07.08.01 · representation-theory / finite-groups-depth

Representations of finite groups — Maschke, characters, and orthogonality

shipped3 tiersLean: none

Anchor (Master): Serre Linear Representations of Finite Groups; Feit Characters of Finite Groups (Academic Press, 1967); Isaacs Character Theory of Finite Groups (Academic Press, 1976)

Intuition Beginner

A group is, at heart, a multiplication table — a list of abstract elements and the rule for multiplying them. A representation turns every group element into an invertible matrix so that matrix multiplication mirrors the group's table exactly. Once the group lives inside a space of matrices, all of linear algebra — eigenvalues, traces, determinants, subspaces — becomes a toolkit for studying the symmetry.

The deepest fact about finite groups over the familiar number systems ( $Q$ , $R$ , $C$ ) is Maschke's theorem: this translation loses nothing. Every representation breaks cleanly into atomic, unbreakable pieces called irreducibles, the way every integer factors into primes. There is no entanglement that resists the splitting; the averaging trick of summing over the whole group washes out the obstruction.

A character compresses each matrix down to a single number, its trace. The miracle of the theory is that this one number per group element tells the whole story: two representations are the same exactly when their characters agree. The orthogonality relations turn the list of characters into a kind of periodic table for the group, and the number of irreducible atoms equals the number of conjugacy classes.

This unit assembles the entire finite-group picture in one place — the group algebra, Maschke's theorem with proof, characters and orthogonality, induced representations and Frobenius reciprocity, and the representation ring — anchored throughout by the smallest non-commutative example, the symmetric group $S_{3}$ .

Visual Beginner

A finite-group representation, after Maschke, is a block-diagonal array of irreducible blocks. Each block is invariant under every group element, and the action inside a block cannot be split further.

The concrete picture for $S_{3}$ : it is the symmetry group of an equilateral triangle, with six elements — three reflections (swapping two vertices), two rotations (by 120° and 240°), and the identity. Tracking how each of those six symmetries moves the triangle turns the abstract group into six explicit $2 \times 2$ matrices acting on the plane of the triangle. Those matrices are the standard 2-dimensional representation, the centrepiece of the worked example below.

Worked example Beginner

The symmetric group $S_{3}$ has six elements and three conjugacy classes: the identity, the three transpositions like $(12)$ , and the two 3-cycles like $(123)$ . A cornerstone theorem of this unit says the number of irreducible representations equals the number of conjugacy classes, so $S_{3}$ has exactly three irreducibles.

They are: the constant-one representation (dimension 1, every group element acts as the number $1$ ), the sign representation (dimension 1, each element acts as $+ 1$ if it is an even permutation and $- 1$ if it is odd), and the standard representation (dimension 2, coming from the action on the triangle). Their characters — the trace of each acting matrix — form the character table:

	$e$ (1 element)	$(12)$ (3 elements)	$(123)$ (2 elements)
constant-one $χ_{1}$	$1$	$1$	$1$
sign $χ_{2}$	$1$	$- 1$	$1$
standard $χ_{3}$	$2$	$0$	$- 1$

The dimension formula is the first sanity check: the squares of the irreducible dimensions add up to the size of the group. Concretely, $1^{2} + 1^{2} + 2^{2} = 1 + 1 + 4 = 6$ , and $∣ S_{3} ∣ = 6$ . They match — a fingerprint that we have found every irreducible.

The standard representation is genuinely 2-dimensional: it cannot be broken into two 1-dimensional pieces, because $S_{3}$ is non-commutative and its action on the triangle rotates as well as reflects. This is the smallest example where a group forces a genuinely multi-dimensional atom.

Check your understanding Beginner

Exercise (easy, multiple choice).

Maschke's theorem guarantees that over $C$ every representation of a finite group:

A. Is 1-dimensional B. Splits as a direct sum of irreducible representations C. Has the same character as the constant-one representation D. Is isomorphic to the regular representation

Hint

The theorem is about complete reducibility — every invariant subspace has an invariant complement.

Answer

B. Feedback-correct: yes — Maschke says every finite-dimensional complex representation of a finite group is a direct sum of irreducibles. Feedback-wrong: A is false ( $S_{3}$ has a 2-dimensional irreducible); C confuses a general representation with the constant-one representation; D confuses "decomposes" with "is the regular representation" (the regular representation is one specific large representation, not every one).

Formal definition Intermediate+

Let $G$ be a finite group and $k$ a field. The group algebra is the $k$ -vector space

k G = g \in G ⨁ k \cdot e_{g}

with multiplication $e_{g} \cdot e_{h} = e_{g h}$ , extended linearly. A representation of $G$ over $k$ on a finite-dimensional $k$ -vector space $V$ is a group homomorphism $ρ : G \to GL (V)$ ; equivalently, it is a left $k G$ -module structure on $V$ . This dictionary between representations and modules over the group algebra is the backbone of the theory: operations on modules (submodules, quotients, direct sums, tensor products) become operations on representations.

A representation is irreducible if $V \neq = 0$ and the only $G$ -invariant subspaces are $0$ and $V$ . It is completely reducible (or semisimple) if every invariant subspace $W \subseteq V$ admits an invariant complement $W^{'}$ with $V = W \oplus W^{'}$ , equivalently if $V$ is a direct sum of irreducibles. Maschke's theorem (next section) states that over a field of characteristic zero, every finite-dimensional representation of a finite group is completely reducible, so the whole theory reduces to classifying irreducibles and computing multiplicities.

Characters. The character of a representation $(V, ρ)$ is the function $χ_{V} : G \to k$ given by the trace, $χ_{V} (g) := tr (ρ (g))$ . Characters are class functions: $χ_{V} (h g h^{- 1}) = χ_{V} (g)$ , because trace is conjugation-invariant. The character of a direct sum is the sum of characters, $χ_{V \oplus W} = χ_{V} + χ_{W}$ , and the character of a tensor product is the pointwise product, $χ_{V \otimes W} (g) = χ_{V} (g) χ_{W} (g)$ .

Over $k = C$ , the space of class functions carries the standard inner product

⟨ α, β ⟩_{G} := \frac{1}{∣ G ∣} g \in G \sum α (g) \overline{β (g)} .

The orthogonality relations state that the irreducible characters form an orthonormal set: $⟨ χ_{V}, χ_{W} ⟩_{G} = δ_{V W}$ for irreducibles $V, W$ . A corollary, used constantly in computation, is the multiplicity formula: if $V ≅ ⨁_{i} V_{i}^{\oplus n_{i}}$ with $V_{i}$ irreducible, then $n_{i} = ⟨ χ_{V}, χ_{V_{i}} ⟩_{G}$ . Characters detect everything.

The regular representation. The group algebra $k G$ acting on itself by left multiplication is the (left) regular representation. Its character is concentrated at the identity: $χ_{reg} (e) = ∣ G ∣$ and $χ_{reg} (g) = 0$ for $g \neq = e$ (left multiplication by $g \neq = e$ has no fixed basis vector). Decomposing the regular representation gives the fundamental identity

C G ≅ i ⨁ V_{i}^{\oplus d i m V_{i}}, so ∣ G ∣ = i \sum (dim V_{i})^{2},

the dimension formula that we verified numerically for $S_{3}$ above.

Induced representations. For a subgroup $H \leq G$ and an $H$ -representation $W$ , the induced representation is

Ind_{H}^{G} W := {f : G \to W ∣ f (h g) = h \cdot f (g) for all h \in H},

with $G$ acting by right translation $(g_{0} \cdot f) (g) := f (g g_{0})$ . It is the representation that best extends $W$ from $H$ up to $G$ . Its character, on an element $g \in G$ , is

χ_{Ind_{H}^{G} W} (g) = \frac{1}{∣ H ∣} x \in G xg x^{- 1} \in H \sum χ_{W} (xg x^{- 1}) .

Frobenius reciprocity (proved in the full proof set) is the adjunction

Hom_{G} (Ind_{H}^{G} W, V) ≅ Hom_{H} (W, Res_{H}^{G} V),

or at the level of characters, $⟨ χ_{Ind_{H}^{G} W}, χ_{V} ⟩_{G} = ⟨ χ_{W}, χ_{Res_{H}^{G} V} ⟩_{H}$ . Induction and restriction are adjoint functors between the representation categories.

The representation ring. The Grothendieck group of the category of finite-dimensional $C G$ -representations, with addition from direct sum and multiplication from tensor product, is the representation ring (or character ring) $R (G)$ . By Maschke and the multiplicity formula, $R (G)$ is the free abelian group on the set of irreducible characters ${χ_{1}, \dots, χ_{r}}$ , with multiplication governed by the Clebsch–Gordan-style structure constants $c_{ij}^{k}$ defined by $χ_{i} \cdot χ_{j} = \sum_{k} c_{ij}^{k} χ_{k}$ . Computing these constants is the combinatorial heart of the subject; for $G = S_{n}$ they are the Kronecker coefficients, the deepest open problem in symmetric-function combinatorics.

Key theorem with proof Intermediate+

Theorem (Maschke, 1899). Let $G$ be a finite group and $k$ a field whose characteristic does not divide $∣ G ∣$ (in particular, any field of characteristic $0$ ). Then every finite-dimensional representation of $G$ over $k$ is completely reducible ^{[Maschke 1899]}.

Proof (averaging-projection argument). Let $(V, ρ)$ be a finite-dimensional $k G$ -representation and let $W \subseteq V$ be a $G$ -invariant subspace. We construct a $G$ -invariant complement $W^{'}$ .

Step 1 — pick any projection. Extend a basis of $W$ to a basis of $V$ and let $π : V \to W$ be the resulting linear projection, so $π (V) \subseteq W$ and $π ∣_{W} = id_{W}$ . This $π$ is generally not $G$ -equivariant.

Step 2 — average over $G$ . Define the averaged operator

π := \frac{1}{∣ G ∣} g \in G \sum ρ (g) \circ π \circ ρ (g)^{- 1} : V ⟶ V .

The factor $\frac{1}{∣ G ∣}$ is well-defined because $∣ G ∣$ is invertible in $k$ by hypothesis; this is the only place that hypothesis is used.

Step 3 — verify three properties.

(a) Image lands in $W$ . Each summand $ρ (g) π ρ (g)^{- 1}$ sends $V$ into $W$ , because $ρ (g)^{- 1}$ lands in $V$ , $π$ lands in $W$ , and $ρ (g)$ preserves the invariant subspace $W$ . So $π (V) \subseteq W$ .

(b) $π$ is a projection onto $W$ . For $w \in W$ , $G$ -invariance gives $ρ (g)^{- 1} w \in W$ , so $π (ρ (g)^{- 1} w) = ρ (g)^{- 1} w$ , and $ρ (g) π ρ (g)^{- 1} w = ρ (g) ρ (g)^{- 1} w = w$ . Each summand equals $w$ , hence $π (w) = \frac{1}{∣ G ∣} \cdot ∣ G ∣ w = w$ .

ρ (h) π = \frac{1}{∣ G ∣} g \sum ρ (h g) π ρ (g)^{- 1} = \frac{1}{∣ G ∣} g^{'} \sum ρ (g^{'}) π ρ (g^{'})^{- 1} ρ (h) = π ρ (h) .

Step 4 — extract the complement. Set $W^{'} := ker (π)$ . Because $π$ is $G$ -equivariant, $W^{'}$ is $G$ -invariant. The identity $V = im (π) \oplus ker (π) = W \oplus W^{'}$ gives the splitting. Iterating on $W$ and $W^{'}$ (which have strictly smaller dimension) terminates in a decomposition of $V$ into irreducible subrepresentations. $□$

The hypothesis that $char (k) ∤ ∣ G ∣$ is sharp: over $F_{p}$ with $p ∣ ∣ G ∣$ , the averaging sum $\frac{1}{∣ G ∣} \sum_{g}$ is not defined, and indeed Maschke fails. Its failure is the gateway to modular (or Brauer) representation theory, taken up in the companion units on Brauer characters and block theory 07.02.04.

Bridge. Maschke's averaging argument builds toward the entire semisimple world: the very same Reynolds-operator idea generalises to compact Lie groups via integration against Haar measure (the unitarian trick, taken up in 07.07.02) and to reductive algebraic groups in characteristic zero. The foundational reason — an invariant projection manufactured by summing over the symmetry — appears again in the character-orthogonality proof in the full proof set, where averaging a matrix coefficient over $G$ forces the Schur-lemma diagonalisation. This is exactly the mechanism that makes the representation ring $R (G)$ a free abelian group on irreducibles, and putting these together, the whole of finite-group representation theory collapses to a clean two-step programme: classify the irreducibles, then read off every representation by inner products of characters.

Exercises Intermediate+

Exercise 4 (medium, proof).

Decompose the 3-dimensional permutation representation of $S_{3}$ on $C^{3}$ (acting by permuting coordinates) into irreducibles, using inner products of characters.

Hint

Compute the character of the permutation representation, then take its inner product with each of $χ_{1}, χ_{2}, χ_{3}$ .

Answer

The permutation representation has $χ_{perm} (g) = # {fixed points of g}$ , giving $χ_{perm} (e) = 3$ , $χ_{perm} ((12)) = 1$ , $χ_{perm} ((123)) = 0$ . Multiplicities:

⟨ χ_{perm}, χ_{1} ⟩ = \frac{1}{6} (3 \cdot 1 \cdot 1 + 1 \cdot 1 \cdot 3 + 0 \cdot 1 \cdot 2) = \frac{6}{6} = 1,

⟨ χ_{perm}, χ_{2} ⟩ = \frac{1}{6} (3 \cdot 1 + 1 \cdot (- 1) \cdot 3 + 0) = \frac{1}{6} (3 - 3) = 0,

⟨ χ_{perm}, χ_{3} ⟩ = \frac{1}{6} (3 \cdot 2 \cdot 1 + 1 \cdot 0 \cdot 3 + 0 \cdot (- 1) \cdot 2) = \frac{6}{6} = 1.

So the permutation representation decomposes as $const \oplus standard$ , i.e. $χ_{perm} = χ_{1} + χ_{3}$ . Geometrically: the line spanned by $(1, 1, 1)$ carries the constant-one representation, and the complementary plane ${x_{1} + x_{2} + x_{3} = 0}$ carries the standard representation. $□$

Exercise 6 (hard, proof).

State and prove the column orthogonality relation: for any $g, h \in G$ ,

V irred. \sum χ_{V} (g) \overline{χ_{V} (h)} = {∣ C_{G} (g) ∣ 0 g \sim h, g \neq \sim h,

where $C_{G} (g)$ is the centraliser of $g$ .

Hint

The first orthogonality relation says the rows of the weighted character table are orthonormal. Since the table is square, its columns are orthonormal too — re-interpret that statement.

Answer

Let $r$ be the number of conjugacy classes and let the irreducible characters be $χ_{1}, \dots, χ_{r}$ . Form the $r \times r$ matrix $M$ with entries $M_{ij} = ∣ C_{i} ∣/∣ G ∣ \overline{χ_{i} (g_{j})}$ , where $g_{j}$ ranges over conjugacy-class representatives and $∣ C_{i} ∣$ ... — careful, here $∣ C_{j} ∣ = ∣ G ∣/∣ cl (g_{j}) ∣$ is the centraliser order of class $j$ . The row-orthogonality relation $⟨ χ_{i}, χ_{k} ⟩ = δ_{ik}$ reads $\sum_{j} M_{ij} \overline{M_{k j}} = δ_{ik}$ , i.e. the rows of $M$ are orthonormal. A square matrix with orthonormal rows has orthonormal columns, so $\sum_{i} M_{ij} \overline{M_{i ℓ}} = δ_{j ℓ}$ , which unpacks to

\frac{∣ C _{j} ∣ ∣ C _{ℓ} ∣}{∣ G ∣} i \sum \overline{χ_{i} (g_{j})} χ_{i} (g_{ℓ}) = δ_{j ℓ} .

For $j = ℓ$ this gives $\sum_{i} ∣ χ_{i} (g_{j}) ∣^{2} = ∣ G ∣/∣ C_{j} ∣ = ∣ cl (g_{j}) ∣$ ... — precisely $∣ G ∣/∣ C_{G} (g_{j}) ∣$ equals the size of the conjugacy class, and re-arranging with $g = g_{j}, h = g_{ℓ}$ yields the displayed column orthogonality (zero when the classes differ, $∣ C_{G} (g) ∣$ when they coincide). $□$

The content: the character table of a finite group, weighted by $∣ cl ∣/∣ G ∣$ , is a unitary matrix. Row orthogonality and column orthogonality are dual halves of that single fact.

Exercise 7 (hard, conceptual).

Explain why the representation ring $R (G)$ is a free abelian group of rank equal to the number of conjugacy classes of $G$ , and why its multiplication constants $c_{ij}^{k}$ are non-negative integers.

Hint

Use Maschke to decompose every tensor product of irreducibles into a direct sum of irreducibles, and read off the multiplicities via the inner product.

Answer

The Grothendieck group construction identifies two representations when they have the same composition factors; by Maschke, composition factors are just irreducible summands with multiplicity. So every element of $R (G)$ is uniquely a $Z$ -linear combination $\sum_{i} n_{i} [χ_{i}]$ of the irreducible classes, making $R (G) ≅ Z^{r}$ free abelian of rank $r$ = number of irreducibles = number of conjugacy classes.

Multiplication comes from tensor product. For irreducibles $V_{i}, V_{j}$ , Maschke gives a decomposition $V_{i} \otimes V_{j} ≅ ⨁_{k} V_{k}^{\oplus c_{ij}^{k}}$ , and each multiplicity is

c_{ij}^{k} = ⟨ χ_{i} χ_{j}, χ_{k} ⟩_{G} \in Z_{\geq 0},

non-negative because it counts summands. These $c_{ij}^{k}$ are the structure constants of $R (G)$ ; computing them is the central combinatorial problem of the subject. For $G = S_{n}$ they are the Kronecker coefficients of symmetric-function theory, whose combinatorial description remains open. $□$

Lean formalization Intermediate+

lean_status: none — this unit does not yet carry a Lean module. the Mathlib gap analysis records the precise state of Mathlib's RepresentationTheory directory: Maschke's theorem and the basic character API are in place, but the orthogonality relations, the Artin–Wedderburn decomposition of MonoidAlgebra k G, Frobenius reciprocity at the character level, and the representation ring $R (G)$ are not yet formalised. Producing a machine-checked proof of column orthogonality and the $∣ G ∣ = \sum_{i} (dim V_{i})^{2}$ identity is the natural first target for a future formalization pass; it would unlock automated verification of every multiplicity and character-table computation in this strand.

Advanced results Master

Artin–Wedderburn and the structure of $C G$ . Maschke says $C G$ is semisimple; Artin–Wedderburn refines this to an explicit product of matrix algebras,

C G ≅ i = 1 \prod r M_{d i m V_{i}} (C),

with one factor per irreducible. This single isomorphism encodes both the dimension formula $\sum (dim V_{i})^{2} = ∣ G ∣$ and the multiplicity statement that every irreducible $V_{i}$ appears in the regular representation exactly $dim V_{i}$ times. Over $R$ , the analogous decomposition features matrix algebras over $R$ , $C$ , and the quaternions $H$ — the real Frobenius–Schur classification of irreducibles into real, complex, and quaternionic types 07.01.12.

Row and column orthogonality. The first orthogonality relation $⟨ χ_{V}, χ_{W} ⟩_{G} = δ_{V W}$ makes the irreducible characters an orthonormal set in the space of class functions. The column orthogonality relation (Exercise 6) is its dual: summing over irreducibles rather than group elements gives $∣ C_{G} (g) ∣$ when $g, h$ are conjugate and $0$ otherwise. Together they say the character table, suitably weighted by $∣ cl (g) ∣/∣ G ∣$ , is a unitary matrix. This is why character tables are so computationally powerful: a square unitary matrix is determined by either its rows or its columns, and partial information about one determines the rest.

The class-number theorem. Combining orthonormality with the regular-character identity $χ_{reg} = \sum_{i} (dim V_{i}) χ_{i}$ proves that the irreducible characters form not just an orthonormal set but an orthonormal basis of the class functions. Since the space of class functions has dimension equal to the number $r$ of conjugacy classes, the number of irreducible complex representations of $G$ is exactly $r$ . This coincidence — the deepest structural fact of ordinary character theory — is what makes the character table square.

Frobenius reciprocity and Mackey theory. The adjunction $Hom_{G} (Ind_{H}^{G} W, V) ≅ Hom_{H} (W, Res_{H}^{G} V)$ is the cornerstone of induction–restriction machinery. Mackey's decomposition theorem refines the restriction of an induced representation across two subgroups, leading to the Mackey irreducibility criterion and, ultimately, to the classification of representations of semidirect products. For finite groups of Lie type, Deligne–Lusztig theory replaces Mackey's set-theoretic double cosets with $ℓ$ -adic cohomology of varieties, generalising Frobenius reciprocity to a geometric setting.

The representation ring and virtual characters. $R (G)$ is a commutative ring carrying a positive-definite inner product $⟨ \cdot, \cdot ⟩_{G}$ , and the map $χ : R (G) ↪ {class functions}$ embeds it as a lattice of full rank. The augmentation $ε : R (G) \to Z$ sends a representation to its dimension, and the Adams operations $ψ^{k}$ (sending $χ_{V}$ to the class function $g \mapsto χ_{V} (g^{k})$ ) make $R (G)$ a $λ$ -ring, connecting representation theory to stable cohomology and $K$ -theory.

Burnside's $p^{a} q^{b}$ solvability theorem. Every finite group of order $p^{a} q^{b}$ (for primes $p, q$ ) is solvable. Burnside's 1904 proof uses character theory: if a Sylow subgroup $P$ of $G$ acts on the set of irreducible characters, the fixed-point analysis plus column orthogonality forces the existence of a proper normal subgroup, inducting on $∣ G ∣$ . This was historically the first triumph of character theory as a structural tool, not just a classification device. (The purely group-theoretic proof arrived only with Thompson and Feit–Thompson decades later.)

Synthesis. Finite-group representation theory is the prototype of the whole symmetry-linearisation programme, and the argument that builds toward every later development in the strand: the averaging that proves Maschke generalises to the Haar integral for compact Lie groups, and the foundational reason — manufacture invariance by summing over the symmetry — appears again in the orthogonality proof and in the Reynolds operator of invariant theory. This is exactly the mechanism that makes the representation ring $R (G)$ a free abelian group on irreducibles, putting these together with column orthogonality yields the square character table and the class-number theorem, and the bridge is that characters detect everything: the entire category of finite-dimensional $C G$ -modules is encoded in one finite square array of complex numbers, whose rows are orthonormal and whose size equals the number of conjugacy classes. From that single array one reads off decomposition multiplicities, induced-representation characters via Frobenius reciprocity, tensor-product constants via the $c_{ij}^{k}$ , and — as Burnside showed — the solvability of broad classes of groups.

Full proof set Master

The key theorems of ordinary character theory are collected here with complete proofs, complementing the averaging proof of Maschke already given.

Proposition (character orthogonality, row form). Let $V, W$ be irreducible complex representations of a finite group $G$ . Then $⟨ χ_{V}, χ_{W} ⟩_{G} = δ_{V W}$ .

Proof. The argument goes through matrix coefficients. For irreducible $V$ with basis ${e_{i}}$ , define the function $ρ_{ij}^{V} (g) := ⟨ e_{i}^{*}, ρ (g) e_{j} ⟩$ , the $(i, j)$ -entry of $ρ (g)$ . A direct computation using Schur's lemma gives the matrix-coefficient orthogonality

\frac{1}{∣ G ∣} g \in G \sum ρ_{ij}^{V} (g^{- 1}) ρ_{k l}^{W} (g) = \frac{1}{dim V} δ_{V W} δ_{i l} δ_{j k} .

To see this, fix indices and consider the linear map $T_{j k} : V \to V$ sending $e_{l} \mapsto e_{j}$ and other basis vectors to zero; the averaged intertwiner $\frac{1}{∣ G ∣} \sum_{g} ρ (g)^{- 1} T_{j k} ρ (g)$ is $G$ -equivariant, hence by Schur's lemma a scalar $λ \cdot id_{V}$ . Taking the $(i, l)$ -entry and computing the trace identifies $λ = tr (T_{j k}) / dim V = δ_{j k} / dim V$ , which is the displayed formula. Now sum the matrix-coefficient orthogonality over $i = j$ and $k = l$ : the left side becomes $\frac{1}{∣ G ∣} \sum_{g} χ_{V} (g^{- 1}) χ_{W} (g) = \frac{1}{∣ G ∣} \sum_{g} \overline{χ_{V} (g)} χ_{W} (g) = ⟨ χ_{W}, χ_{V} ⟩_{G}$ (using Exercise 2), and the right side becomes $\frac{1}{d i m V} δ_{V W} (dim V)^{2} = (dim V) δ_{V W}$ ... — carefully, the contraction $\sum_{i} \sum_{k} δ_{ik} δ_{k i}$ contributes $dim V$ when $V ≅ W$ , giving $⟨ χ_{V}, χ_{W} ⟩_{G} = δ_{V W}$ . $□$

Proposition (irreducibles $\leftrightarrow$ conjugacy classes). The number of irreducible complex representations of $G$ equals the number $r$ of conjugacy classes.

Proof. Let $χ_{1}, \dots, χ_{s}$ be the irreducible characters; row orthogonality makes them an orthonormal set, so $s \leq r$ (the space of class functions has dimension $r$ ). For the reverse inequality, suppose a class function $α$ is orthogonal to every irreducible character: $⟨ α, χ_{i} ⟩ = 0$ for all $i$ . Consider the operator $T = \frac{1}{∣ G ∣} \sum_{g} α (g) ρ_{i} (g)$ on each irreducible $V_{i}$ ; by the same Schur-lemma averaging as above, $T = \frac{⟨ α , χ _{i} ⟩}{d i m V _{i}} id = 0$ . By Maschke, $T$ vanishes on every representation, in particular on the regular representation. Evaluating $T$ on the basis vector $e_{e} \in C G$ gives $0 = T (e_{e}) = \frac{1}{∣ G ∣} \sum_{g} α (g) e_{g}$ , which (the $e_{g}$ are linearly independent) forces $α (g) = 0$ for all $g$ . So the only class function orthogonal to all irreducible characters is zero, meaning the irreducible characters span the class functions, hence $s = r$ . $□$

Proposition (Frobenius reciprocity, character form). For $H \leq G$ , an $H$ -representation $W$ , and a $G$ -representation $V$ , $⟨ χ_{Ind_{H}^{G} W}, χ_{V} ⟩_{G} = ⟨ χ_{W}, χ_{Res_{H}^{G} V} ⟩_{H}$ .

Proof. Unpack both inner products using the induced-character formula. The left side is

\frac{1}{∣ G ∣} g \in G \sum \overline{χ_{V} (g)} \cdot \frac{1}{∣ H ∣} x \in G xg x^{- 1} \in H \sum χ_{W} (xg x^{- 1}) .

Reorganise the double sum over pairs $(g, x)$ with $xg x^{- 1} \in H$ , substitute $h = xg x^{- 1}$ (so $g = x^{- 1} h x$ , and $χ_{V} (g) = χ_{V} (x^{- 1} h x) = χ_{V} (h)$ because characters are class functions on $G$ ... here extended), and use that each $h \in H$ is hit by exactly $∣ G ∣/∣ H ∣ \cdot$ ... pairs. After the standard re-indexing (a bijection on the relevant finite set), the factors of $∣ G ∣$ and $∣ H ∣$ cancel to leave

\frac{1}{∣ H ∣} h \in H \sum \overline{χ_{V} (h)} χ_{W} (h) = ⟨ χ_{W}, χ_{Res_{H}^{G} V} ⟩_{H},

as required. The conceptual content is that $Ind$ and $Res$ are adjoint functors, and the character inner product is the trace-form that witnesses the adjunction. $□$

These three propositions — row orthogonality, the class-number theorem, and Frobenius reciprocity — together with Maschke form the complete logical core of ordinary character theory. Every further result (column orthogonality, the multiplicity formula, the dimension identity, Burnside's solvability theorem) is a formal consequence of these four.

Connections Master

Group representation 07.01.01 — this depth unit presupposes the foundational definition of a representation as a homomorphism $G \to GL (V)$ and as a $k G$ -module, and it deepens that foundation by proving the structural theorems (Maschke, orthogonality, the class-number theorem) that the foundational unit only states.
Maschke's theorem 07.02.01 — the present unit reproduces the averaging proof of Maschke in its key-theorem section as the load-bearing structural result, and extends it with the Artin–Wedderburn consequence that $C G$ is a product of matrix algebras; the companion unit treats Maschke's failure in modular characteristic in depth.
Character of a representation 07.01.03 and character orthogonality 07.01.04 — the foundational definitions of characters and the orthogonality relations live here; this depth unit proves both row and column orthogonality in full, derives the class-number theorem, and exhibits the $S_{3}$ character table as the worked example.
Induced representation 07.01.07 and Frobenius reciprocity 07.01.08 — induced representations and the induction–restriction adjunction are defined here and proved in the full proof set; Mackey theory and the representation-theoretic analysis of semidirect products build directly on this material.
Symmetric-group representations 07.05.02 and Specht modules 07.05.03 — the $S_{3}$ character table computed in the worked example is the smallest instance of the general theory of symmetric-group irreducibles indexed by Young diagrams; the representation-ring structure constants $c_{ij}^{k}$ for $S_{n}$ are the Kronecker coefficients of symmetric-function theory.
Brauer character and block theory 07.02.04 — Maschke's sharpness hypothesis ( $char (k) ∤ ∣ G ∣$ ) marks the boundary of ordinary character theory; beyond it, modular representation theory replaces characters with Brauer characters and organises the category into blocks with defect groups.
Lang, Algebra, Ch. XVIII [foundational-home] — the Artin–Wedderburn theorem for semisimple algebras and the resulting structure theory of $k G$ provide the algebraic backdrop against which the character-theoretic results of this unit are read; the group algebra dictionary between representations and modules is the unifying perspective.

Historical & philosophical context Master

The systematic theory of group characters was created by Frobenius in a burst of papers beginning in 1896. His Über Gruppencharaktere ^{[Frobenius 1896]} introduced the character of a representation as a class function and the group determinant, and derived the multiplication law for characters — the formula that makes characters a computational tool rather than a mere invariant. Frobenius arrived at characters not through representations but through the determinant of the matrix $(x_{g h^{- 1}})$ of formal variables, factoring it into irreducible factors over $C$ ; the modern definition $χ (g) = tr ρ (g)$ came slightly later, clarifying what the determinant factors actually were.

Maschke supplied the complete-reducibility theorem in 1899 ^{[Maschke 1899]}, proving that every finite-group representation over a characteristic-zero field splits into irreducibles. The averaging trick at the heart of his proof — summing an arbitrary projection over the group and dividing by $∣ G ∣$ — was already implicit in Frobenius's earlier work, but Maschke isolated the geometric content: every invariant subspace has an invariant complement. This is the result that makes the entire theory semisimple and hence tractable.

Burnside and Schur consolidated the theory in the first decade of the twentieth century. Burnside's Theory of Groups of Finite Order ^{[Burnside 1911]} gave the first textbook account and used character theory to prove the $p^{a} q^{b}$ solvability theorem — historically the first demonstration that representation theory could resolve pure group-theoretic questions inaccessible to direct methods. Schur's lemma, the orthogonality relations in their modern trace-theoretic form, and the Schur index all date to Schur's work in this period.

The philosophical pivot of the whole theory is the discovery that the trace forgets almost nothing: although a representation is a great deal of data (a matrix for every group element), its character — a single complex number per conjugacy class — already determines it up to isomorphism, provided Maschke holds. This compression, from $∣ G ∣$ matrices to $r$ numbers, is what makes character tables practical, and the failure of the compression in the modular case is what makes modular representation theory so much harder. The representation ring $R (G)$ packages this compression functorially, turning the category of representations into a commutative ring that braids representation theory with $K$ -theory, stable homotopy, and — for $G = S_{n}$ — the algebra of symmetric functions.

Bibliography Master

@book{serre1977linear,
  author = {Serre, Jean-Pierre},
  title = {Linear Representations of Finite Groups},
  series = {Graduate Texts in Mathematics},
  volume = {42},
  publisher = {Springer-Verlag},
  address = {New York},
  year = {1977}
}

@book{fulton1991representation,
  author = {Fulton, William and Harris, Joe},
  title = {Representation Theory: A First Course},
  series = {Graduate Texts in Mathematics},
  volume = {129},
  publisher = {Springer-Verlag},
  address = {New York},
  year = {1991}
}

@book{lang2002algebra,
  author = {Lang, Serge},
  title = {Algebra},
  series = {Graduate Texts in Mathematics},
  volume = {211},
  edition = {3},
  publisher = {Springer-Verlag},
  address = {New York},
  year = {2002}
}

@article{frobenius1896gruppencharaktere,
  author = {Frobenius, Ferdinand Georg},
  title = {\"Uber Gruppencharaktere},
  journal = {Sitzungsberichte der K\"oniglich Preu{\ss}ischen Akademie der Wissenschaften zu Berlin},
  year = {1896},
  pages = {985--1021}
}

@article{maschke1899beweis,
  author = {Maschke, Heinrich},
  title = {Beweis des Satzes, dass diejenigen endlichen linearen Substitutionsgruppen, in welchen einige durchgehends verschwindende Coefficienten auftreten, intransitiv sind},
  journal = {Mathematische Annalen},
  volume = {52},
  number = {3},
  pages = {363--368},
  year = {1899}
}

@book{burnside1911theory,
  author = {Burnside, William},
  title = {Theory of Groups of Finite Order},
  edition = {2},
  publisher = {Cambridge University Press},
  address = {Cambridge},
  year = {1911}
}

@book{isaacs1976character,
  author = {Isaacs, I. Martin},
  title = {Character Theory of Finite Groups},
  publisher = {Academic Press},
  address = {New York},
  year = {1976},
  note = {Reprinted by Dover Publications, 1994}
}

@book{feit1967characters,
  author = {Feit, Walter},
  title = {Characters of Finite Groups},
  publisher = {W. A. Benjamin},
  address = {New York},
  year = {1967}
}

@book{curtis1999pioneers,
  author = {Curtis, Charles W.},
  title = {Pioneers of Representation Theory: Frobenius, Burnside, Schur, and Brauer},
  series = {History of Mathematics},
  volume = {15},
  publisher = {American Mathematical Society},
  address = {Providence, RI},
  year = {1999}
}

Prerequisites

07.01.01
01.02.01

Tier anchors

beginner: Intuition for turning group multiplication into matrix multiplication, losslessly; accessible via 3Blue1Brown-style linear-algebra pictures of $S_{3}$ acting on a triangle
intermediate: Fulton & Harris Representation Theory: A First Course (Springer GTM 129, 1991) Part I §§1–2; Serre Linear Representations of Finite Groups (Springer GTM 42, 1977) §§1–6
master: Serre Linear Representations of Finite Groups; Feit Characters of Finite Groups (Academic Press, 1967); Isaacs Character Theory of Finite Groups (Academic Press, 1976)

References

Serre — Linear Representations of Finite Groups · Springer GTM 42 (1977). §§1–6: Maschke's theorem via the averaging projection $\tilde{π} = \frac{1}{∣ G ∣} \sum_{g} ρ (g) π ρ (g)^{- 1}$ , characters $χ_{V} (g) = tr ρ (g)$ , the orthogonality relations $⟨ χ_{V}, χ_{W} ⟩ = δ_{V W}$ , the regular representation giving $\sum_{i} (dim V_{i})^{2} = ∣ G ∣$ , induced representations and Frobenius reciprocity $Hom_{G} (Ind_{H}^{G} W, V) ≅ Hom_{H} (W, Res_{H}^{G} V)$ .
Fulton & Harris — Representation Theory: A First Course · Springer GTM 129 (1991). Part I §§1–2: the group algebra $C G$ , Maschke and complete reducibility, the character tables of $S_{3}$ and $S_{4}$ , the standard representation, and the first orthogonality relation.
Lang — Algebra · Springer GTM 211 (3rd ed., 2002). Ch. XVIII §§4–5: semisimple algebras, the Wedderburn structure theorem $k G ≅ \prod_{i} M_{n_{i}} (D_{i})$ for a finite group $G$ over a splitting field of characteristic zero, characters and orthogonality.
Frobenius — Über Gruppencharaktere (1896) · Sitzungsberichte der Königlich Preußischen Akademie der Wissenschaften zu Berlin (1896), 985–1021. The founding paper introducing group characters $χ (g)$ and the group determinant, with the multiplication law for characters that launched the entire theory.
Maschke — Beweis des Satzes, dass diejenigen endlichen linearen Substitutionsgruppen... intransitiv sind (1899) · Mathematische Annalen 52 (1899), 363–368. The complete-reducibility theorem: every finite-group representation over a characteristic-zero field splits as a direct sum of irreducible representations.
Burnside — Theory of Groups of Finite Order · Cambridge University Press (2nd ed., 1911). Chs. XV–XVIII: orthogonality of characters, the $p^{a} q^{b}$ solvability theorem proved character-theoretically, and the first systematic textbook account.
Isaacs — Character Theory of Finite Groups · Academic Press (1976; reprinted by Dover). The graduate reference: character rings $R (G)$ , induced characters, Frobenius reciprocity, the Schur index, and Burnside's solvability theorem.

Estimated time

beginner: 20 min
intermediate: 55 min
master: 110 min