Posts 01 and 02 studied groups one at a time: their elements, subgroups, cosets and quotients. This post studies the maps between groups and the ways a group can move a set. Homomorphisms explain what quotient groups really are (every quotient is the image of a homomorphism, and every image is a quotient). Group actions turn a group into a counting machine: the orbit–stabilizer theorem, the class equation and Burnside's lemma all come from one idea. The payoff is the Sylow theorems, which let you prove facts such as "every group of order 15 is cyclic" using nothing but divisibility. The chapter closes with the classification of finite abelian groups.

Series map: 00 Integers, Relations & Modular Arithmetic → 01 Groups & Subgroups → 02 Cosets, Lagrange & Quotient Groups → 03 Homomorphisms, Group Actions & Sylow (this post) → 04 Rings, Ideals & Fields → 05 Field Extensions & Galois Theory.


Learning Outcome

  • Define a group homomorphism, its kernel and its image, and prove that the kernel is a normal subgroup.
  • Decide whether two groups are isomorphic by comparing invariants (order, element orders, abelian or not).
  • Prove that Aut⁡(Z/nZ)≅(Z/nZ)×\operatorname{Aut}(\mathbb{Z}/n\mathbb{Z}) \cong (\mathbb{Z}/n\mathbb{Z})^\times.
  • State and prove the First Isomorphism Theorem, and state the Second, Third and Correspondence theorems.
  • Prove Cayley's theorem: every group is a group of permutations.
  • Define a group action, orbits and stabilizers, and prove the orbit–stabilizer theorem.
  • Derive the class equation and use it to prove that a pp-group has a nontrivial center.
  • Count colourings up to symmetry with Burnside's lemma.
  • State Cauchy's and Sylow's theorems and use them to analyse groups of order 12, 15 and 20.
  • List all abelian groups of a given order, in both elementary-divisor and invariant-factor form.

Chapter 1 — Homomorphisms

Terminology

  • Homomorphism — a map φ:G→H\varphi : G \to H between groups with φ(ab)=φ(a)φ(b)\varphi(ab) = \varphi(a)\varphi(b) for all a,b∈Ga, b \in G. The product on the left is in GG, the one on the right is in HH.
  • Kernel — ker⁡φ={g∈G:φ(g)=eH}\ker \varphi = \{ g \in G : \varphi(g) = e_H \}, the elements sent to the identity.
  • Image — im⁡φ=φ(G)={φ(g):g∈G}\operatorname{im} \varphi = \varphi(G) = \{ \varphi(g) : g \in G \}.
  • Monomorphism / epimorphism — an injective / surjective homomorphism.
  • Isomorphism — a bijective homomorphism. Write G≅HG \cong H when one exists.
  • Automorphism — an isomorphism G→GG \to G. The automorphisms of GG form a group Aut⁡(G)\operatorname{Aut}(G) under composition.
  • Inner automorphism — conjugation by a fixed element: cg(x)=gxg−1c_g(x) = gxg^{-1}.

First properties

Let φ:G→H\varphi : G \to H be a homomorphism. Then:

  1. φ(eG)=eH\varphi(e_G) = e_H. Proof: φ(e)=φ(ee)=φ(e)φ(e)\varphi(e) = \varphi(ee) = \varphi(e)\varphi(e); multiply both sides by φ(e)−1\varphi(e)^{-1}.
  2. φ(g−1)=φ(g)−1\varphi(g^{-1}) = \varphi(g)^{-1}. Proof: φ(g)φ(g−1)=φ(gg−1)=φ(e)=eH\varphi(g)\varphi(g^{-1}) = \varphi(gg^{-1}) = \varphi(e) = e_H.
  3. φ(gk)=φ(g)k\varphi(g^k) = \varphi(g)^k for every integer kk (induction, plus property 2 for negative kk).
  4. If gg has finite order nn, then the order of φ(g)\varphi(g) divides nn, because φ(g)n=φ(gn)=eH\varphi(g)^n = \varphi(g^n) = e_H.
  5. ker⁡φ\ker \varphi is a subgroup of GG and im⁡φ\operatorname{im}\varphi is a subgroup of HH.

Property 4 is the workhorse for ruling maps out. There is no nontrivial homomorphism Z/3Z→Z/4Z\mathbb{Z}/3\mathbb{Z} \to \mathbb{Z}/4\mathbb{Z}: the image of the generator would need an order dividing both 3 and 4, so it is the identity.

Examples worth knowing

Homomorphism Kernel Image
det⁡:GLn(R)→R×\det : GL_n(\mathbb{R}) \to \mathbb{R}^\times SLn(R)SL_n(\mathbb{R}) all of R×\mathbb{R}^\times
sgn⁡:Sn→{±1}\operatorname{sgn} : S_n \to \{\pm 1\}, n≥2n \ge 2 AnA_n {±1}\{\pm 1\}
Z→Z/nZ\mathbb{Z} \to \mathbb{Z}/n\mathbb{Z}, k↦k mod nk \mapsto k \bmod n nZn\mathbb{Z} all of Z/nZ\mathbb{Z}/n\mathbb{Z}
exp⁡:(R,+)→(R>0,⋅)\exp : (\mathbb{R}, +) \to (\mathbb{R}_{>0}, \cdot) {0}\{0\} all of R>0\mathbb{R}_{>0} (isomorphism)
(R,+)→S1(\mathbb{R}, +) \to S^1, t↦e2πitt \mapsto e^{2\pi i t} Z\mathbb{Z} the unit circle S1S^1
cg:G→Gc_g : G \to G, x↦gxg−1x \mapsto gxg^{-1} {e}\{e\} GG (automorphism)

Homomorphisms out of a cyclic group. A homomorphism φ:Z/nZ→Z/mZ\varphi : \mathbb{Z}/n\mathbb{Z} \to \mathbb{Z}/m\mathbb{Z} is fixed by a=φ(1)a = \varphi(1), since φ(k)=ka\varphi(k) = ka. The only condition is that n⋅1=0n \cdot 1 = 0 must go to 00, that is na≡0(modm)na \equiv 0 \pmod m. The solutions are the multiples of m/gcd⁡(n,m)m/\gcd(n,m), so there are exactly gcd⁡(n,m)\gcd(n, m) homomorphisms. For n=12n = 12, m=18m = 18: 12a≡0(mod18)12a \equiv 0 \pmod{18} holds for a∈{0,3,6,9,12,15}a \in \{0, 3, 6, 9, 12, 15\}, which is gcd⁡(12,18)=6\gcd(12, 18) = 6 maps.

Common pitfall: "k↦k mod mk \mapsto k \bmod m" is not a well-defined map Z/nZ→Z/mZ\mathbb{Z}/n\mathbb{Z} \to \mathbb{Z}/m\mathbb{Z} unless m∣nm \mid n. In Z/4Z\mathbb{Z}/4\mathbb{Z}, 0=40 = 4, but 0 mod 3≠4 mod 30 \bmod 3 \ne 4 \bmod 3. Always check that a map defined on representatives does not depend on the representative chosen.

Kernels are normal, and injectivity

Theorem. If φ:G→H\varphi : G \to H is a homomorphism, then ker⁡φ\ker\varphi is a normal subgroup of GG.

Proof. It is a subgroup: φ(e)=e\varphi(e) = e; if φ(a)=φ(b)=e\varphi(a) = \varphi(b) = e then φ(ab−1)=φ(a)φ(b)−1=e\varphi(ab^{-1}) = \varphi(a)\varphi(b)^{-1} = e. It is normal: for k∈ker⁡φk \in \ker\varphi and any g∈Gg \in G,

φ(gkg−1)=φ(g) φ(k) φ(g)−1=φ(g) e φ(g)−1=e,\varphi(gkg^{-1}) = \varphi(g)\,\varphi(k)\,\varphi(g)^{-1} = \varphi(g)\,e\,\varphi(g)^{-1} = e,

so gkg−1∈ker⁡φgkg^{-1} \in \ker\varphi. ∎

Theorem. φ\varphi is injective if and only if ker⁡φ={e}\ker\varphi = \{e\}.

Proof. If φ\varphi is injective, only ee maps to ee. Conversely, if ker⁡φ={e}\ker\varphi = \{e\} and φ(a)=φ(b)\varphi(a) = \varphi(b), then φ(ab−1)=e\varphi(ab^{-1}) = e, so ab−1=eab^{-1} = e and a=ba = b. ∎

More precisely, the fibres of φ\varphi are exactly the cosets of the kernel: φ(a)=φ(b)\varphi(a) = \varphi(b) if and only if aker⁡φ=bker⁡φa\ker\varphi = b\ker\varphi. This is the whole content of the First Isomorphism Theorem in Chapter 2.

Isomorphism invariants

Isomorphic groups are "the same group with renamed elements", so any property defined only from the group operation is shared. To prove two groups are not isomorphic, find a property one has and the other lacks:

  • the order ∣G∣|G|;
  • being abelian or cyclic;
  • the number of elements of each order;
  • the size of the center, the number of subgroups of a given order, and so on.

Z/4Z\mathbb{Z}/4\mathbb{Z} has an element of order 4; in Z/2Z×Z/2Z\mathbb{Z}/2\mathbb{Z} \times \mathbb{Z}/2\mathbb{Z} every non-identity element has order 2. So they are not isomorphic, although both have order 4 and are abelian. On the other hand Z/6Z≅Z/2Z×Z/3Z\mathbb{Z}/6\mathbb{Z} \cong \mathbb{Z}/2\mathbb{Z} \times \mathbb{Z}/3\mathbb{Z} via k↦(k mod 2,k mod 3)k \mapsto (k \bmod 2, k \bmod 3); this is the Chinese Remainder Theorem from post 00. In general Z/mZ×Z/nZ\mathbb{Z}/m\mathbb{Z} \times \mathbb{Z}/n\mathbb{Z} is cyclic if and only if gcd⁡(m,n)=1\gcd(m, n) = 1.

Watch out: to prove two groups are isomorphic you must produce a map and check it. Matching every invariant you can think of is evidence, not proof.

Automorphisms of cyclic groups

Theorem. Aut⁡(Z/nZ)≅(Z/nZ)×\operatorname{Aut}(\mathbb{Z}/n\mathbb{Z}) \cong (\mathbb{Z}/n\mathbb{Z})^\times, the group of units modulo nn. In particular ∣Aut⁡(Z/nZ)∣=ϕ(n)\lvert\operatorname{Aut}(\mathbb{Z}/n\mathbb{Z})\rvert = \phi(n), Euler's totient function from post 00. (This post writes Euler's function as ϕ\phi to keep φ\varphi free for homomorphisms.)

Proof. Every endomorphism of Z/nZ\mathbb{Z}/n\mathbb{Z} is μa(k)=ak\mu_a(k) = ak for a=μa(1)a = \mu_a(1) (see "Homomorphisms out of a cyclic group" with m=nm = n). The map μa\mu_a is a bijection exactly when its image contains 11, that is when ab≡1(modn)ab \equiv 1 \pmod n for some bb, that is when aa is a unit. Define Ψ:(Z/nZ)×→Aut⁡(Z/nZ)\Psi : (\mathbb{Z}/n\mathbb{Z})^\times \to \operatorname{Aut}(\mathbb{Z}/n\mathbb{Z}), a↦μaa \mapsto \mu_a. It is surjective by what was just said and injective since μa(1)=a\mu_a(1) = a recovers aa. It is a homomorphism because μa∘μb(k)=abk=μab(k)\mu_a \circ \mu_b (k) = abk = \mu_{ab}(k). ∎

Example: (Z/8Z)×={1,3,5,7}(\mathbb{Z}/8\mathbb{Z})^\times = \{1, 3, 5, 7\} and each element squares to 11 (32=93^2 = 9, 52=255^2 = 25, 72=497^2 = 49 are all ≡1\equiv 1). So Aut⁡(Z/8Z)≅Z/2Z×Z/2Z\operatorname{Aut}(\mathbb{Z}/8\mathbb{Z}) \cong \mathbb{Z}/2\mathbb{Z} \times \mathbb{Z}/2\mathbb{Z}, not cyclic. In the other direction, Aut⁡(Z/2Z×Z/2Z)≅S3\operatorname{Aut}(\mathbb{Z}/2\mathbb{Z} \times \mathbb{Z}/2\mathbb{Z}) \cong S_3: an automorphism may permute the three non-identity elements in any of the 3!=63! = 6 ways.

Inner automorphisms. The map G→Aut⁡(G)G \to \operatorname{Aut}(G), g↦cgg \mapsto c_g, is a homomorphism (cgch=cghc_g c_h = c_{gh}). Its kernel is the set of gg with gxg−1=xgxg^{-1} = x for all xx, which is the center Z(G)Z(G) from post 02. Its image is Inn⁡(G)\operatorname{Inn}(G). By the First Isomorphism Theorem (next chapter), Inn⁡(G)≅G/Z(G)\operatorname{Inn}(G) \cong G/Z(G). For D4D_4 (order 8) the center is {e,r2}\{e, r^2\}, so Inn⁡(D4)\operatorname{Inn}(D_4) has order 8/2=48/2 = 4.


Chapter 2 — The isomorphism theorems

The First Isomorphism Theorem

Theorem (First Isomorphism Theorem). Let φ:G→H\varphi : G \to H be a homomorphism with kernel KK. Then KK is normal in GG and

G/K  ≅  im⁡φ,gK↦φ(g).G/K \;\cong\; \operatorname{im}\varphi, \qquad gK \mapsto \varphi(g).

Proof. Normality of KK was proved in Chapter 1. Define φˉ:G/K→im⁡φ\bar\varphi : G/K \to \operatorname{im}\varphi by φˉ(gK)=φ(g)\bar\varphi(gK) = \varphi(g).

  • Well defined. If gK=g′KgK = g'K then g′=gkg' = gk for some k∈Kk \in K, so φ(g′)=φ(g)φ(k)=φ(g)\varphi(g') = \varphi(g)\varphi(k) = \varphi(g).
  • Homomorphism. φˉ(gK⋅g′K)=φˉ(gg′K)=φ(gg′)=φ(g)φ(g′)=φˉ(gK) φˉ(g′K)\bar\varphi(gK \cdot g'K) = \bar\varphi(gg'K) = \varphi(gg') = \varphi(g)\varphi(g') = \bar\varphi(gK)\,\bar\varphi(g'K).
  • Injective. If φˉ(gK)=e\bar\varphi(gK) = e then φ(g)=e\varphi(g) = e, so g∈Kg \in K and gK=KgK = K, the identity of G/KG/K. A homomorphism with trivial kernel is injective.
  • Surjective. Every element of im⁡φ\operatorname{im}\varphi is φ(g)=φˉ(gK)\varphi(g) = \bar\varphi(gK). ∎

Read it as a factorisation: every homomorphism is "quotient map, then isomorphism, then inclusion", G→G/K→  ≅  im⁡φ↪HG \to G/K \xrightarrow{\;\cong\;} \operatorname{im}\varphi \hookrightarrow H. For finite GG it also gives the counting rule ∣G∣=∣ker⁡φ∣⋅∣im⁡φ∣|G| = |\ker\varphi| \cdot |\operatorname{im}\varphi|.

Worked applications of the table in Chapter 1:

  • GLn(R)/SLn(R)≅R×GL_n(\mathbb{R}) / SL_n(\mathbb{R}) \cong \mathbb{R}^\times (via det⁡\det).
  • Sn/An≅{±1}≅Z/2ZS_n / A_n \cong \{\pm 1\} \cong \mathbb{Z}/2\mathbb{Z} (via sgn⁡\operatorname{sgn}), so [Sn:An]=2[S_n : A_n] = 2 and ∣An∣=n!/2|A_n| = n!/2.
  • Z/nZ\mathbb{Z}/n\mathbb{Z} really is Z\mathbb{Z} modulo the kernel nZn\mathbb{Z} of reduction.
  • R/Z≅S1\mathbb{R}/\mathbb{Z} \cong S^1 (via t↦e2πitt \mapsto e^{2\pi i t}): "the real line wrapped around a circle".

Trick: to identify a quotient G/NG/N, do not work with cosets directly. Find a surjective homomorphism out of GG whose kernel is exactly NN; its target is the answer.

The Second and Third Isomorphism Theorems

Theorem (Second, or "diamond"). Let H≤GH \le G and N⊴GN \trianglelefteq G. Then HNHN is a subgroup of GG, H∩NH \cap N is normal in HH, and

HN/N  ≅  H/(H∩N).HN/N \;\cong\; H/(H \cap N).

(Proof idea: H→HN/NH \to HN/N, h↦hNh \mapsto hN, is surjective with kernel H∩NH \cap N; apply the First theorem.) For finite groups it gives ∣HN∣=∣H∣ ∣N∣/∣H∩N∣|HN| = |H|\,|N| / |H \cap N|.

Theorem (Third). Let N⊆KN \subseteq K be normal subgroups of GG. Then K/NK/N is normal in G/NG/N and

(G/N)/(K/N)  ≅  G/K.(G/N)\big/(K/N) \;\cong\; G/K.

(Proof idea: G/N→G/KG/N \to G/K, gN↦gKgN \mapsto gK, is well defined and surjective with kernel K/NK/N.) Example: (Z/12Z)/(4Z/12Z)≅Z/4Z(\mathbb{Z}/12\mathbb{Z}) / (4\mathbb{Z}/12\mathbb{Z}) \cong \mathbb{Z}/4\mathbb{Z}. The subgroup 4Z/12Z={0,4,8}4\mathbb{Z}/12\mathbb{Z} = \{0, 4, 8\} has order 3, and 12/3=412/3 = 4.

Theorem (Correspondence). Let N⊴GN \trianglelefteq G and π:G→G/N\pi : G \to G/N the quotient map. Then K↦K/N=π(K)K \mapsto K/N = \pi(K) is a bijection between subgroups KK of GG that contain NN and subgroups of G/NG/N. It preserves inclusion, indices ([G:K]=[G/N:K/N][G:K] = [G/N : K/N]) and normality.

So the subgroup lattice of G/NG/N is exactly the part of the lattice of GG that sits above NN.

Cayley's theorem

Theorem (Cayley). Every group GG is isomorphic to a subgroup of the symmetric group Sym⁡(G)\operatorname{Sym}(G). If ∣G∣=n|G| = n, then GG is isomorphic to a subgroup of SnS_n.

Proof. For g∈Gg \in G let λg:G→G\lambda_g : G \to G, λg(x)=gx\lambda_g(x) = gx. It is a bijection with inverse λg−1\lambda_{g^{-1}}, so λg∈Sym⁡(G)\lambda_g \in \operatorname{Sym}(G). The map Λ:G→Sym⁡(G)\Lambda : G \to \operatorname{Sym}(G), g↦λgg \mapsto \lambda_g, is a homomorphism because λgλh(x)=g(hx)=(gh)x=λgh(x)\lambda_g\lambda_h(x) = g(hx) = (gh)x = \lambda_{gh}(x). Its kernel is trivial: if λg\lambda_g is the identity then g=ge=λg(e)=eg = ge = \lambda_g(e) = e. By the First Isomorphism Theorem G≅im⁡Λ≤Sym⁡(G)G \cong \operatorname{im}\Lambda \le \operatorname{Sym}(G). ∎

Cayley's theorem is historically important (it says the abstract definition of a group loses nothing compared with "groups of permutations") but it is wasteful: it embeds a group of order 8 in S8S_8, of order 40320. Group actions on smaller sets give much better information, which is the next chapter.


Chapter 3 — Group actions

Terminology

  • Action — a group GG acts on a set XX if there is a map G×X→XG \times X \to X, (g,x)↦g⋅x(g, x) \mapsto g \cdot x, with e⋅x=xe \cdot x = x and g⋅(h⋅x)=(gh)⋅xg \cdot (h \cdot x) = (gh) \cdot x.
  • Permutation representation — equivalently, a homomorphism ρ:G→Sym⁡(X)\rho : G \to \operatorname{Sym}(X), with ρ(g)(x)=g⋅x\rho(g)(x) = g\cdot x. The two descriptions are the same thing.
  • Faithful action — one whose homomorphism ρ\rho is injective (only ee acts as the identity).
  • Orbit — G⋅x={g⋅x:g∈G}G\cdot x = \{ g\cdot x : g \in G \}, everywhere xx can be moved to.
  • Stabilizer — Gx={g∈G:g⋅x=x}G_x = \{ g \in G : g\cdot x = x \}, the elements that fix xx. It is a subgroup of GG.
  • Transitive action — one with a single orbit.
  • Fixed set — Xg={x∈X:g⋅x=x}X^g = \{ x \in X : g\cdot x = x \}, the points a given gg fixes.

Examples

  1. SnS_n acts on {1,…,n}\{1, \dots, n\} by σ⋅i=σ(i)\sigma\cdot i = \sigma(i). Transitive and faithful.
  2. D4D_4 (order 8, the symmetries of a square, notation from post 01) acts on the four vertices. Label them 1,2,3,41, 2, 3, 4 counter-clockwise; then r=(1 2 3 4)r = (1\,2\,3\,4) and, taking ss to be the reflection in the diagonal through vertices 1 and 3, s=(2 4)s = (2\,4).
  3. The rotation group of a cube (24 elements) acts on the 6 faces, the 8 vertices and the 12 edges.
  4. Any group acts on itself by left multiplication, g⋅x=gxg\cdot x = gx (this is Cayley's action), and on the left cosets G/HG/H by g⋅xH=gxHg\cdot xH = gxH.
  5. Any group acts on itself by conjugation, g⋅x=gxg−1g\cdot x = gxg^{-1}. Orbits are conjugacy classes cl⁡(x)\operatorname{cl}(x), and the stabilizer of xx is the centralizer CG(x)={g:gx=xg}C_G(x) = \{ g : gx = xg \}.

The orbits of an action partition XX: the relation "y=g⋅xy = g\cdot x for some gg" is an equivalence relation (reflexive via ee, symmetric via g−1g^{-1}, transitive via products), and orbits are its classes (equivalence relations are from post 00).

The orbit–stabilizer theorem

Theorem (orbit–stabilizer). Let a finite group GG act on XX and let x∈Xx \in X. Then

∣G⋅x∣  =  [G:Gx]  =  ∣G∣∣Gx∣.|G\cdot x| \;=\; [G : G_x] \;=\; \frac{|G|}{|G_x|}.

Proof. Define ff from the set of left cosets of GxG_x to the orbit by f(gGx)=g⋅xf(gG_x) = g\cdot x.

  • Well defined and injective. For g,h∈Gg, h \in G: g⋅x=h⋅x  ⟺  h−1g⋅x=x  ⟺  h−1g∈Gx  ⟺  gGx=hGxg\cdot x = h\cdot x \iff h^{-1}g\cdot x = x \iff h^{-1}g \in G_x \iff gG_x = hG_x. Reading the chain left to right gives injectivity; right to left gives well-definedness.
  • Surjective. Every orbit element is g⋅x=f(gGx)g\cdot x = f(gG_x).

So the orbit is in bijection with the cosets of GxG_x, and there are [G:Gx]=∣G∣/∣Gx∣[G:G_x] = |G|/|G_x| of them by Lagrange's theorem (post 02). ∎

Worked example: D4D_4 on the square. The orbit of vertex 1 is all four vertices (rotations reach each one). The stabilizer of vertex 1 is {e,s}\{e, s\}: the only non-identity symmetry fixing vertex 1 is the reflection in the diagonal through it. So ∣G⋅1∣⋅∣G1∣=4⋅2=8=∣D4∣|G\cdot 1| \cdot |G_1| = 4 \cdot 2 = 8 = |D_4|. The proof's bijection is concrete here. The four cosets of {e,s}\{e, s\} are {e,s}\{e, s\}, {r,rs}\{r, rs\}, {r2,r2s}\{r^2, r^2s\}, {r3,r3s}\{r^3, r^3s\}, and both elements of rk{e,s}r^k\{e, s\} send vertex 1 to vertex k+1k+1.

Orbit–stabilizer for the square: the 8 elements of D4 fall into 4 cosets of the stabilizer {e, s} of vertex 1, and each coset sends vertex 1 to one vertex, so orbit size 4 times stabilizer size 2 equals 8

Worked example: the cube. The rotation group acts transitively on faces, vertices and edges. A face is fixed by the 4 rotations about the axis through its centre; a vertex by the 3 rotations about the long diagonal through it; an edge by the identity and the half-turn about the axis through its midpoint. All three counts agree:

6⋅4  =  8⋅3  =  12⋅2  =  24.6 \cdot 4 \;=\; 8 \cdot 3 \;=\; 12 \cdot 2 \;=\; 24.

This is also the standard way to find the order of a symmetry group: pick an object, count where it can go, and count what fixes it.

Trick: points in the same orbit have conjugate stabilizers: Gg⋅x=g Gx g−1G_{g\cdot x} = g\,G_x\,g^{-1}. So all stabilizers along one orbit have the same size, as the formula requires.

The class equation

Apply orbit–stabilizer to conjugation. The orbit of xx is cl⁡(x)\operatorname{cl}(x) and its stabilizer is CG(x)C_G(x), so ∣cl⁡(x)∣=[G:CG(x)]|\operatorname{cl}(x)| = [G : C_G(x)], which divides ∣G∣|G|. An element has a class of size 1 exactly when it commutes with everything, that is when x∈Z(G)x \in Z(G). Splitting GG into classes and collecting the size-1 classes gives the class equation:

∣G∣  =  ∣Z(G)∣  +  ∑i=1k[G:CG(xi)],|G| \;=\; |Z(G)| \;+\; \sum_{i=1}^{k} [G : C_G(x_i)],

where x1,…,xkx_1, \dots, x_k are representatives of the classes with more than one element. Each term in the sum is a divisor of ∣G∣|G| bigger than 1.

Group Order Class sizes Center
S3S_3 6 1, 2, 3 trivial
D4D_4 8 1, 1, 2, 2, 2 {e,r2}\{e, r^2\}
A4A_4 12 1, 3, 4, 4 trivial
S4S_4 24 1, 3, 6, 6, 8 trivial
A5A_5 60 1, 12, 12, 15, 20 trivial

For D4D_4 the classes are {e}\{e\}, {r2}\{r^2\}, {r,r3}\{r, r^3\}, {s,r2s}\{s, r^2s\}, {rs,r3s}\{rs, r^3s\}. In SnS_n two permutations are conjugate exactly when they have the same cycle type; in S4S_4 the classes are the identity (1), double transpositions (3), transpositions (6), 4-cycles (6), 3-cycles (8).

p-groups have a nontrivial center

A pp-group is a group of order pnp^n for a prime pp and some n≥1n \ge 1.

Theorem. If ∣G∣=pn|G| = p^n with n≥1n \ge 1, then Z(G)≠{e}Z(G) \ne \{e\}.

Proof. In the class equation, each [G:CG(xi)][G : C_G(x_i)] is a divisor of pnp^n bigger than 1, so it is divisible by pp. The left side pnp^n is divisible by pp. Hence ∣Z(G)∣=pn−∑[G:CG(xi)]|Z(G)| = p^n - \sum [G : C_G(x_i)] is divisible by pp. Since e∈Z(G)e \in Z(G), ∣Z(G)∣≥1|Z(G)| \ge 1, so ∣Z(G)∣≥p|Z(G)| \ge p. ∎

Lemma. If G/Z(G)G/Z(G) is cyclic, then GG is abelian.

Proof. Let gZ(G)gZ(G) generate G/Z(G)G/Z(G). Every element of GG has the form gizg^i z with z∈Z(G)z \in Z(G). Two such elements commute: giz⋅gjw=gi+jzw=gjw⋅gizg^i z \cdot g^j w = g^{i+j} z w = g^j w \cdot g^i z, since z,wz, w commute with everything. ∎

Corollary. Every group of order p2p^2 is abelian, hence isomorphic to Z/p2Z\mathbb{Z}/p^2\mathbb{Z} or Z/pZ×Z/pZ\mathbb{Z}/p\mathbb{Z} \times \mathbb{Z}/p\mathbb{Z}.

Proof. ∣Z(G)∣|Z(G)| is pp or p2p^2. If it were pp, then G/Z(G)G/Z(G) would have order pp, hence be cyclic, hence GG abelian by the lemma, so Z(G)=GZ(G) = G has order p2p^2: a contradiction. So Z(G)=GZ(G) = G. The two possibilities come from Chapter 5. ∎

D4D_4 shows that the corollary stops at p2p^2: it has order 232^3, is not abelian, and its center has order 22.

Burnside's lemma

Theorem (Burnside's lemma, also called the Cauchy–Frobenius lemma). If a finite group GG acts on a finite set XX, the number of orbits is

#orbits  =  1∣G∣∑g∈G∣Xg∣,\#\text{orbits} \;=\; \frac{1}{|G|} \sum_{g \in G} |X^g|,

the average number of points fixed by an element.

Proof. Count the pairs (g,x)(g, x) with g⋅x=xg\cdot x = x in two ways. By gg first: ∑g∣Xg∣\sum_g |X^g|. By xx first: ∑x∣Gx∣=∑x∣G∣/∣G⋅x∣\sum_x |G_x| = \sum_x |G|/|G\cdot x|. Group the xx by orbits. An orbit OO contributes ∣O∣⋅∣G∣/∣O∣=∣G∣|O| \cdot |G|/|O| = |G|. So the total is ∣G∣|G| times the number of orbits. ∎

The typical use is counting colourings "up to symmetry". If gg permutes the positions with c(g)c(g) cycles, a colouring with kk colours is fixed by gg exactly when it is constant on each cycle, so ∣Xg∣=kc(g)|X^g| = k^{c(g)}.

Worked example: necklaces. Colour 6 beads on a circle black or white; two necklaces are the same if one is a rotation of the other. The group is the cyclic group of rotations by 0,1,…,50, 1, \dots, 5 steps. A rotation by jj steps has gcd⁡(j,6)\gcd(j, 6) cycles on the beads: 6,1,2,3,2,16, 1, 2, 3, 2, 1 cycles for j=0,…,5j = 0, \dots, 5. So

#necklaces=26+21+22+23+22+216=64+2+4+8+4+26=846=14.\#\text{necklaces} = \frac{2^6 + 2^1 + 2^2 + 2^3 + 2^2 + 2^1}{6} = \frac{64 + 2 + 4 + 8 + 4 + 2}{6} = \frac{84}{6} = 14.

If flips are also allowed (the dihedral group of order 12, so "bracelets"), the count drops to 13. With 3 colours and rotations only, the same formula gives (729+3+9+27+9+3)/6=780/6=130(729 + 3 + 9 + 27 + 9 + 3)/6 = 780/6 = 130.

Worked example: colouring a cube's faces. The 24 rotations of the cube, sorted by type, with the number of cycles each makes on the 6 faces:

Rotation type How many Cycles on faces
identity 1 6
90° or 270° about a face axis 6 3 (two fixed faces, one 4-cycle)
180° about a face axis 3 4 (two fixed faces, two 2-cycles)
120° or 240° about a vertex diagonal 8 2 (two 3-cycles)
180° about an edge axis 6 3 (three 2-cycles)

With 3 colours:

36+6⋅33+3⋅34+8⋅32+6⋅3324=729+162+243+72+16224=136824=57.\frac{3^6 + 6\cdot 3^3 + 3\cdot 3^4 + 8\cdot 3^2 + 6\cdot 3^3}{24} = \frac{729 + 162 + 243 + 72 + 162}{24} = \frac{1368}{24} = 57.

With 2 colours the same table gives (64+48+48+32+48)/24=240/24=10(64 + 48 + 48 + 32 + 48)/24 = 240/24 = 10.

Common pitfall: dividing the number of colourings by ∣G∣|G| (36/243^6/24 is not even an integer) is wrong because symmetric colourings have smaller orbits. Burnside's average of fixed points corrects for exactly this.


Chapter 4 — Cauchy and Sylow

Lagrange's theorem says the order of a subgroup divides ∣G∣|G|. The converse is false: A4A_4 has order 12 but no subgroup of order 6. The theorems in this chapter recover as much of the converse as is true, for prime powers.

Terminology

  • Sylow pp-subgroup — if ∣G∣=pkm|G| = p^k m with pp prime and p∤mp \nmid m, a subgroup of order exactly pkp^k (the largest power of pp allowed by Lagrange).
  • npn_p — the number of Sylow pp-subgroups of GG.
  • Normalizer — NG(H)={g:gHg−1=H}N_G(H) = \{ g : gHg^{-1} = H \}, the stabilizer of HH when GG acts on its subgroups by conjugation.
  • Simple group — a nontrivial group whose only normal subgroups are {e}\{e\} and itself (post 02).

The theorems

Theorem (Cauchy). If a prime pp divides ∣G∣|G|, then GG has an element of order pp.

(A short proof, due to McKay, lets Z/pZ\mathbb{Z}/p\mathbb{Z} rotate the pp-tuples (g1,…,gp)(g_1, \dots, g_p) with g1g2⋯gp=eg_1 g_2 \cdots g_p = e and counts fixed points modulo pp.)

Theorem (Sylow). Let ∣G∣=pkm|G| = p^k m with pp prime, k≥1k \ge 1 and p∤mp \nmid m.

  1. Existence. GG has a subgroup of order pkp^k. More generally, every pp-subgroup of GG is contained in some Sylow pp-subgroup.
  2. Conjugacy. Any two Sylow pp-subgroups are conjugate: if P,QP, Q are Sylow, then Q=gPg−1Q = gPg^{-1} for some g∈Gg \in G.
  3. Counting. np≡1(modp)n_p \equiv 1 \pmod p, and npn_p divides mm. Moreover np=[G:NG(P)]n_p = [G : N_G(P)] for any Sylow pp-subgroup PP.

Two consequences are used constantly:

  • np=1n_p = 1 if and only if the Sylow pp-subgroup is normal. (Conjugates of a Sylow subgroup are Sylow, and by part 2 all Sylow subgroups are conjugates of one of them.)
  • When pp divides ∣G∣|G| only once, the Sylow pp-subgroups have prime order pp. Two distinct subgroups of prime order meet only in {e}\{e\}, so the npn_p of them contain np(p−1)n_p(p-1) elements of order pp.

Sylow counts in familiar groups (checked by listing subgroups):

Group Order n2n_2 n3n_3 n5n_5
S3S_3 2⋅32 \cdot 3 3 1 —
A4A_4 22⋅32^2 \cdot 3 1 4 —
S4S_4 23⋅32^3 \cdot 3 3 4 —
A5A_5 22⋅3⋅52^2 \cdot 3 \cdot 5 5 10 6

Every entry obeys part 3: for A5A_5, n5=6≡1(mod5)n_5 = 6 \equiv 1 \pmod 5 and 6∣126 \mid 12; n3=10≡1(mod3)n_3 = 10 \equiv 1 \pmod 3 and 10∣2010 \mid 20.

Application 1 — every group of order 15 is cyclic

Let ∣G∣=15=3⋅5|G| = 15 = 3 \cdot 5.

  • n3n_3 divides 5 and n3≡1(mod3)n_3 \equiv 1 \pmod 3. The divisors of 5 are 1 and 5, and 5≡25 \equiv 2. So n3=1n_3 = 1.
  • n5n_5 divides 3 and n5≡1(mod5)n_5 \equiv 1 \pmod 5. The divisors of 3 are 1 and 3, and 3≢13 \not\equiv 1. So n5=1n_5 = 1.

So GG has a normal subgroup PP of order 3 and a normal subgroup QQ of order 5. Their intersection is a subgroup of both, so its order divides gcd⁡(3,5)=1\gcd(3, 5) = 1: P∩Q={e}P \cap Q = \{e\}. For a∈Pa \in P, b∈Qb \in Q, the commutator aba−1b−1aba^{-1}b^{-1} lies in QQ (as aba−1∈Qaba^{-1} \in Q) and in PP (as ba−1b−1∈Pba^{-1}b^{-1} \in P), so it is ee and ab=baab = ba. Then abab has order lcm⁡(3,5)=15\operatorname{lcm}(3, 5) = 15, and G=⟨ab⟩≅Z/15ZG = \langle ab \rangle \cong \mathbb{Z}/15\mathbb{Z}.

The same argument proves: if p<qp < q are primes and p∤q−1p \nmid q - 1, every group of order pqpq is cyclic. (For p=2p = 2, q=3q = 3 it fails, correctly: S3S_3 has order 6 and 2∣3−12 \mid 3 - 1.)

Application 2 — no group of order 12 is simple

Let ∣G∣=12=22⋅3|G| = 12 = 2^2 \cdot 3. Then n3∣4n_3 \mid 4 with n3≡1(mod3)n_3 \equiv 1 \pmod 3, so n3∈{1,4}n_3 \in \{1, 4\}; and n2∣3n_2 \mid 3 with n2n_2 odd, so n2∈{1,3}n_2 \in \{1, 3\}.

  • If n3=1n_3 = 1, the Sylow 3-subgroup is normal and GG is not simple.
  • If n3=4n_3 = 4, the four subgroups of order 3 contain 4⋅2=84 \cdot 2 = 8 elements of order 3. That leaves 12−8=412 - 8 = 4 elements. A Sylow 2-subgroup has 4 elements, none of order 3, so it must be exactly these 4 elements. There is room for only one, so n2=1n_2 = 1 and it is normal.

Either way GG has a proper nontrivial normal subgroup. A4A_4 is the second case: n3=4n_3 = 4 and n2=1n_2 = 1, the normal subgroup being V4={e,(1 2)(3 4),(1 3)(2 4),(1 4)(2 3)}V_4 = \{e, (1\,2)(3\,4), (1\,3)(2\,4), (1\,4)(2\,3)\}.

Application 3 — order 20

If ∣G∣=20=22⋅5|G| = 20 = 2^2 \cdot 5, then n5∣4n_5 \mid 4 and n5≡1(mod5)n_5 \equiv 1 \pmod 5 force n5=1n_5 = 1. So every group of order 20 has a normal subgroup of order 5 and is not simple.

Trick: the recipe for "no simple group of order nn" is: list the allowed npn_p for each prime; if some prime forces np=1n_p = 1, you are done; if not, count elements of prime order and look for a contradiction, as for order 12.

Watch out: Sylow part 3 gives only candidates. n2∈{1,3}n_2 \in \{1, 3\} for order 12 does not mean both values occur in every group; A4A_4 has n2=1n_2 = 1, while the dihedral group of order 12 has n2=3n_2 = 3.


Chapter 5 — Finite abelian groups

For abelian groups the story is complete: they are classified by lists of numbers.

Theorem (fundamental theorem of finite abelian groups). Every finite abelian group AA is isomorphic to a direct product of cyclic groups of prime-power order,

A  ≅  Z/p1a1Z×⋯×Z/prarZ,A \;\cong\; \mathbb{Z}/p_1^{a_1}\mathbb{Z} \times \cdots \times \mathbb{Z}/p_r^{a_r}\mathbb{Z},

and the list of prime powers piaip_i^{a_i} (the elementary divisors) is unique up to order. Equivalently, A≅Z/d1Z×⋯×Z/dsZA \cong \mathbb{Z}/d_1\mathbb{Z} \times \cdots \times \mathbb{Z}/d_s\mathbb{Z} with d1∣d2∣⋯∣dsd_1 \mid d_2 \mid \cdots \mid d_s and d1>1d_1 > 1, and the invariant factors did_i are unique.

To go from elementary divisors to invariant factors, write the prime powers of each prime in decreasing order, then multiply across: the largest power of every prime gives dsd_s, the next largest gives ds−1d_{s-1}, and so on. This uses Z/mZ×Z/nZ≅Z/mnZ\mathbb{Z}/m\mathbb{Z} \times \mathbb{Z}/n\mathbb{Z} \cong \mathbb{Z}/mn\mathbb{Z} for coprime m,nm, n (Chapter 1).

Counting. The number of abelian groups of order p1e1⋯prerp_1^{e_1} \cdots p_r^{e_r} is the product of the partition numbers p(e1)⋯p(er)p(e_1) \cdots p(e_r), since each prime's part is chosen independently as a partition of its exponent.

Worked example: order 72. 72=23⋅3272 = 2^3 \cdot 3^2. The exponent 3 has three partitions (33, 2+12+1, 1+1+11+1+1) and the exponent 2 has two (22, 1+11+1), so there are 3⋅2=63 \cdot 2 = 6 abelian groups of order 72:

Elementary divisors Invariant factors
8,98, 9 Z/72\mathbb{Z}/72
8,3,38, 3, 3 Z/3×Z/24\mathbb{Z}/3 \times \mathbb{Z}/24
4,2,94, 2, 9 Z/2×Z/36\mathbb{Z}/2 \times \mathbb{Z}/36
4,2,3,34, 2, 3, 3 Z/6×Z/12\mathbb{Z}/6 \times \mathbb{Z}/12
2,2,2,92, 2, 2, 9 Z/2×Z/2×Z/18\mathbb{Z}/2 \times \mathbb{Z}/2 \times \mathbb{Z}/18
2,2,2,3,32, 2, 2, 3, 3 Z/2×Z/6×Z/6\mathbb{Z}/2 \times \mathbb{Z}/6 \times \mathbb{Z}/6

(Here Z/n\mathbb{Z}/n abbreviates Z/nZ\mathbb{Z}/n\mathbb{Z}.) Check one row: for 4,2,3,34, 2, 3, 3 the powers of 2 in decreasing order are 4,24, 2 and of 3 are 3,33, 3; multiplying across gives 4⋅3=124\cdot 3 = 12 and 2⋅3=62 \cdot 3 = 6, so Z/6×Z/12\mathbb{Z}/6 \times \mathbb{Z}/12, and indeed 6∣126 \mid 12.

Common pitfall: Z/2×Z/36\mathbb{Z}/2 \times \mathbb{Z}/36 and Z/4×Z/18\mathbb{Z}/4 \times \mathbb{Z}/18 are the same group (both have elementary divisors 4,2,94, 2, 9). Always reduce to one normal form before comparing.

The same theorem, stated for finitely generated modules over a principal ideal domain, is the structure theory behind the Jordan normal form; it is planned as a later post.


Practice Exercises

1. How many homomorphisms are there from Z/20Z\mathbb{Z}/20\mathbb{Z} to Z/8Z\mathbb{Z}/8\mathbb{Z}? List the possible images of 11.

Solution

A homomorphism is fixed by a=φ(1)a = \varphi(1), subject to 20a≡0(mod8)20a \equiv 0 \pmod 8, i.e. 4a≡0(mod8)4a \equiv 0 \pmod 8, i.e. aa even. So a∈{0,2,4,6}a \in \{0, 2, 4, 6\}: four homomorphisms, which is gcd⁡(20,8)=4\gcd(20, 8) = 4.

2. Show that Z/8Z\mathbb{Z}/8\mathbb{Z}, Z/4Z×Z/2Z\mathbb{Z}/4\mathbb{Z} \times \mathbb{Z}/2\mathbb{Z} and (Z/2Z)3(\mathbb{Z}/2\mathbb{Z})^3 are pairwise non-isomorphic.

Solution

Compare the largest element order. Z/8Z\mathbb{Z}/8\mathbb{Z} has an element of order 8. In Z/4Z×Z/2Z\mathbb{Z}/4\mathbb{Z} \times \mathbb{Z}/2\mathbb{Z} the order of (a,b)(a, b) is lcm⁡\operatorname{lcm} of the orders, at most lcm⁡(4,2)=4\operatorname{lcm}(4, 2) = 4, and (1,0)(1, 0) has order 4. In (Z/2Z)3(\mathbb{Z}/2\mathbb{Z})^3 every non-identity element has order 2. The maximal orders 8, 4, 2 differ, so no two are isomorphic. (These are the three abelian groups of order 8, from the partitions 33, 2+12+1, 1+1+11+1+1.)

3. S4S_4 acts on the set of its three "pairings" {{1,2},{3,4}}\{\{1,2\},\{3,4\}\}, {{1,3},{2,4}}\{\{1,3\},\{2,4\}\}, {{1,4},{2,3}}\{\{1,4\},\{2,3\}\}. Use orbit–stabilizer to find the order of the stabilizer of one pairing, and find the kernel of the action.

Solution

The action is transitive (for instance (2 3)(2\,3) swaps the first two pairings, (2 4)(2\,4) swaps the first and third), so the orbit has size 3 and the stabilizer has order 24/3=824/3 = 8. It is a Sylow 2-subgroup (a copy of D4D_4). The kernel is the set of permutations fixing all three pairings: V4={e,(1 2)(3 4),(1 3)(2 4),(1 4)(2 3)}V_4 = \{e, (1\,2)(3\,4), (1\,3)(2\,4), (1\,4)(2\,3)\}. By the First Isomorphism Theorem, S4/V4S_4/V_4 is isomorphic to the image in S3S_3, which has order 24/4=624/4 = 6, so S4/V4≅S3S_4/V_4 \cong S_3.

4. How many ways are there to colour the 4 vertices of a square with 3 colours, up to rotation? Up to all 8 symmetries of D4D_4?

Solution

Rotations by 0∘,90∘,180∘,270∘0^\circ, 90^\circ, 180^\circ, 270^\circ have 4,1,2,14, 1, 2, 1 cycles on the vertices: (34+3+32+3)/4=(81+3+9+3)/4=96/4=24(3^4 + 3 + 3^2 + 3)/4 = (81 + 3 + 9 + 3)/4 = 96/4 = 24.

For D4D_4 add the reflections. The two diagonal reflections fix 2 vertices and swap 2: 3 cycles each. The two reflections through edge midpoints swap two pairs: 2 cycles each. Total (96+2⋅27+2⋅9)/8=(96+54+18)/8=168/8=21(96 + 2\cdot 27 + 2 \cdot 9)/8 = (96 + 54 + 18)/8 = 168/8 = 21.

5. Prove that no group of order 35 is simple, and in fact that every group of order 35 is cyclic.

Solution

35=5⋅735 = 5 \cdot 7. n5∣7n_5 \mid 7 and n5≡1(mod5)n_5 \equiv 1 \pmod 5: the divisors are 1, 7 and 7≡27 \equiv 2, so n5=1n_5 = 1. n7∣5n_7 \mid 5 and n7≡1(mod7)n_7 \equiv 1 \pmod 7: the divisors are 1, 5 and 5≢15 \not\equiv 1, so n7=1n_7 = 1. Both Sylow subgroups are normal, meet trivially, and commute elementwise, exactly as for order 15, so G≅Z/5×Z/7≅Z/35G \cong \mathbb{Z}/5 \times \mathbb{Z}/7 \cong \mathbb{Z}/35. (This is the "p∤q−1p \nmid q - 1" criterion with 5∤65 \nmid 6.)

6. List the abelian groups of order 36 in both normal forms.

Solution

36=22⋅3236 = 2^2 \cdot 3^2, and p(2)⋅p(2)=2⋅2=4p(2) \cdot p(2) = 2 \cdot 2 = 4 groups:

  • 4,94, 9 → Z/36\mathbb{Z}/36
  • 4,3,34, 3, 3 → Z/3×Z/12\mathbb{Z}/3 \times \mathbb{Z}/12
  • 2,2,92, 2, 9 → Z/2×Z/18\mathbb{Z}/2 \times \mathbb{Z}/18
  • 2,2,3,32, 2, 3, 3 → Z/6×Z/6\mathbb{Z}/6 \times \mathbb{Z}/6

7. Let GG be a finite group and HH a subgroup of index nn. Using the action of GG on the cosets G/HG/H, show that GG has a normal subgroup K⊆HK \subseteq H with [G:K][G:K] dividing n!n!. Deduce that a simple group of order 60 has no subgroup of index 2, 3 or 4.

Solution

The action gives a homomorphism ρ:G→Sym⁡(G/H)≅Sn\rho : G \to \operatorname{Sym}(G/H) \cong S_n. Its kernel KK is normal; anything in KK fixes the coset HH, so K⊆HK \subseteq H. By the First Isomorphism Theorem G/KG/K embeds in SnS_n, so [G:K][G:K] divides n!n! by Lagrange.

If GG is simple of order 60 and [G:H]=n∈{2,3,4}[G:H] = n \in \{2, 3, 4\}, then KK is {e}\{e\} or GG. K=GK = G is impossible since K⊆H≠GK \subseteq H \ne G. So K={e}K = \{e\} and 6060 divides n!≤24n! \le 24, a contradiction.


Cheat-sheet

Idea Statement
Homomorphism φ(ab)=φ(a)φ(b)\varphi(ab) = \varphi(a)\varphi(b); then φ(e)=e\varphi(e) = e, φ(g−1)=φ(g)−1\varphi(g^{-1}) = \varphi(g)^{-1}
Kernel always a normal subgroup; trivial exactly when φ\varphi is injective
Maps between cyclic groups gcd⁡(n,m)\gcd(n, m) homomorphisms Z/n→Z/m\mathbb{Z}/n \to \mathbb{Z}/m, fixed by the image of 1
Automorphisms Aut⁡(Z/n)≅(Z/n)×\operatorname{Aut}(\mathbb{Z}/n) \cong (\mathbb{Z}/n)^\times, order ϕ(n)\phi(n); Inn⁡(G)≅G/Z(G)\operatorname{Inn}(G) \cong G/Z(G)
First Isomorphism Theorem G/ker⁡φ≅im⁡φG/\ker\varphi \cong \operatorname{im}\varphi
Second HN/N≅H/(H∩N)HN/N \cong H/(H \cap N) when NN is normal
Third (G/N)/(K/N)≅G/K(G/N)/(K/N) \cong G/K
Cayley GG embeds in Sym⁡(G)\operatorname{Sym}(G); order nn embeds in SnS_n
Orbit–stabilizer orbit size times stabilizer size =∣G∣= \lvert G\rvert
Class equation ∣G∣=∣Z(G)∣+∑[G:CG(xi)]\lvert G\rvert = \lvert Z(G)\rvert + \sum [G : C_G(x_i)]
pp-groups nontrivial center; order p2p^2 implies abelian
Burnside orbits = average number of fixed points; colourings fixed by gg: kc(g)k^{c(g)}
Cauchy prime pp divides the order, so an element of order pp exists
Sylow exist; all conjugate; np≡1 mod pn_p \equiv 1 \bmod p and np∣mn_p \mid m
Normal Sylow np=1n_p = 1 exactly when the Sylow pp-subgroup is normal
Order pqpq, p<qp < q cyclic if p∤q−1p \nmid q - 1
Finite abelian groups product of cyclic prime-power groups; count = product of partition numbers

What's next

Post 04 moves from one operation to two. A ring has addition and multiplication tied by the distributive law; the integers, the integers modulo nn, polynomials and matrices are the models. Every idea of this post has a ring version: ring homomorphisms and their kernels (now called ideals), quotient rings, and a First Isomorphism Theorem with the same proof. Prime and maximal ideals then explain why Z/pZ\mathbb{Z}/p\mathbb{Z} is a field, and the same construction applied to polynomials builds C\mathbb{C} and the finite fields that post 05 needs.


References

  • D. S. Dummit and R. M. Foote, Abstract Algebra, 3rd edition, Wiley, 2004 — Chapters 3.3 (isomorphism theorems), 4 (group actions, class equation, Sylow) and 5.2 (finite abelian groups).
  • M. Artin, Algebra, 2nd edition, Pearson, 2011 — Chapters 6 (symmetry, group actions, counting) and 7 (class equation, Sylow theorems).
  • J. A. Gallian, Contemporary Abstract Algebra, 10th edition, CRC Press, 2021 — homomorphisms, Burnside's lemma worked examples, Sylow theorems.
  • J. J. Rotman, An Introduction to the Theory of Groups, 4th edition, Springer GTM 148, 1995 — full proofs of Sylow and the basis theorem.
  • MIT OpenCourseWare, 18.701 Algebra I (Fall 2010) and 18.702 Algebra II (Spring 2011) — lecture notes and problem sets.
  • K. Conrad, Expository papers ("blurbs") — short notes on group actions, Sylow theorems and finite abelian groups.
  • J. H. McKay, "Another proof of Cauchy's group theorem", American Mathematical Monthly 66 (1959), p. 119.

Note: this post is in progress. Planned additions: full proofs of the three Sylow theorems (via the action on subsets of size pkp^k and on Sylow subgroups by conjugation), semidirect products and the classification of groups of order 12 and of order pqpq, and the Pólya enumeration theorem as a refinement of Burnside's lemma.