Skip to content

Worked Examples

Problem. Show that S4S_4 has no normal subgroup of order 8.

Solution. Suppose N⊴S4N \trianglelefteq S_4 with ∣N∣=8|N| = 8. By Lagrange, [S4:N]=24/8=3[S_4 : N] = 24/8 = 3. The action of S4S_4 on the cosets of NN gives a homomorphism ϕ:S4→S3\phi : S_4 \to S_3. Since NN is normal, ker⁡(ϕ)⊆N\ker(\phi) \subseteq N So ∣ker⁡(ϕ)∣|\ker(\phi)| divides 88. Since im(ϕ)⊆S3\mathrm{im}(\phi) \subseteq S_3 has order dividing 66, ∣ker⁡(ϕ)∣|\ker(\phi)| divides 24/6=424/6 = 4. But ∣ker⁡(ϕ)∣|\ker(\phi)| divides 88 So ∣ker⁡(ϕ)∣|\ker(\phi)| divides gcd⁡(8,4)=4\gcd(8, 4) = 4. Since ker⁡(ϕ)⊆N\ker(\phi) \subseteq N and ∣N∣=8|N| = 8 And ∣ker⁡(ϕ)∣|\ker(\phi)| divides 44 This is possible. However, we need ker⁡(ϕ)=N\ker(\phi) = N (since NN is the kernel of the action on cosets). Then ∣ker⁡(ϕ)∣=8|\ker(\phi)| = 8 Contradicting ∣ker⁡(ϕ)∣≤4|\ker(\phi)| \leq 4. So no such NN exists. ■\blacksquare

Problem. Show that Z[2]\mathbb{Z}[\sqrt{2}] is a Euclidean domain.

Solution. Define δ(a+b2)=∣a2−2b2∣\delta(a + b\sqrt{2}) = |a^2 - 2b^2|. For α,β≠0\alpha, \beta \neq 0 in Z[2]\mathbb{Z}[\sqrt{2}] Compute α/β∈Q(2)\alpha/\beta \in \mathbb{Q}(\sqrt{2}). Choose q∈Z[2]q \in \mathbb{Z}[\sqrt{2}] With ∣α/β−q∣<1|\alpha/\beta - q| \lt 1 in both coordinates (round to nearest integers). Then α=qβ+r\alpha = q\beta + r where r=α−qβr = \alpha - q\beta. We have δ(r)=∣N(r)∣=∣N(α−qβ)∣=∣N(β)∣⋅∣N(α/β−q)∣\delta(r) = |N(r)| = |N(\alpha - q\beta)| = |N(\beta)| \cdot |N(\alpha/\beta - q)|. Since ∣N(α/β−q)∣=∣a2−2b2∣<1|N(\alpha/\beta - q)| = |a^2 - 2b^2| \lt 1 For the coordinate differences a,ba, b (each less than 11 in absolute value), we get δ(r)<δ(β)\delta(r) \lt \delta(\beta). ■\blacksquare

Problem. Determine the Galois group of x4−5x^4 - 5 over Q\mathbb{Q}.

Solution. The roots are ±54\pm\sqrt[4]{5} and ±i54\pm i\sqrt[4]{5}. The splitting field is E=Q(54,i)E = \mathbb{Q}(\sqrt[4]{5}, i). We have [Q(54):Q]=4[\mathbb{Q}(\sqrt[4]{5}) : \mathbb{Q}] = 4 (since x4−5x^4 - 5 is irreducible by Eisenstein with p=5p = 5), and [E:Q(54)]=2[E : \mathbb{Q}(\sqrt[4]{5})] = 2 (since i∉Q(54)⊂Ri \notin \mathbb{Q}(\sqrt[4]{5}) \subset \mathbb{R}). Thus [E:Q]=8[E : \mathbb{Q}] = 8.

The Galois group is generated by:

  • σ:54↦i54, i↦i\sigma : \sqrt[4]{5} \mapsto i\sqrt[4]{5},\ i \mapsto i (order 4)
  • τ:54↦54, i↦−i\tau : \sqrt[4]{5} \mapsto \sqrt[4]{5},\ i \mapsto -i (order 2)

With τστ−1=σ−1\tau\sigma\tau^{-1} = \sigma^{-1} We get Gal(E/Q)≅D8\mathrm{Gal}(E/\mathbb{Q}) \cong D_8 (the dihedral Group of order 8). ■\blacksquare

Worked examples are where abstract theory meets concrete calculation. Showing that S4 has no normal subgroup of order 8 requires understanding how group actions on cosets create homomorphisms and why counting constraints make certain structures impossible. Determining whether a ring is a Euclidean domain tests whether division with remainder is possible. Computing Galois groups reveals the hidden symmetry of polynomial roots, connecting the algebraic structure of a field extension to the geometric arrangement of roots. These examples demonstrate that algebra is not about memorising definitions but about building intuition for how structures fit together.

Mistake 1: Assuming every normal subgroup is unique for a given order A group can have multiple normal subgroups of the same order, or none at all. Lagrange’s theorem only constrains the possible orders of subgroups, not their existence or uniqueness. Always verify normality directly rather than assuming it from the order alone.

Mistake 2: Confusing the Galois group of a polynomial with the Galois group of its splitting field The Galois group acts on the roots of the polynomial, but it is defined as the automorphism group of the splitting field. For reducible polynomials, the Galois group may be a proper subgroup of the permutation group on roots. Always identify the splitting field first before computing its automorphisms.

Mistake 3: Assuming irreducibility over Q implies irreducibility over Z A polynomial with rational coefficients that is irreducible over Q may still factor over Z after clearing denominators. Gauss’s lemma guarantees that if a polynomial factors over Q, it factors over Z with the same degree factors, but the converse requires checking primitivity of the coefficients.

Problem. Let GG be a finite group and H⊴GH \trianglelefteq G with gcd⁡(∣H∣,[G:H])=1\gcd(|H|, [G:H]) = 1. Show that HH has a complement in GG: there exists K≤GK \leq G with HK=GHK = G and H∩K={e}H \cap K = \{e\}.

Solution

Solution. This result is known as the Schur-Zassenhaus theorem. We prove a special case When HH is abelian. Let ∣H∣=m|H| = m and [G:H]=n[G : H] = n with gcd⁡(m,n)=1\gcd(m, n) = 1.

Consider the action of GG on HH by conjugation. Since HH is abelian and normal, GG acts by Automorphisms on HH. The group of automorphisms of HH, Aut(H)\mathrm{Aut}(H) Has order dividing ∣H∣!|H|! But we need a more refined argument.

Here is a cleaner approach for the case when one factor is cyclic. Let G/H=⟨gH⟩G/H = \langle gH \rangle be cyclic (which always holds when [G:H][G : H] is the smallest prime dividing ∣G∣|G|). Pick a representative gg with gn∈Hg^n \in H. We need to find k∈Gk \in G with kn=ek^n = e and kH=gHkH = gH.

Since gcd⁡(m,n)=1\gcd(m, n) = 1 There exist a,b∈Za, b \in \mathbb{Z} with am+bn=1am + bn = 1. Let x=gn∈Hx = g^n \in H. Then xa=x1−bn=x⋅(xn)−b=x⋅x−bnx^a = x^{1 - bn} = x \cdot (x^n)^{-b} = x \cdot x^{-bn}… Actually, the general …/1-number-and-algebra/3_proof-and-logic requires Cohomology and is beyond our current scope. The key takeaway: when gcd⁡(∣H∣,[G:H])=1\gcd(|H|, [G:H]) = 1 A complement exists (Schur-Zassenhaus theorem). ■\blacksquare

Problem. Show that if GG is a group of order pqrpqr where p<q<rp \lt q \lt r are primes, then GG is not simple.

Solution

Solution. Consider the Sylow rr-subgroups. nr≡1(modr)n_r \equiv 1 \pmod{r} and nrn_r divides pqpq. Since r>q>pr > q > p We have r>pqr > pq is possible but nr=pq+1,2pq+1,…n_r = pq + 1, 2pq + 1, \ldots would need To divide pqpq. The only divisors of pqpq are 1,p,q,pq1, p, q, pq. If nr=1n_r = 1 We are done (Sylow rr-subgroup is normal).

If nr≠1n_r \neq 1 Then nr≥r+1>q+1>p+1n_r \geq r + 1 > q + 1 > p + 1. Since nrn_r divides pqpq We must have nr=pqn_r = pq. This means there are pq(r−1)pq(r - 1) non-identity elements in Sylow rr-subgroups.

Similarly, nq≡1(modq)n_q \equiv 1 \pmod{q} and nqn_q divides prpr. If nq=prn_q = pr There are pr(q−1)pr(q - 1) non-identity elements in Sylow qq-subgroups.

If both nr=pqn_r = pq and nq=prn_q = pr The total number of non-identity elements is at least pq(r−1)+pr(q−1)=pqr−pq+pqr−pr=2pqr−p(q+r)pq(r-1) + pr(q-1) = pqr - pq + pqr - pr = 2pqr - p(q+r) Which exceeds pqr−1pqr - 1 for most values. So at least one of nr=1n_r = 1 or nq=1n_q = 1 must hold, giving a normal Sylow subgroup. ■\blacksquare

Problem. Prove that there are exactly two groups of order 1010 up to isomorphism.

Solution

Solution. ∣G∣=10=2⋅5|G| = 10 = 2 \cdot 5. By Sylow’s third theorem: n5≡1(mod5)n_5 \equiv 1 \pmod{5} and n5n_5 Divides 22 So n5=1n_5 = 1. The Sylow 55-subgroup P=⟨a⟩≅Z/5ZP = \langle a \rangle \cong \mathbb{Z}/5\mathbb{Z} is normal.

n2≡1(mod2)n_2 \equiv 1 \pmod{2} and n2n_2 divides 55 So n2=1n_2 = 1 or 55.

Let bb be an element of order 22 (exists by Cauchy’s theorem). Since P⊴GP \trianglelefteq G bab−1∈Pbab^{-1} \in P. So bab−1=akbab^{-1} = a^k for some k∈{0,1,2,3,4}k \in \{0, 1, 2, 3, 4\}. Applying conjugation twice: b2ab−2=ak2b^2ab^{-2} = a^{k^2} I.e., a=ak2a = a^{k^2} So k2≡1(mod5)k^2 \equiv 1 \pmod{5} Giving k≡1k \equiv 1 or k≡4(mod5)k \equiv 4 \pmod{5}.

Case k=1k = 1: bab−1=abab^{-1} = a So aa and bb commute. G≅Z/5Z×Z/2Z≅Z/10ZG \cong \mathbb{Z}/5\mathbb{Z} \times \mathbb{Z}/2\mathbb{Z} \cong \mathbb{Z}/10\mathbb{Z}.

Case k=4k = 4: bab−1=a4=a−1bab^{-1} = a^4 = a^{-1} So ba=a−1bba = a^{-1}b. This gives the dihedral group D5D_5.

These are the only two possibilities, so there are exactly two groups of order 1010. ■\blacksquare

  • Group Actions — The orbit-stabilizer theorem and Burnside’s lemma used in the worked examples are developed in this chapter.
  • The Sylow Theorems — Sylow theory is the primary tool for the group classification arguments in several worked examples.
  • Polynomial Rings — The ring-theoretic examples rely on irreducibility criteria and Euclidean division developed in this chapter.
  • Euclidean Domains, PIDs, and UFDs — The chain of implications between ring types underpins the factorisation arguments in the worked examples.