathfrak{k}$.

Worked example [Beginner]

Consider $\mathfrak{g}=\mathfrak{su}(2)\oplus \mathfrak{su}(2)$ with the swap involution $s(X,Y)=(Y,X)$.

Step 1. Check $s$ is an involution: $s(s(X,Y))=s(Y,X)=(X,Y)$, so $s^2=\mathrm{id}$.

Step 2. Check $s$ preserves the bracket: $s([(X_1,Y_1),(X_2,Y_2)])=s([X_1,X_2],[Y_1,Y_2])=([Y_1,Y_2],[X_1,X_2])$. This equals $[(Y_1,X_1),(Y_2,X_2)]=[s(X_1,Y_1),s(X_2,Y_2)]$.

Step 3. Find the eigenspaces. The $+1$ eigenspace satisfies $(X,Y)=(Y,X)$, so $X=Y$. This gives $\mathfrak{k}=\{(X,X):X\in \mathfrak{su}(2)\}$, dimension 3. The $-1$ eigenspace satisfies $(X,Y)=-(Y,X)$, so $X=-Y$. This gives $\mathfrak{p}=\{(X,-X):X\in \mathfrak{su}(2)\}$, dimension 3.

What this tells us: the pair $(\mathfrak{su}(2)\oplus \mathfrak{su}(2),\text{swap})$ is an orthogonal symmetric Lie algebra of Type II (compact type, group case). The fixed-point set $\mathfrak{k}\cong \mathfrak{su}(2)$ is compact. This pair encodes the symmetric space structure of the Lie group $SU(2)$.

Check your understanding [Beginner]

Formal definition [Intermediate+]

Let $\mathfrak{g}$ be a real finite-dimensional Lie algebra.

Definition (Orthogonal symmetric Lie algebra). An orthogonal symmetric Lie algebra is a pair $(\mathfrak{g},s)$ where:

1. $s:\mathfrak{g}\to \mathfrak{g}$ is an involutive automorphism: $s^2=\mathrm{id}$ and $s([X,Y])=[s(X),s(Y)]$ for all $X,Y\in \mathfrak{g}$.
2. The fixed-point subalgebra $\mathfrak{k}=\{X\in \mathfrak{g}:s(X)=X\}$ is compactly embedded: the adjoint group ${\mathrm{Ad}}_{\mathfrak{g}}(\mathfrak{k})$ (the subgroup of $\mathrm{Aut}(\mathfrak{g})$ generated by $\{\exp (\mathrm{ad}(X)):X\in \mathfrak{k}\}$) has compact closure.

Write $\mathfrak{g}=\mathfrak{k}\oplus \mathfrak{p}$ where $\mathfrak{p}=\{X\in \mathfrak{g}:s(X)=-X\}$. The bracket relations are $[\mathfrak{k},\mathfrak{k}]\subset \mathfrak{k}$, $[\mathfrak{k},\mathfrak{p}]\subset \mathfrak{p}$, $[\mathfrak{p},\mathfrak{p}]\subset \mathfrak{k}$.

Definition (Effectiveness). The pair $(\mathfrak{g},s)$ is effective if $\mathfrak{k}$ contains no nonzero ideal of $\mathfrak{g}$.

Definition (Irreducibility). The pair $(\mathfrak{g},s)$ is irreducible if the adjoint representation ${\mathrm{ad}}_{\mathfrak{g}}{|}_{\mathfrak{k}}:\mathfrak{k}\to \mathfrak{gl}(\mathfrak{p})$ is irreducible.

Counterexamples to common slips

• The identity map does not produce an effective pair. If $s=\mathrm{id}$, then $\mathfrak{p}=0$ and $\mathfrak{k}=\mathfrak{g}$. Every ideal of $\mathfrak{g}$ lies in $\mathfrak{k}$, so effectiveness fails.

• An arbitrary involution on a non-compact algebra may not satisfy the compact-embedding condition. For example, the involution $\theta (X)=-X^T$ on $\mathfrak{gl}(n,\mathbb{R})$ has fixed-point set $\mathfrak{so}(n)$, which is compact, but $\mathfrak{gl}(n,\mathbb{R})$ is not semisimple (it has a center), so the pair is not irreducible.

• The direct sum of two orthogonal symmetric Lie algebras is not irreducible. If $({\mathfrak{g}}_1,s_1)$ and $({\mathfrak{g}}_2,s_2)$ are both nontrivial, then $({\mathfrak{g}}_1\oplus {\mathfrak{g}}_2,s_1\oplus s_2)$ decomposes as $\mathfrak{p}={\mathfrak{p}}_1\oplus {\mathfrak{p}}_2$ with each summand invariant under $\mathrm{ad}(\mathfrak{k})$.

Key theorem with proof [Intermediate+]

Theorem (Classification of irreducible orthogonal symmetric Lie algebras). Let $(\mathfrak{g},s)$ be an irreducible, effective orthogonal symmetric Lie algebra. Then exactly one of the following holds:

1. (Type I) $\mathfrak{g}$ is compact simple.
2. (Type II) $\mathfrak{g}=\mathfrak{u}\oplus \mathfrak{u}$ for a compact simple Lie algebra $\mathfrak{u}$, and $s(X,Y)=(Y,X)$.
3. (Type III) $\mathfrak{g}$ is non-compact simple and $s$ is a Cartan involution 07.04.03.

Types I and III are dual via complexification (see Step 5 below).

Proof.

Step 1 (Semisimplicity). Effectiveness forces $\mathfrak{g}$ to be semisimple. Suppose the radical $\mathfrak{r}\ne 0$ (maximal solvable ideal). Since $s$ is an automorphism, $s(\mathfrak{r})=\mathfrak{r}$, so $\mathfrak{r}=(\mathfrak{r}\cap \mathfrak{k})\oplus (\mathfrak{r}\cap \mathfrak{p})$. The subspace $\mathfrak{r}\cap \mathfrak{k}$ is an ideal of $\mathfrak{g}$ contained in $\mathfrak{k}$ (it is stable under $[\mathfrak{g},\mathfrak{g}]$ since $\mathfrak{r}$ is an ideal and $[\mathfrak{r}\cap \mathfrak{k},\mathfrak{g}]\subset \mathfrak{r}\cap [\mathfrak{k},\mathfrak{g}]\subset \mathfrak{r}\cap \mathfrak{k}$). By effectiveness, $\mathfrak{r}\cap \mathfrak{k}=0$, so $\mathfrak{r}\subset \mathfrak{p}$.

Now $[\mathfrak{r},\mathfrak{r}]\subset \mathfrak{r}\cap [\mathfrak{p},\mathfrak{p}]\subset \mathfrak{r}\cap \mathfrak{k}=0$, so $\mathfrak{r}$ is abelian. Then $[\mathfrak{k},\mathfrak{r}]\subset \mathfrak{r}\cap \mathfrak{p}$ (an invariant subspace of $\mathfrak{p}$ under $\mathrm{ad}(\mathfrak{k})$). By irreducibility and $\mathfrak{r}\ne 0$, either $\mathfrak{r}\cap \mathfrak{p}=\mathfrak{p}$ or $\mathfrak{r}=0$. If $\mathfrak{r}=\mathfrak{p}$, then $[\mathfrak{p},\mathfrak{p}]=0$ and $\mathfrak{k}$ acts on the abelian space $\mathfrak{p}$, making $\mathfrak{k}+[\mathfrak{p},\mathfrak{p}]=\mathfrak{k}$ an ideal, contradicting effectiveness. So $\mathfrak{r}=0$ and $\mathfrak{g}$ is semisimple.

Step 2 (Simple ideal decomposition). Write $\mathfrak{g}={\mathfrak{g}}_1\oplus \cdots \oplus {\mathfrak{g}}_r$ as a direct sum of simple ideals. Since $s$ is an automorphism, each $s({\mathfrak{g}}_i)$ is a simple ideal, hence $s({\mathfrak{g}}_i)={\mathfrak{g}}_j$ for some $j$.

Step 3 (Irreducibility forces $r\le 2$). For each $i$, the subspace $({\mathfrak{g}}_i\cap \mathfrak{p})$ is invariant under $\mathrm{ad}(\mathfrak{k})$: if $K\in \mathfrak{k}$ and $P\in {\mathfrak{g}}_i\cap \mathfrak{p}$, then $[K,P]\in \mathfrak{p}$ (bracket relations) and $[K,P]\in {\mathfrak{g}}_i$ (${\mathfrak{g}}_i$ is an ideal), so $[K,P]\in {\mathfrak{g}}_i\cap \mathfrak{p}$.

By irreducibility of $\mathfrak{p}$ under $\mathrm{ad}(\mathfrak{k})$, there can be at most one index $i$ with ${\mathfrak{g}}_i\cap \mathfrak{p}\ne 0$ and $s({\mathfrak{g}}_i)={\mathfrak{g}}_i$. For pairs $(i,j)$ with $i\ne j$ and $s({\mathfrak{g}}_i)={\mathfrak{g}}_j$, the pair contributes the invariant subspace $\{(X,-s(X)):X\in {\mathfrak{g}}_i\}\cap \mathfrak{p}$ to $\mathfrak{p}$. Irreducibility allows at most one such pair. So either $r=1$ (one simple ideal, $s$-stable) or $r=2$ ($\mathfrak{g}=\mathfrak{u}\oplus \mathfrak{u}$ with $s$ swapping the two factors).

Step 4 (Type classification). Case $r=1$: $\mathfrak{g}$ is simple. If $\mathfrak{g}$ is compact, this is Type I. If $\mathfrak{g}$ is non-compact, then compact embedding of $\mathfrak{k}$ forces the Killing form to be negative definite on $\mathfrak{k}$, making $s$ a Cartan involution: Type III.

Case $r=2$: $\mathfrak{g}=\mathfrak{u}\oplus \mathfrak{u}$, $s(X,Y)=(Y,X)$, $\mathfrak{u}$ simple. Compact embedding of $\mathfrak{k}=\{(X,X)\}$ requires $\mathfrak{u}$ compact (otherwise the adjoint group of $\mathfrak{k}\cong \mathfrak{u}$ is not compact). This is Type II.

Step 5 (Duality I $\leftrightarrow$ III). Given a Type I pair $(\mathfrak{g},s)$ with $\mathfrak{g}$ compact simple, write $\mathfrak{g}=\mathfrak{k}\oplus \mathfrak{p}$. Complexify: ${\mathfrak{g}}_{\mathbb{C}}={\mathfrak{k}}_{\mathbb{C}}\oplus {\mathfrak{p}}_{\mathbb{C}}$. Define ${\mathfrak{g}}^{\ast }=\mathfrak{k}+i\mathfrak{p}$ as a real subalgebra of ${\mathfrak{g}}_{\mathbb{C}}$, with involution $s^{\ast }=s{|}_{{\mathfrak{g}}^{\ast }}$.

The algebra ${\mathfrak{g}}^{\ast }$ is simple (since $\mathfrak{g}$ is simple, ${\mathfrak{g}}^{\ast }$ is a real form of ${\mathfrak{g}}_{\mathbb{C}}$, hence simple as a real algebra). The Killing form ${\kappa }^{\ast }$ of ${\mathfrak{g}}^{\ast }$ satisfies: ${\kappa }^{\ast }$ is negative definite on $\mathfrak{k}$ (same as in $\mathfrak{g}$) and positive definite on $i\mathfrak{p}$ (since ${\kappa }^{\ast }(iX,iX)=-\kappa (X,X)$ for $X\in \mathfrak{p}$, and $\kappa$ is negative definite on all of compact $\mathfrak{g}$, so $-\kappa (X,X)>0$). The involution $s^{\ast }$ has $+1$ eigenspace $\mathfrak{k}$ (negative-definite Killing form) and $-1$ eigenspace $i\mathfrak{p}$ (positive-definite Killing form), so $s^{\ast }$ is a Cartan involution of ${\mathfrak{g}}^{\ast }$. The pair $({\mathfrak{g}}^{\ast },s^{\ast })$ is Type III. The reverse construction (Type III $\to$ Type I) replaces $i\mathfrak{p}$ by $-i(i\mathfrak{p})=\mathfrak{p}$, recovering a compact real form. $\square$

Bridge. The classification theorem builds toward the Riemannian symmetric space classification in 07.04.07, where each orthogonal symmetric Lie algebra exponentiates to a unique simply-connected symmetric space. The foundational reason is that the involution $s$ encodes the geodesic reflection at the base point, and this is exactly the algebraic data needed to reconstruct the full local and global geometry. The bridge is the correspondence that identifies each pair $(\mathfrak{g},s)$ with a symmetric space $G/K$. Putting these together with the Cartan involution of 07.04.03, the duality between Types I and III generalises to a geometric duality between compact and non-compact symmetric spaces.

Exercises [Intermediate+]

Advanced results [Master]

Theorem 1 (Curvature tensor of a symmetric space). Let $(\mathfrak{g},s)$ be an orthogonal symmetric Lie algebra with decomposition $\mathfrak{g}=\mathfrak{k}\oplus \mathfrak{p}$ and an $\mathrm{ad}(\mathfrak{g})$-invariant bilinear form $B$ that is positive definite on $\mathfrak{p}$. The curvature tensor of the associated symmetric space at the base point is $R(X,Y)Z=-[[X,Y],Z]$ for $X,Y,Z\in \mathfrak{p}$.

Theorem 2 (De Rham decomposition). Every orthogonal symmetric Lie algebra decomposes as an orthogonal direct sum $(\mathfrak{g},s)={⨁}_i({\mathfrak{g}}_i,s_i)$ of irreducible orthogonal symmetric Lie algebras and a flat (abelian) factor. At the geometric level, every simply-connected symmetric space decomposes as a Riemannian product of irreducible symmetric spaces and a Euclidean factor.

Theorem 3 (Exponentiation theorem). Let $(\mathfrak{g},s)$ be an effective orthogonal symmetric Lie algebra. Then there exists a unique (up to isomorphism) simply-connected symmetric space $M=G/K$ whose associated orthogonal symmetric Lie algebra is $(\mathfrak{g},s)$. Here $G$ is the simply-connected Lie group with Lie algebra $\mathfrak{g}$, and $K$ is the connected Lie subgroup with Lie algebra $\mathfrak{k}$.

Theorem 4 (Lie triple system). The subspace $\mathfrak{p}$ of an orthogonal symmetric Lie algebra, equipped with the ternary operation $[X,Y,Z]:=[[X,Y],Z]$, forms a Lie triple system: it satisfies $[X,Y,Z]+[Y,Z,X]+[Z,X,Y]=0$ and $[X,Y,[Z,W,V]]=[[X,Y,Z],W,V]+[Z,[X,Y,W],V]+[Z,W,[X,Y,V]]$.

Theorem 5 (Holonomy of symmetric spaces). The restricted holonomy group of a Riemannian symmetric space $M=G/K$ at the base point equals $\mathrm{Ad}(K^0)$ acting on $\mathfrak{p}\cong T_{eK}M$. The holonomy representation coincides with the isotropy representation ${\mathrm{ad}}_{\mathfrak{g}}{|}_{\mathfrak{k}}$.

Theorem 6 (Complete classification table). The irreducible orthogonal symmetric Lie algebras of Type III are in bijection with non-compact real simple Lie algebras (equivalently, with Satake diagrams). The irreducible types of Type I are in bijection with involutive automorphisms of compact simple Lie algebras (equivalently, with symmetric spaces of compact type). Type II gives one family for each compact simple Lie algebra.

Synthesis. The orthogonal symmetric Lie algebra is the foundational reason that the classification of symmetric spaces reduces to the classification of involutive automorphisms of simple Lie algebras. The central insight is that the pair $(\mathfrak{g},s)$ identifies the infinitesimal structure of a symmetric space with the eigenspace decomposition of an involution, and this is exactly the algebraic data needed to reconstruct the full geometry via the exponentiation theorem. Putting these together with the compact-non-compact duality, every irreducible symmetric space comes in dual pairs (Type I and Type III), and the bridge is the complexification map $\mathfrak{p}\mapsto i\mathfrak{p}$ that interchanges the two types.

The pattern generalises: the de Rham decomposition of a general symmetric space into irreducible factors mirrors the decomposition of the orthogonal symmetric Lie algebra into irreducible components, and appears again in 07.04.08 where restricted root systems refine the structure of each irreducible factor. The Lie triple system structure on $\mathfrak{p}$ is dual to the holonomy algebra on $\mathfrak{k}$, and this is exactly the bridge between the tangent-space geometry and the isotropy representation.

Full proof set [Master]

Proposition 1 (Curvature tensor formula). The curvature tensor of the symmetric space $G/K$ associated to $(\mathfrak{g},s)$ at the base point satisfies $R(X,Y)Z=-[[X,Y],Z]$ for $X,Y,Z\in \mathfrak{p}$.

Proof. The Levi-Civita connection at the base point is determined by the $G$-invariant metric $B$ on $\mathfrak{p}$. For vector fields extending $X,Y\in \mathfrak{p}$, the Koszul formula gives:

$$2B({\nabla }_{X}Y,Z)=X\cdot B(Y,Z)+Y\cdot B(Z,X)-Z\cdot B(X,Y)+B([X,Y],Z)-B([Y,Z],X)+B([Z,X],Y)$$

At the base point $eK$, the Lie brackets of left-invariant extensions of elements of $\mathfrak{p}$ lie in $\mathfrak{k}$ for $[X,Y]$ with $X,Y\in \mathfrak{p}$ (by the bracket relations). The directional derivative terms $X\cdot B(Y,Z)$ vanish because $B$ is $G$-invariant and hence constant along orbits. Using $[\mathfrak{p},\mathfrak{p}]\subset \mathfrak{k}$ and $B(\mathfrak{k},\mathfrak{p})=0$, we get ${\nabla }_{X}Y=\frac{1}{2}[X,Y{]}_{\mathfrak{p}}=0$ at the base point (since $[X,Y]\in \mathfrak{k}$ for $X,Y\in \mathfrak{p}$). So ${\nabla }_{X}Y{|}_{eK}=0$ for all $X,Y\in \mathfrak{p}$.

The curvature is $R(X,Y)Z={\nabla }_X{\nabla }_{Y}Z-{\nabla }_Y{\nabla }_{X}Z-{\nabla }_{[X,Y]}Z$. Since ${\nabla }_{X}Y{|}_{eK}=0$, the first two terms vanish at $eK$. For the third term, $[X,Y]\in \mathfrak{k}$, and ${\nabla }_{U}Z$ for $U\in \mathfrak{k}$, $Z\in \mathfrak{p}$ equals $\frac{1}{2}[U,Z]$ (by the Koszul formula with $[U,Z]\in \mathfrak{p}$). So:

$$R(X,Y)Z{|}_{eK}=-{\nabla }_{[X,Y]}Z{|}_{eK}=-\frac{1}{2}[[X,Y],Z]-\frac{1}{2}[[X,Y],Z]=-[[X,Y],Z]$$

using the fact that $[X,Y]\in \mathfrak{k}$ and $[\mathfrak{k},\mathfrak{p}]\subset \mathfrak{p}$. $\square$

Proposition 2 (Lie triple system axioms). The space $\mathfrak{p}$ with $[X,Y,Z]=[[X,Y],Z]$ satisfies the Jacobi identity for Lie triple systems.

Proof. The ternary Jacobi identity $[X,Y,Z]+[Y,Z,X]+[Z,X,Y]=0$ expands to $[[X,Y],Z]+[[Y,Z],X]+[[Z,X],Y]=0$, which is the ordinary Jacobi identity for the Lie bracket, hence automatic.

The derivation identity $[X,Y,[Z,W,V]]=[[X,Y,Z],W,V]+[Z,[X,Y,W],V]+[Z,W,[X,Y,V]]$ expands via the Jacobi identity for the binary bracket. The left side is $[[X,Y],[[Z,W],V]]$. By the Jacobi identity applied to the outer bracket: $[[X,Y],[[Z,W],V]]=[[[X,Y],[Z,W]],V]+[[Z,W],[[X,Y],V]]$. The first term equals $[[[X,Y],Z],W],V]+[[Z,[[X,Y],W]],V]$ by another application. Combining gives the desired identity. $\square$

Connections [Master]

• Cartan involution 07.04.03. The Type III orthogonal symmetric Lie algebras are precisely the pairs $(\mathfrak{g},\theta )$ where $\theta$ is a Cartan involution of a non-compact semisimple Lie algebra $\mathfrak{g}$. The classification of Cartan involutions from 07.04.03 directly feeds the classification of Type III symmetric spaces. The compact embedding condition in the definition of orthogonal symmetric Lie algebra specialises to the positive-definiteness of $B_{\theta }$ for Cartan involutions.

• Compact real form 07.04.02. The duality between Type I and Type III orthogonal symmetric Lie algebras relies on the compact real form of the complexification. Given a compact simple algebra $\mathfrak{g}$ (Type I), the dual non-compact algebra ${\mathfrak{g}}^{\ast }=\mathfrak{k}\oplus i\mathfrak{p}$ is constructed inside ${\mathfrak{g}}_{\mathbb{C}}$ by choosing a different real form, precisely mirroring the compact real form construction of 07.04.02.

• Riemannian symmetric space 07.04.07. Every orthogonal symmetric Lie algebra $(\mathfrak{g},s)$ exponentiates to a unique simply-connected Riemannian symmetric space $G/K$. The curvature tensor, holonomy, and de Rham decomposition at the geometric level all descend from the Lie algebra structure documented here. The unit 07.04.07 develops the geometric side of this correspondence.

Historical & philosophical context [Master]

Elie Cartan introduced orthogonal symmetric Lie algebras in his 1926 memoir on symmetric spaces (Bull. Soc. Math. France 54) ^[Cartan1926], extending his earlier 1914 classification of real simple Lie groups. Cartan's insight was that the involutive automorphism of the Lie algebra encodes the local symmetry of the corresponding Riemannian manifold, and the global classification of symmetric spaces reduces to the algebraic classification of these involutions on simple Lie algebras.

The modern formulation, distinguishing the three types and establishing the compact-non-compact duality systematically, was crystallised by Helgason in his 1962 lectures and 1978 monograph Differential Geometry, Lie Groups, and Symmetric Spaces ^{[Helgason1978]}. Kobayashi and Nomizu (Foundations of Differential Geometry Vol. II, 1969) ^{[KobayashiNomizu1969]} gave the curvature tensor formula and the Lie triple system structure their standard modern treatment.

Bibliography [Master]

@article{Cartan1926,
  author = {Cartan, Elie},
  title = {Sur une classe remarquable d'espaces de {R}iemann},
  journal = {Bull. Soc. Math. France},
  volume = {54},
  year = {1926},
  pages = {214--264},
}

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

@book{KobayashiNomizu1969,
  author = {Kobayashi, Shoshichi and Nomizu, Katsumi},
  title = {Foundations of Differential Geometry, Vol. II},
  publisher = {Wiley-Interscience},
  year = {1969},
}

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