Representations of Compact Lie Groups

Representations of compact Lie groups are continuous homomorphisms from a compact Lie group

G

into the group of invertible linear operators on a (finite-dimensional) vector space. The central result is that every such representation is completely reducible into a direct sum of irreducibles, each irreducible appears in the regular representation, and the irreducibles are classified by highest weights lying in a dominant Weyl chamber. Applies across harmonic analysis, particle physics, number theory, and differential geometry. Developed by Schur, Weyl, Cartan, and Peter (1890s–1930s).

A representation of a compact Lie group $G$ on a finite-dimensional complex vector space $V$ is a smooth homomorphism $ρ : G \to G L (V)$ . Compactness is the decisive hypothesis: it guarantees an invariant inner product (via averaging over the Haar measure), forcing every representation to be unitarizable and hence completely reducible. The theory culminates in the Peter–Weyl theorem (the matrix coefficients of irreducibles span $L^{2} (G)$ ) and the Theorem of the Highest Weight (irreducibles correspond bijectively to dominant integral weights). The surprising fact: although $G$ is a curved, often nonabelian manifold, its entire harmonic analysis is governed by a finite combinatorial gadget — the root system and its Weyl group — making representation theory essentially a problem in lattice combinatorics.

The Core Insight

Compactness = unitarity = decomposability. Averaging over a compact group with its finite-total-mass Haar measure converts any representation into a unitary one. Unitarity makes orthogonal complements of invariant subspaces invariant, so representations break into irreducible "atoms." These atoms are then organized by the weights of a maximal torus: every irreducible has a unique highest weight, and the dominant integral weights enumerate all irreducibles without repetition. The continuous, infinite object

G

is thereby reduced to discrete data on a lattice.

Formal Statement

Compact Lie Group and Representation

A compact Lie group $G$ is a group that is also a compact smooth manifold, with smooth multiplication and inversion. A (finite-dimensional, complex) representation is a smooth homomorphism

ρ : G ⟶ G L (V), V ≅ C^{n},

⟳A homomorphism preserves products,

ρ (g h) = ρ (g) ρ (h)

; smoothness is automatic from continuity by a theorem of von Neumann for Lie groups.

Two representations $(ρ, V)$ , $(σ, W)$ are equivalent if there is a linear isomorphism $T : V \to W$ with $T ρ (g) = σ (g) T$ for all $g$ . A representation is irreducible if $V$ has no $G$ -invariant subspace other than $0$ and $V$ . The symbols: $V$ is the representation space, $n = dim V$ the dimension, and $G L (V)$ the invertible linear maps.

Haar Measure

On a compact Lie group there exists a unique bi-invariant Haar measure $μ$ normalized so that $μ (G) = 1$ :

\int_{G} f (h g) d μ (g) = \int_{G} f (g h) d μ (g) = \int_{G} f (g) d μ (g),

⟳Build a left-invariant top-degree form from a basis of the Lie algebra, transport it by left translation; compactness gives finite mass, and bi-invariance follows since the modular function of a compact group is trivial.

Unitarizability and Complete Reducibility

Every finite-dimensional representation $(ρ, V)$ of a compact Lie group $G$ admits a $G$ -invariant Hermitian inner product, making $ρ$ unitary. Consequently $V$ decomposes as an orthogonal direct sum of irreducible subrepresentations:

V ≅ i ⨁ V_{i}, V_{i} irreducible .

⟳Average any inner product over the group using Haar measure; orthogonal complement of an invariant subspace is invariant; induct on dimension.

Schur's Lemma

If $(ρ, V)$ and $(σ, W)$ are irreducible and $T : V \to W$ is an intertwiner ( $T ρ (g) = σ (g) T$ ), then either $T = 0$ or $T$ is an isomorphism. Moreover, any self-intertwiner of an irreducible complex representation is a scalar:

T ρ (g) = ρ (g) T \forall g ⟹ T = λ Id, λ \in C .

⟳

ker T

and

im T

are invariant; irreducibility forces them to be trivial. For the scalar claim, an eigenvalue

λ

T

gives invariant kernel

T - λ

, which must be all of

V

Peter–Weyl

Let $G$ be the set of equivalence classes of irreducible unitary representations of $G$ . The matrix coefficients $g \mapsto ⟨ ρ (g) v, w ⟩$ for $ρ \in G$ span a dense subspace of $L^{2} (G)$ , and

L^{2} (G) ≅ ρ \in G ⨁ V_{ρ} \otimes V_{ρ}^{*}, dim (ρ -isotypic block) = (dim V_{ρ})^{2} .

⟳The convolution operator by a continuous function is compact and self-adjoint (when the function is suitable); the spectral theorem decomposes

L^{2} (G)

into finite-dimensional eigenspaces, each carrying irreducible bi-representations of

G \times G

Theorem of the Highest Weight

Let $T \subset G$ be a maximal torus with weight lattice $Λ \subset t^{*}$ and choose a set of positive roots determining a dominant chamber. Then equivalence classes of irreducible representations of a compact connected $G$ are in bijection with dominant integral weights $λ \in Λ^{+}$ :

G 1 : 1 Λ^{+} = {λ \in Λ : ⟨ λ, α^{\lor} ⟩ \geq 0 for all simple α} .

⟳Restrict an irreducible to

T

; its weights are a Weyl-invariant set with a unique extreme (highest) weight

λ

; show

λ

is dominant integral and uniquely recovers the representation via Verma-module / Borel–Weil construction.

Why It Holds / Derivation

Complete Reducibility via Averaging

Step 1 — Build an invariant inner product. Start with any Hermitian inner product

⟨ \cdot, \cdot ⟩_{0}

V

. Define

⟨ v, w ⟩ := \int_{G} ⟨ ρ (g) v, ρ (g) w ⟩_{0} d μ (g) .

⟳The integrand is continuous in

g

and

G

is compact, so the integral converges; positivity is preserved because the integrand is positive for

v \neq = 0

Step 2 — Verify invariance. For fixed

h \in G

, substitute

g \mapsto g h

and use right-invariance of

μ

⟨ ρ (h) v, ρ (h) w ⟩ = \int_{G} ⟨ ρ (g h) v, ρ (g h) w ⟩_{0} d μ (g) = ⟨ v, w ⟩ .

⟳

ρ (g) ρ (h) = ρ (g h)

; right-invariance of Haar measure absorbs the shift.

Step 3 — Orthogonal complement is invariant. If

W \subseteq V

is invariant, then for

w^{⊥} \in W^{⊥}

v \in W

, and any

g

⟨ ρ (g) w^{⊥}, v ⟩ = ⟨ w^{⊥}, ρ (g^{- 1}) v ⟩ = 0

since

ρ (g^{- 1}) v \in W

. Hence

W^{⊥}

is invariant. Step 4 — Induct. Decompose

V = W \oplus W^{⊥}

and recurse on each factor until only irreducibles remain. Finite dimension guarantees termination.

■

Orthogonality of Characters

Define the character $χ_{ρ} (g) = tr ρ (g)$ . Using Schur's lemma and the averaging operator $P = \int_{G} ρ (g) \otimes σ (g)^{*} d μ (g)$ projecting onto invariants of $Hom (W, V)$ , one obtains

\int_{G} χ_{ρ} (g) \overline{χ_{σ} (g)} d μ (g) = dim Hom_{G} (σ, ρ) = {1, 0, ρ ≅ σ, ρ \neq ≅ σ .

⟳

dim Hom_{G} (W, V) = tr P = \int_{G} χ_{ρ} \overline{χ_{σ}}

; Schur's lemma makes this

1

for equivalent irreducibles,

0

otherwise.

Thus the irreducible characters form an orthonormal set in $L^{2} (G)$ , and by Peter–Weyl they are an orthonormal basis of the space of class functions. $■$

Hermann Weyl | 1885–1955

Weyl unified Cartan's structural Lie algebra theory with Hurwitz's averaging trick into the modern representation theory of compact groups. In a celebrated 1925–1926 series of papers he proved complete reducibility globally (the "unitarian trick"), derived the Weyl character formula and dimension formula, and (with Fritz Peter) established the completeness theorem for $L^{2} (G)$ . Weyl insisted that the global group, not merely the Lie algebra, carried the essential information — a viewpoint that anticipated harmonic analysis on homogeneous spaces and the entire Langlands program. His book The Classical Groups (1939) became the canonical text.

Building Intuition

Think of a compact Lie group as a "smoothed-out finite group." For a finite group, you average by summing over elements and dividing by the order; for a compact Lie group you integrate against Haar measure of total mass $1$ . Everything finite-group representation theory gives you survives. The maximal torus $T$ is the group's "diagonalizable core" — like simultaneously diagonalizing commuting unitary matrices, restricting to $T$ splits $V$ into one-dimensional weight spaces labeled by characters of $T$ . The Weyl group $W = N (T) / T$ permutes these weights, imposing a symmetry. The highest weight is simply the most extreme weight in the dominant chamber — a single point that "generates" all the others by lowering operators, exactly as the highest-spin state of an $S U (2)$ multiplet generates the rest by applying $J_{-}$ .

The SU(2) Paradigm

In quantum mechanics, angular momentum is representation theory of $S U (2)$ (the double cover of $S O (3)$ ). The irreducible representations $V_{j}$ are labeled by spin $j \in {0, \frac{1}{2}, 1, \frac{3}{2}, \dots}$ — precisely the dominant weights of $S U (2)$ , which is a single nonnegative half-integer. The dimension $2 j + 1$ counts the magnetic quantum numbers $m = - j, \dots, + j$ , which are the weights. The ladder operators $J_{\pm}$ are the root vectors, raising and lowering $m$ by $1$ . Clebsch–Gordan coefficients are nothing but the decomposition of tensor products $V_{j_{1}} \otimes V_{j_{2}} = ⨁_{j = ∣ j_{1} - j_{2} ∣}^{j_{1} + j_{2}} V_{j}$ into irreducibles. Every word a physicist says about angular momentum is a sentence in compact group representation theory.

Root system $A_2$ of $SU(3)$ with dominant Weyl chamber and fundamental weights

Worked Examples

Irreducibles of SU(2) and a Tensor Decomposition

Given:

G = S U (2)

, maximal torus

T = {diag (e^{i θ}, e^{- i θ})}

. Weights are integers

m

via

diag (e^{i θ}, e^{- i θ}) \mapsto e^{im θ}

. We want to decompose

V_{1} \otimes V_{1}

, where

V_{1}

is the spin-1 (adjoint, 3-dimensional) representation. Step 1 — Identify weights of $V_{1}$ . Spin

j = 1

has weights

{- 2, 0, 2}

in the integer normalization (equivalently

m = - 1, 0, 1

). Character:

χ_{1} (θ) = e^{2 i θ} + 1 + e^{- 2 i θ} .

⟳Character of

V_{j}

is the sum of

e^{im θ}

over the weights; here three weights.

Step 2 — Multiply characters.

χ_{1} (θ)^{2} = (e^{2 i θ} + 1 + e^{- 2 i θ})^{2} = e^{4 i θ} + 2 e^{2 i θ} + 3 + 2 e^{- 2 i θ} + e^{- 4 i θ} .

⟳Expand the square term by term and collect powers of

e^{i θ}

Step 3 — Subtract highest characters greedily. The top weight

e^{4 i θ}

forces a spin-

2

factor (

χ_{2} = e^{4 i θ} + e^{2 i θ} + 1 + e^{- 2 i θ} + e^{- 4 i θ}

). The remainder is

χ_{1} + χ_{0}

χ_{1}^{2} = χ_{2} + χ_{1} + χ_{0} .

⟳Peel off the irreducible character with the highest remaining weight, repeat until zero.

Boxed answer:

V_{1} \otimes V_{1} ≅ V_{2} \oplus V_{1} \oplus V_{0}

(dimensions

9 = 5 + 3 + 1

, the spin-1 ⊗ spin-1 = spin-2 ⊕ spin-1 ⊕ spin-0 of physics).

Weyl Dimension Formula for SU(3)

Given:

G = S U (3)

, type

A_{2}

. Simple roots

α_{1}, α_{2}

; fundamental weights

ω_{1}, ω_{2}

. Compute the dimension of the irreducible with highest weight

λ = ω_{1} + ω_{2}

(the adjoint representation). Step 1 — Weyl dimension formula.

dim V_{λ} = α > 0 \prod \frac{⟨ λ + δ , α ^{\lor} ⟩}{⟨ δ , α ^{\lor} ⟩}, δ = \frac{1}{2} α > 0 \sum α = ω_{1} + ω_{2} .

⟳Take the ratio of the Weyl character formula's numerator and denominator and evaluate the limit at the identity using L'Hôpital / Weyl's specialization argument.

Step 2 — List positive roots of $A_{2}$ .

α_{1}, α_{2}, α_{1} + α_{2}

. Pair

λ + δ = 2 ω_{1} + 2 ω_{2}

against each coroot. Using

⟨ ω_{i}, α_{j}^{\lor} ⟩ = δ_{ij}

⟨ λ + δ, α_{1}^{\lor} ⟩ = 2, ⟨ λ + δ, α_{2}^{\lor} ⟩ = 2, ⟨ λ + δ, (α_{1} + α_{2})^{\lor} ⟩ = 2 + 2 = 4.

⟳

(α_{1} + α_{2})^{\lor}

pairs with

ω_{i}

as the sum of the individual pairings in simply-laced type.

Step 3 — Denominator with $δ = ω_{1} + ω_{2}$ .

⟨ δ, α_{1}^{\lor} ⟩ = 1, ⟨ δ, α_{2}^{\lor} ⟩ = 1, ⟨ δ, (α_{1} + α_{2})^{\lor} ⟩ = 2

. Step 4 — Form the product.

dim V_{ω_{1} + ω_{2}} = \frac{2}{1} \cdot \frac{2}{1} \cdot \frac{4}{2} = 8.

⟳Multiply the three ratios from Steps 2–3.

Boxed answer:

dim V_{ω_{1} + ω_{2}} = 8

— the adjoint representation of

S U (3)

, the "eightfold way" octet of Gell-Mann.

Common Misconceptions

Compactness is essential, not cosmetic. For non-compact groups (e.g.,

S L (2, R)

), finite-dimensional representations need not be unitary or completely reducible, and the interesting unitary representations are infinite-dimensional. The averaging argument breaks because Haar measure has infinite total mass. Do not transplant complete reducibility to non-compact settings. "Irreducible over $R$ " $\neq =$ "irreducible over $C$ ." A real representation can be irreducible over

R

yet split when complexified (e.g., the 2-dimensional rotation representation of

S O (2)

R

-irreducible but

C

-reducible into

\pm 1

characters). The highest-weight theory is a complex theory. Weights are not eigenvalues of group elements directly — they are characters of the maximal torus, i.e., homomorphisms

T \to U (1)

, encoded as elements of the weight lattice

Λ \subset t^{*}

Consider the additive group $R$ (a non-compact abelian Lie group) acting on $C^{2}$ by

ρ (t) = (10 t 1) .

⟳Check

ρ (s + t) = ρ (s) ρ (t)

by matrix multiplication; the off-diagonal entries add.

This representation is not completely reducible: the subspace spanned by $e_{1}$ is invariant, but it has no invariant complement (any complement would have to be a $ρ$ -stable line, and the only $ρ$ -stable line is $C e_{1}$ ). The Jordan block cannot be diagonalized. This is impossible for a compact group precisely because there is no invariant inner product — illustrating why the compactness hypothesis in [3] cannot be dropped.

Connections

Representation theory of compact Lie groups sits at a crossroads. Via the unitarian trick it is equivalent to the Representation Theory of Complex Semisimple Lie Algebras: each compact $G$ has a complexified Lie algebra $g_{C}$ , and finite-dimensional reps of $G$ ↔ finite-dimensional reps of $g_{C}$ . The classification by highest weights mirrors Verma Modules and the Borel–Weil–Bott geometric realization on flag varieties. Harmonic analysis generalizes Fourier Analysis: Peter–Weyl is the nonabelian Fourier transform, with characters playing the role of $e^{2 π in x}$ . The combinatorial backbone — Root Systems and the Weyl Group — links to Coxeter Groups and Lie Algebra Cohomology. Tensor product decompositions connect to Schur–Weyl Duality and the Symmetric Group via Young tableaux.

Real-World Applications

Quantum angular momentum & spin — $S U (2)$ and $S O (3)$ representations classify atomic orbitals, nuclear spin states, and selection rules in spectroscopy.
The Standard Model of particle physics — the gauge group $S U (3) \times S U (2) \times U (1)$ ; quarks live in the fundamental $3$ of $S U (3)$ color, and Gell-Mann's "eightfold way" is the octet $8$ .
Crystallography and molecular vibration — finite and compact symmetry groups predict normal modes, IR/Raman activity, and degeneracies via character tables.
Quantum information — the Schur–Weyl Duality underlies optimal quantum state estimation, decoherence-free subspaces, and the Schur transform algorithm.
Signal processing on the sphere — spherical harmonics are $S O (3)$ irreducibles, used in geodesy, cosmic microwave background analysis, and computer graphics (environment lighting).
Lattice gauge theory and QCD simulations — Haar-measure integration over $S U (3)$ link variables is the computational heart of Monte Carlo lattice QCD.
Robotics and computer vision — $S O (3)$ and $S E (3)$ representations govern rotation interpolation, pose estimation, and equivariant neural networks.

Historical Development

1888–1894 — Wilhelm Killing and Élie Cartan classify simple complex Lie algebras via root systems, producing the $A, B, C, D, E, F, G$ types.
1897 — Adolf Hurwitz introduces invariant integration over $S O (n)$ and $U (n)$ — the seed of the averaging/Haar trick.
1905 — Issai Schur develops the orthogonality relations and "Schur's lemma" in the context of $G L_{n}$ .
1913 — Cartan proves the theorem of the highest weight for complex semisimple Lie algebras, classifying finite-dimensional irreducibles.
1925–1926 — Hermann Weyl proves global complete reducibility (unitarian trick), the character formula, and the dimension formula for compact semisimple groups.
1927 — Fritz Peter and Hermann Weyl prove the completeness theorem for $L^{2} (G)$ — the Peter–Weyl theorem.
1933 — Alfred Haar constructs invariant measure on general locally compact groups; John von Neumann proves uniqueness for compact groups.
1950s — Harish-Chandra and others extend the theory to non-compact semisimple groups via the unitarian trick and infinite-dimensional representations.
1957 — Raoul Bott completes the Borel–Weil–Bott theorem, realizing irreducibles as cohomology of line bundles on flag manifolds.

Subtleties & Limits

While the finite-dimensional representation theory of compact Lie groups is completely solved, several deep questions extend beyond it. Tensor product multiplicities (Littlewood–Richardson and their generalizations to all types, computed by Littelmann paths and crystal bases) remain combinatorially intricate; the saturation conjecture for general types and the geometry of the eigenvalue/Horn problem connect to algebraic geometry and remain active. The transition to non-compact and infinite-dimensional representations (the unitary dual of a general reductive group) is far from complete — the classification of the unitary dual is open for many real reductive groups, central to the Langlands program. For infinite-dimensional (compact-type) groups such as loop groups and their affine Lie algebra representations, positive-energy representation theory is rich but the full picture (e.g., for higher-dimensional current groups) is unresolved.