Skip to content
Part IA Michaelmas, Lent Term

Examinable Proofs Reference

This document consolidates all proofs you should be able to reproduce under exam conditions, organised by topic.


Part I: Proofs

1. Irrationality of 2\sqrt{2}

Theorem: 2\sqrt{2} is irrational.

Proof:

Suppose for contradiction 2=pq\sqrt{2} = \frac{p}{q} where gcd(p,q)=1\gcd(p, q) = 1.

Squaring: 2=p2q22 = \frac{p^2}{q^2}, so p2=2q2p^2 = 2q^2.

Thus p2p^2 is even, so pp is even. Write p=2kp = 2k.

Substituting: 4k2=2q24k^2 = 2q^2, so q2=2k2q^2 = 2k^2.

Thus q2q^2 is even, so qq is even.

But pp and qq both even contradicts gcd(p,q)=1\gcd(p, q) = 1. \square


2. Infinitude of Primes (Euclid)

Theorem: There are infinitely many primes.

Proof:

Suppose finitely many primes: p1,,pnp_1, \ldots, p_n.

Let N=p1pn+1N = p_1 \cdots p_n + 1.

Since N>1N > 1, NN has a prime divisor pp.

If p=pip = p_i for some ii, then pp1pnp \mid p_1 \cdots p_n and pNp \mid N, so p(Np1pn)=1p \mid (N - p_1 \cdots p_n) = 1.

No prime divides 1. Contradiction.

Thus pp is not in our list. \square


3. Fermat’s Little Theorem

Theorem: If pp is prime and pap \nmid a, then ap11(modp)a^{p-1} \equiv 1 \pmod{p}.

Proof:

Consider a,2a,,(p1)a(modp)a, 2a, \ldots, (p-1)a \pmod{p}.

Non-zeroness: Each ia≢0(modp)ia \not\equiv 0 \pmod{p} since pap \nmid a and pip \nmid i.

Distinctness: If iaja(modp)ia \equiv ja \pmod{p}, then p(ij)ap \mid (i-j)a. By Euclid’s lemma, p(ij)p \mid (i-j). But ij<p|i-j| < p, so i=ji = j.

Thus {a,2a,,(p1)a}\{a, 2a, \ldots, (p-1)a\} = {1,2,,p1}(modp)\{1, 2, \ldots, p-1\} \pmod{p}.

Taking products: ap1(p1)!(p1)!(modp)a^{p-1}(p-1)! \equiv (p-1)! \pmod{p}.

Since gcd((p1)!,p)=1\gcd((p-1)!, p) = 1, cancel: ap11(modp)a^{p-1} \equiv 1 \pmod{p}. \square


Part II: Sets, Relations, Functions

4. Relational Composition is Associative

Theorem: (TS)R=T(SR)(T \circ S) \circ R = T \circ (S \circ R).

Proof:

a((TS)R)d    c.aRcc(TS)da((T \circ S) \circ R)d \iff \exists c. aRc \land c(T \circ S)d     c,b.aRccSbbTd\iff \exists c, b. aRc \land cSb \land bTd

a(T(SR))d    b.a(SR)bbTda(T \circ (S \circ R))d \iff \exists b. a(S \circ R)b \land bTd     b,c.aRccSbbTd\iff \exists b, c. aRc \land cSb \land bTd

Both equivalent to b,c.aRccSbbTd\exists b, c. aRc \land cSb \land bTd. \square


5. Identity Laws for Composition

Theorem: idBR=R=RidA\text{id}_B \circ R = R = R \circ \text{id}_A.

Proof:

a(idBR)b    c.aRccidBb    c.aRcc=b    aRba(\text{id}_B \circ R)b \iff \exists c. aRc \land c\,\text{id}_B\,b \iff \exists c. aRc \land c = b \iff aRb.

a(RidA)b    c.aidAccRb    c.a=ccRb    aRba(R \circ \text{id}_A)b \iff \exists c. a\,\text{id}_A\,c \land cRb \iff \exists c. a = c \land cRb \iff aRb. \square


6. Reflexive-Transitive Closure is Smallest Preorder

Theorem: RR^{\circ *} is the smallest preorder containing RR.

Proof:

RR^{\circ *} is a preorder: reflexive (contains id\text{id}), transitive (path concatenation).

RRR \subseteq R^{\circ *}: Clear since R1=RR^{\circ 1} = R.

Smallest: Let PP be a preorder with RPR \subseteq P. By induction on nn: RnPR^{\circ n} \subseteq P.

  • Base: idP\text{id} \subseteq P by reflexivity.
  • Step: R(n+1)=RRnPPPR^{\circ(n+1)} = R \circ R^{\circ n} \subseteq P \circ P \subseteq P by transitivity.

Thus R=nRnPR^{\circ *} = \bigcup_n R^{\circ n} \subseteq P. \square


7. Equivalence Relations \leftrightarrow Partitions Bijection

Theorem: Φ:EqRel(A)Part(A)\Phi : \text{EqRel}(A) \to \text{Part}(A) by Φ(E)={[a]E:aA}\Phi(E) = \{[a]_E : a \in A\} is a bijection.

Proof:

Well-defined: Equivalence classes partition AA (non-empty by reflexivity, disjoint by transitivity/symmetry, cover by definition).

Inverse: Given partition PP, define E={(a,b):BP.a,bB}E = \{(a, b) : \exists B \in P. a, b \in B\}.

This is an equivalence relation (reflexive: aa in unique block; symmetric: trivial; transitive: blocks disjoint).

Inverses: Check Φ1(Φ(E))=E\Phi^{-1}(\Phi(E)) = E (elements in same class iff equivalent) and Φ(Φ1(P))=P\Phi(\Phi^{-1}(P)) = P (blocks are the equivalence classes). \square


8. Isomorphisms are Bijective

Theorem: f:ABf : A \to B is an isomorphism iff ff is bijective.

Proof:

()(\Rightarrow): If f1f^{-1} exists: injective (if f(a1)=f(a2)f(a_1) = f(a_2), then a1=f1(f(a1))=f1(f(a2))=a2a_1 = f^{-1}(f(a_1)) = f^{-1}(f(a_2)) = a_2); surjective (for bb, b=f(f1(b))b = f(f^{-1}(b))).

()(\Leftarrow): Define f1(b)=f^{-1}(b) = the unique aa with f(a)=bf(a) = b. Verify f1f=idf^{-1} \circ f = \text{id} and ff1=idf \circ f^{-1} = \text{id}. \square


9. Sections and Idempotents

Theorem: (s,r)(s, r) section-retraction pair \leftrightarrow e=sre = s \circ r idempotent.

Proof:

()(\Rightarrow): ee=srsr=sidr=sr=ee \circ e = s \circ r \circ s \circ r = s \circ \text{id} \circ r = s \circ r = e.

()(\Leftarrow): Given idempotent ee, let B=Im(e)B = \text{Im}(e). Define r(a)=e(a)r(a) = e(a), s(b)=bs(b) = b (inclusion). Then r(s(b))=e(b)=br(s(b)) = e(b) = b since bIm(e)b \in \text{Im}(e) means b=e(a)b = e(a) for some aa. \square


10. Cantor’s Theorem

Theorem: No surjection AP(A)A \twoheadrightarrow \mathcal{P}(A) exists.

Proof:

Suppose e:AP(A)e : A \twoheadrightarrow \mathcal{P}(A). Define U={xA:xe(x)}U = \{x \in A : x \notin e(x)\}.

Since ee surjective, a.e(a)=U\exists a. e(a) = U.

Is aUa \in U?

  • If yes: aUae(a)=Ua \in U \Rightarrow a \notin e(a) = U. Contradiction.
  • If no: aUae(a)=Ua \notin U \Rightarrow a \in e(a) = U. Contradiction.

Thus no such ee exists. \square


11. Well-Founded Induction

Theorem: If (yx.P(y))P(x)(\forall y \prec x. P(y)) \Rightarrow P(x) for all xx, then x.P(x)\forall x. P(x).

Proof:

Suppose S={x:¬P(x)}S = \{x : \neg P(x)\} \neq \emptyset.

By well-foundedness, SS has minimal element mm.

For all ymy \prec m: ySy \notin S, so P(y)P(y) holds.

By assumption, P(m)P(m) holds, contradicting mSm \in S.

Thus S=S = \emptyset. \square


Part III: Formal Languages and Automata

12. Rule Induction

Theorem: If SS is closed under all rules in RR, then XRSX_R \subseteq S.

Proof:

Induction on derivation height.

Base: Axioms have height 1. Since SS closed under rules with no premises, axioms are in SS.

Step: Derivation of height n+1n+1 uses rule u1,,ukv\frac{u_1, \ldots, u_k}{v} where each uiu_i derived by height n\leq n.

By IH, each uiSu_i \in S. Since SS closed under the rule, vSv \in S. \square


13. Subset Construction Correctness

Theorem: L(P(M))=L(M)L(\mathcal{P}(M)) = L(M).

Proof:

Soundness (L(P(M))L(M)L(\mathcal{P}(M)) \subseteq L(M)): DFA path sSnFs' \to S_n \cap F \neq \emptyset. Build NFA path by tracing elements in each SiS_i back to NFA states. Some qfSnFq_f \in S_n \cap F reachable from ss.

Completeness (L(M)L(P(M))L(M) \subseteq L(\mathcal{P}(M))): NFA path swqfs \Rightarrow_w q_f. Track reachable states SiS_i at each step. qfSnq_f \in S_n, so SnFS_n \cap F \neq \emptyset. \square


14. Kleene’s Theorem

Theorem: LL is regular iff L=L(R)L = L(R) for some regex RR.

Proof:

Regex \Rightarrow NFA: Inductive construction (Thompson) for ε\varepsilon, \emptyset, symbols, union, concat, star. Each preserves language.

NFA \Rightarrow Regex: Region decomposition. Compute R(q,S,q)R(q, S, q') for paths through regions. Eliminate all states: L(M)=qfFR(s,Q,qf)L(M) = \bigcup_{q_f \in F} R(s, Q, q_f). \square


15. Pumping Lemma

Theorem: For LL regular, p.wL\exists p. \forall w \in L with wp|w| \geq p: w=xyzw = xyz with xyp|xy| \leq p, y1|y| \geq 1, and i.xyizL\forall i. xy^iz \in L.

Proof:

Let DFA have nn states. Set p=np = n.

For wn|w| \geq n: computation visits n+1n+1 states. By pigeonhole, some state repeated.

Let xx = prefix before loop, yy = loop, zz = suffix.

  • xyn=p|xy| \leq n = p (loop within first nn states)
  • y1|y| \geq 1 (loop non-trivial)
  • Can repeat or omit loop; stays accepting. \square

Summary

These 15 proofs form the core of examinable material. Key strategies:

  1. Contradiction: Assume negation, derive 2\sqrt{2}, primes, Cantor
  2. Induction: Natural numbers, derivation height, state elimination
  3. Construction: Explicit bijections, automata constructions
  4. Unfolding definitions: Relational composition, closure properties