← 2026 Paper 2

UPSC 2026 Maths Optional Paper 2 Q2a — Step-by-Step Solution

15 marks · Section A

Groups: definition, axioms, examples · Algebra · asked 3× in 14 yrs · Read the full method →

Question

Let GG be a finite group generated by two elements uu and vv of order 2 each. Show that GG is isomorphic to the dihedral group of order 2n2n, where O(uv)=nO(uv) = n.

Technique

The recognisable trigger is two involutions: the moment a group is generated by two elements of order 22, put r=uvr=uv and s=us=u — then v=srv=sr, so {r,s}\{r,s\} still generates, and the pair (r,s)(r,s) satisfies the defining dihedral relations rn=s2=er^{n}=s^{2}=e, srs−1=r−1srs^{-1}=r^{-1}. From those relations alone every element collapses to the normal form rkr^{k} or rksr^{k}s, giving ∣G∣≤2n|G|\le 2n; the whole marking weight then sits on proving those 2n2n elements are distinct, which reduces to the single claim s∉⟨r⟩s\notin\langle r\rangle. Do not merely assert the isomorphism — exhibit the bijection and check it multiplies correctly.

Solution

Step 0 — Fix the model of the dihedral group.

Write Zn={0,1,…,n−1}\mathbb{Z}_n=\{0,1,\dots,n-1\} with addition mod nn. Define the dihedral group of order 2n2n as the semidirect product Dn=Zn⋊Z2D_n=\mathbb{Z}_n\rtimes\mathbb{Z}_2: as a set

Dn={(k,ε)  :  k∈Zn, ε∈{0,1}},∣Dn∣=2n,D_n=\bigl\{(k,\varepsilon)\;:\;k\in\mathbb{Z}_n,\ \varepsilon\in\{0,1\}\bigr\},\qquad |D_n|=2n,

with product

(k,ε)(l,δ)=(k+(−1)εl (mod n), ε+δ (mod 2)).(k,\varepsilon)(l,\delta)=\bigl(k+(-1)^{\varepsilon}l\ (\mathrm{mod}\ n),\ \varepsilon+\delta\ (\mathrm{mod}\ 2)\bigr).

Put a=(1,0)a=(1,0) and b=(0,1)b=(0,1). A direct computation gives

an=b2=e,bab−1=a−1,Dn={ak, akb:0≤k≤n−1}  (all 2n distinct),a^{n}=b^{2}=e,\qquad bab^{-1}=a^{-1},\qquad D_n=\{a^{k},\,a^{k}b:0\le k\le n-1\}\ \ (\text{all }2n\text{ distinct}),

so Dn=⟨a,b∣an=b2=e, bab−1=a−1⟩D_n=\langle a,b\mid a^{n}=b^{2}=e,\ bab^{-1}=a^{-1}\rangle and, for n≥3n\ge3, DnD_n is exactly the symmetry group of the regular nn-gon (nn rotations aka^k and nn reflections akba^kb).

Why state this first: “the dihedral group of order 2n2n” must be a group we already know has order 2n2n, otherwise the counting argument below is circular.

Step 1 — Choose the right generators.

We are given u2=v2=eu^{2}=v^{2}=e with u≠eu\neq e, v≠ev\neq e, and G=⟨u,v⟩G=\langle u,v\rangle finite. Set

r=uv,s=u.\boxed{r=uv,\qquad s=u.}

Since GG is finite, rr has a finite order; by hypothesis

n=O(r)=O(uv).n=O(r)=O(uv).

Step 2 — {r,s}\{r,s\} generates GG.

Clearly r,s∈Gr,s\in G, so ⟨r,s⟩⊆G\langle r,s\rangle\subseteq G. Conversely u=su=s and, since u2=eu^{2}=e,

sr=u(uv)=u2v=v.sr=u(uv)=u^{2}v=v .

So both uu and vv lie in ⟨r,s⟩\langle r,s\rangle, whence G=⟨u,v⟩⊆⟨r,s⟩G=\langle u,v\rangle\subseteq\langle r,s\rangle. Therefore

G=⟨r,s⟩.G=\langle r,s\rangle .

Step 3 — Verify the three dihedral relations.

  1. rn=er^{n}=e by the definition of n=O(r)n=O(r), and no smaller positive power of rr is ee.
  2. s2=u2=es^{2}=u^{2}=e, and s=u≠es=u\neq e, so O(s)=2O(s)=2.
  3. Using u−1=uu^{-1}=u and v−1=vv^{-1}=v,
srs−1=u(uv)u−1=(u2)vu=vu,r−1=(uv)−1=v−1u−1=vu.srs^{-1}=u(uv)u^{-1}=(u^{2})vu=vu,\qquad r^{-1}=(uv)^{-1}=v^{-1}u^{-1}=vu .

Hence

srs−1=r−1.(∗)srs^{-1}=r^{-1}. \tag{$\ast$}

Iterating (∗)(\ast): srks−1=(srs−1)k=r−ks r^{k}s^{-1}=(srs^{-1})^{k}=r^{-k}, and since s−1=ss^{-1}=s,

s rk=r−k sfor every k∈Z.(∗∗)s\,r^{k}=r^{-k}\,s\qquad\text{for every }k\in\mathbb{Z}. \tag{$\ast\ast$}

Relation (∗∗)(\ast\ast) is the working form: it lets every ss be pushed to the right past any power of rr.

Step 4 — Every element of GG has the normal form rkr^{k} or rksr^{k}s.

Let

H={ rk  :  0≤k≤n−1 } ∪ { rks  :  0≤k≤n−1 }.H=\{\,r^{k}\;:\;0\le k\le n-1\,\}\ \cup\ \{\,r^{k}s\;:\;0\le k\le n-1\,\}.

We show HH is a subgroup. All indices are read mod nn, legitimate because rn=er^{n}=e. Using (∗∗)(\ast\ast) and s2=es^{2}=e:

ri⋅rj=ri+j,ri⋅(rjs)=ri+js,r^{i}\cdot r^{j}=r^{i+j},\qquad r^{i}\cdot(r^{j}s)=r^{i+j}s, (ris)⋅rj=ri(s rj)=rir−js=ri−js,(ris)(rjs)=ri(s rj)s=ri−js2=ri−j.(4.1)(r^{i}s)\cdot r^{j}=r^{i}(s\,r^{j})=r^{i}r^{-j}s=r^{i-j}s,\qquad (r^{i}s)(r^{j}s)=r^{i}(s\,r^{j})s=r^{i-j}s^{2}=r^{i-j}. \tag{4.1}

So HH is closed under multiplication. Also e=r0∈He=r^{0}\in H; and HH is closed under inverses, since (rk)−1=rn−k∈H(r^{k})^{-1}=r^{n-k}\in H and

(rks)2=rk−k=e ⇒ (rks)−1=rks∈H(r^{k}s)^{2}=r^{k-k}=e\ \Rightarrow\ (r^{k}s)^{-1}=r^{k}s\in H

(so incidentally every element rksr^{k}s is an involution — the “reflections”).

Thus H≤GH\le G. Moreover u=s=r0s∈Hu=s=r^{0}s\in H, and by Step 2 together with (∗∗)(\ast\ast), v=sr=r−1s∈Hv=sr=r^{-1}s\in H. Since G=⟨u,v⟩G=\langle u,v\rangle is the smallest subgroup containing uu and vv, we get G⊆HG\subseteq H; and H⊆GH\subseteq G is clear. Hence

G={e,r,r2,…,rn−1}  ∪  {s,rs,r2s,…,rn−1s},so ∣G∣≤2n.(4.2)G=\{e,r,r^{2},\dots,r^{n-1}\}\;\cup\;\{s,rs,r^{2}s,\dots,r^{n-1}s\},\qquad\text{so } |G|\le 2n. \tag{4.2}

Step 5 — The key lemma: s∉⟨r⟩s\notin\langle r\rangle.

This is where the marks are. Suppose, for contradiction, that s∈⟨r⟩s\in\langle r\rangle. Then G=⟨r,s⟩=⟨r⟩G=\langle r,s\rangle=\langle r\rangle is cyclic, hence abelian. Being abelian, uu and vv commute, so

r2=(uv)(uv)=u(vu)v=u(uv)v=u2v2=e,r^{2}=(uv)(uv)=u(vu)v=u(uv)v=u^{2}v^{2}=e,

and therefore n=O(r)∈{1,2}n=O(r)\in\{1,2\}.

Both cases are impossible. Hence

s∉⟨r⟩.(5.1)s\notin\langle r\rangle. \tag{5.1}

Step 6 — The 2n2n listed elements are distinct, so ∣G∣=2n|G|=2n.

Three checks, and all three are needed:

  1. Rotations are distinct. If ri=rjr^{i}=r^{j} with 0≤i<j≤n−10\le i<j\le n-1 then r j−i=er^{\,j-i}=e with 0<j−i<n0<j-i<n, contradicting O(r)=nO(r)=n.
  2. Reflections are distinct. If ris=rjsr^{i}s=r^{j}s then right-multiplying by s−1s^{-1} gives ri=rjr^{i}=r^{j}, so i=ji=j by (1).
  3. No rotation equals a reflection. If ri=rjsr^{i}=r^{j}s for some i,ji,j, then s=r i−j∈⟨r⟩s=r^{\,i-j}\in\langle r\rangle, contradicting (5.1).

So the list in (4.2) has exactly 2n2n distinct entries and

∣G∣=2n.(6.1)|G|=2n. \tag{6.1}

(Structural aside, worth one line: ⟨r⟩\langle r\rangle has index 22 in GG by (6.1), hence ⟨r⟩⊴G\langle r\rangle\trianglelefteq G, and (∗)(\ast) says ss acts on it by inversion — i.e. G=⟨r⟩⋊⟨s⟩G=\langle r\rangle\rtimes\langle s\rangle, which is precisely the shape of DnD_n in Step 0.)

Step 7 — Construct the isomorphism explicitly.

By Step 0 the 2n2n elements ak,akba^{k},a^{k}b (0≤k≤n−1)(0\le k\le n-1) of DnD_n are distinct and exhaust DnD_n; by Step 6 the 2n2n elements rk,rksr^{k},r^{k}s are distinct and exhaust GG. So the assignment

φ:G⟶Dn,φ(rk)=ak,φ(rks)=akb(0≤k≤n−1)\varphi:G\longrightarrow D_n,\qquad \varphi(r^{k})=a^{k},\qquad \varphi(r^{k}s)=a^{k}b\qquad(0\le k\le n-1)

is well defined (each element of GG has exactly one normal form) and is a bijection (it matches the two lists term by term).

φ\varphi is a homomorphism: in DnD_n the relations an=b2=ea^{n}=b^{2}=e, baj=a−jbb a^{j}=a^{-j}b give, exactly as in (4.1),

aiaj=ai+j,ai(ajb)=ai+jb,(aib)aj=ai−jb,(aib)(ajb)=ai−j.a^{i}a^{j}=a^{i+j},\quad a^{i}(a^{j}b)=a^{i+j}b,\quad (a^{i}b)a^{j}=a^{i-j}b,\quad (a^{i}b)(a^{j}b)=a^{i-j}.

Comparing with (4.1) case by case:

φ(ri⋅rj)=φ(ri+j)=ai+j=φ(ri)φ(rj),\varphi(r^{i}\cdot r^{j})=\varphi(r^{i+j})=a^{i+j}=\varphi(r^{i})\varphi(r^{j}), φ(ri⋅rjs)=φ(ri+js)=ai+jb=ai(ajb)=φ(ri)φ(rjs),\varphi\bigl(r^{i}\cdot r^{j}s\bigr)=\varphi(r^{i+j}s)=a^{i+j}b=a^{i}(a^{j}b)=\varphi(r^{i})\varphi(r^{j}s), φ((ris)⋅rj)=φ(ri−js)=ai−jb=(aib)aj=φ(ris)φ(rj),\varphi\bigl((r^{i}s)\cdot r^{j}\bigr)=\varphi(r^{i-j}s)=a^{i-j}b=(a^{i}b)a^{j}=\varphi(r^{i}s)\varphi(r^{j}), φ((ris)(rjs))=φ(ri−j)=ai−j=(aib)(ajb)=φ(ris)φ(rjs).\varphi\bigl((r^{i}s)(r^{j}s)\bigr)=\varphi(r^{i-j})=a^{i-j}=(a^{i}b)(a^{j}b)=\varphi(r^{i}s)\varphi(r^{j}s).

These four cases cover every product in GG. Hence φ\varphi is a bijective homomorphism, i.e.

G≅Dn,n=O(uv),∣G∣=2n.■G\cong D_n,\qquad n=O(uv),\qquad |G|=2n.\qquad\blacksquare

(Alternative one-line finish, if von Dyck’s theorem is available: r,s∈Gr,s\in G satisfy the defining relations of Dn=⟨a,b∣an=b2=e, bab−1=a−1⟩D_n=\langle a,b\mid a^{n}=b^{2}=e,\,bab^{-1}=a^{-1}\rangle, so there is a homomorphism Dn→GD_n\to G with a↦ra\mapsto r, b↦sb\mapsto s; it is onto because ⟨r,s⟩=G\langle r,s\rangle=G (Step 2), and ∣Dn∣=2n=∣G∣|D_n|=2n=|G| by Step 6, so it is injective as well. Steps 5–6 are still needed — they supply ∣G∣=2n|G|=2n — so nothing is saved except the four-case check.)

Step 8 — The small cases (honesty at the boundary).

The proof above is uniform in n≥1n\ge1, but it is worth recording what the conclusion means at the two degenerate values, since “dihedral” is usually pictured as a polygon group:

In particular GG is abelian   ⟺  n≤2\iff n\le2.

Step 9 — Where finiteness was used.

Finiteness of GG was used exactly once — to guarantee n=O(uv)<∞n=O(uv)<\infty in Step 1. It is not an extra assumption beyond that: by (6.1), ∣G∣=2 O(uv)|G|=2\,O(uv), so GG is finite iff uvuv has finite order. If O(uv)=∞O(uv)=\infty the same Steps 2–4 give G≅D∞=⟨a,b∣b2=e, bab−1=a−1⟩G\cong D_\infty=\langle a,b\mid b^{2}=e,\ bab^{-1}=a^{-1}\rangle, the infinite dihedral group.

Answer

  With r=uv, s=u:rn=s2=e,srs−1=r−1,G=⟨r,s⟩,G={rk}k=0n−1∪{rks}k=0n−1  (2n distinct elements),∣G∣=2n,φ(rk)=ak,  φ(rks)=akb  is an isomorphism G→ ∼ Dn,n=O(uv).  \boxed{\; \begin{aligned} &\text{With }r=uv,\ s=u:\quad r^{n}=s^{2}=e,\quad srs^{-1}=r^{-1},\quad G=\langle r,s\rangle,\\[2pt] &G=\{r^{k}\}_{k=0}^{n-1}\cup\{r^{k}s\}_{k=0}^{n-1}\ \ (2n\ \text{distinct elements}),\qquad |G|=2n,\\[2pt] &\varphi(r^{k})=a^{k},\ \ \varphi(r^{k}s)=a^{k}b\ \ \text{is an isomorphism } G\xrightarrow{\ \sim\ }D_n,\qquad n=O(uv). \end{aligned}\;}
We post more of this — worked solutions, CSAT trap breakdowns, guide chapters — a few times a week on Telegram. Free, no sign-in. Join

This solution is part of the Maths Coverage Map — 14 years, mapped. Get the take-away PDF free.