Skip to content
← Explore another subjectPearson Edexcel International AS & A Level · WMA

Proof by contradiction: the method and the irrationality of √2

Mathematics (WMA) · Proof; algebra and functions; coordinate geometry in the (x, y) plane; binomial expansion

ONE COMPLETE FREE SECTION

Your course level

This syllabus has no tier selection.

The sample stays complete. Sign up when you’re ready to continue the unit.

0 highlighted

Proof by contradiction: the method and the irrationality of √2

Pure Mathematics P4 (WMA14/01) is the second A Level pure paper, and proof has been part of every recent sitting.

On each of the three recent P4 papers studied for these notes, one question (or one part of a question) asked for a proof by contradiction, and it was worth 4 marks every time: twice as a question on its own and once as part (b) of a binomial question.

Those 4 marks are some of the hardest on the paper, because the examiners report that even the opening line is often wrong.

P4 assumes everything in P1, P2 and P3, and this section leans on two P2 ideas: what a proof is, and how to write odd and even numbers as 2k+12k + 1 and 2k2k.

The idea: show that the opposite is impossible

Most proofs you met in P2 go forwards.

You start from facts you know and reason step by step until you reach the statement.

Proof by contradiction goes the other way round.

You suppose the statement is false, reason correctly from that supposition, and arrive at something that cannot possibly be true.

Since correct reasoning from a true starting point can never produce an impossibility, the starting point must have been false.

So the statement was true all along.

You already use this kind of reasoning outside mathematics.

A detective says: "Suppose the butler did it.

Then he was in the library at nine.

But at nine he was on a train, seen by forty passengers.

So the butler did not do it."

The detective never proves directly who did it; she shows one possibility leads to an impossibility and rules it out.

The method works because a mathematical statement is either true or false, with no third option.

If "the statement is false" leads to nonsense, then "the statement is true" is the only option left.

When to reach for it

Proof by contradiction is the natural tool when the statement says something cannot happen or does not exist:

  • "there are no integers aa and bb such that ...";
  • "2\sqrt{2} is irrational" (it cannot be written as a fraction);
  • "there is no largest prime" (the list of primes does not end);
  • "these two curves never meet".

A direct proof of "there is no such thing" is hard, because you would have to check every case.

Assuming there is such a thing gives you an object to work with: a fraction ab\frac{a}{b}, a largest prime pp, a point where the curves meet.

Then you show that object leads to an impossibility.

Step 1 is the negation: getting the assumption right

Everything depends on assuming exactly the opposite of the statement, no more and no less.

Assume too little and the contradiction does not prove the claim; assume something else and the proof is about a different statement.

The examiners' reports say this is where most proofs go wrong, and a mark scheme awards nothing until the assumption is correct.

Statement to prove

The assumption (its negation)

2\sqrt{2} is irrational

2\sqrt{2} is rational, so 2=ab\sqrt{2} = \dfrac{a}{b} for integers a,ba, b with b≠0b \neq 0

There are no integers a,ba, b with 6a+15b=16a + 15b = 1

There are integers a,ba, b with 6a+15b=16a + 15b = 1

For every integer nn, if n2n^2 is even then nn is even

There is an integer nn such that n2n^2 is even and nn is odd

For all real xx, x2+1≥2xx^2 + 1 \geq 2x

There is a real xx for which x2+1<2xx^2 + 1 < 2x

There are infinitely many primes

There are only finitely many primes

Three patterns are worth learning.

  • "For all" becomes "there is at least one". To deny that every integer has a property, you only need one integer without it.
  • "If P then Q" becomes "P and not Q". The statement "if n2n^2 is even then nn is even" is false only when n2n^2 is even and yet nn is odd. Assuming "n2n^2 is odd" or "nn is odd" alone negates something else.
  • "≥\geq" becomes "<<", and ">>" becomes "≤\leq". The negation of a non-strict inequality is strict, and the other way round. A proof that negates ≥\geq as ≤\leq has assumed the wrong thing.

Proof: if n2n^2 is even, then nn is even

This small result is the key to the proof that 2\sqrt{2} is irrational, so it comes first.

Assume there is an integer nn such that n2n^2 is even and nn is odd.

Since nn is odd, n=2k+1n = 2k + 1 for some integer kk.

Then

n2=(2k+1)2=4k2+4k+1=2(2k2+2k)+1.n^2 = (2k + 1)^2 = 4k^2 + 4k + 1 = 2(2k^2 + 2k) + 1.

Since 2k2+2k2k^2 + 2k is an integer, n2n^2 is one more than an even number, so n2n^2 is odd.

This contradicts the assumption that n2n^2 is even.

So there is no such nn: for every integer nn, if n2n^2 is even then nn is even.

Notice the letter kk.

The odd number is written 2k+12k + 1 with a new letter.

Writing "n=2n+1n = 2n + 1" reuses nn for two different things, and the examiners record that this slip costs the last mark.

Proof: 2\sqrt{2} is irrational

This is a named proof you must know, and it is the model for many others, so learn its shape.

Assume 2\sqrt{2} is rational.

Then 2=ab\sqrt{2} = \dfrac{a}{b}, where aa and bb are integers, b≠0b \neq 0, and the fraction is in its lowest terms, so aa and bb have no common factor other than 1.

Every fraction can be cancelled down to lowest terms, so this costs nothing.

Squaring, 2=a2b22 = \dfrac{a^2}{b^2}, so a2=2b2a^2 = 2b^2.

So a2a^2 is even, and by the result above, aa is even.

Write a=2ca = 2c for an integer cc.

Then 4c2=2b24c^2 = 2b^2, so b2=2c2b^2 = 2c^2.

So b2b^2 is even, and therefore bb is even.

Now aa and bb are both even, so they have the common factor 2.

This contradicts the fraction being in lowest terms.

So 2\sqrt{2} is not rational: 2\sqrt{2} is irrational.

Why does "lowest terms" matter so much?

It is the fact the contradiction breaks.

Without it, finding that aa and bb are both even would be no problem at all: 46\frac{4}{6} has an even top and bottom and is a perfectly good fraction.

The phrase is not decoration; it is the trap the proof is built to spring.

Why 4\sqrt{4} does not fall into the same trap

A good test of understanding is to run the same argument on a number that is rational.

Try 4=ab\sqrt{4} = \frac{a}{b} in lowest terms.

Then a2=4b2a^2 = 4b^2, so a2a^2 is even, so aa is even: a=2ca = 2c.

Then 4c2=4b24c^2 = 4b^2, so b2=c2b^2 = c^2.

Nothing forces bb to be even, so no contradiction appears, and the argument stops.

That is exactly what should happen, since 4=21\sqrt{4} = \frac{2}{1}.

The proof for 2\sqrt{2} works because the 2 in a2=2b2a^2 = 2b^2 survives to make b2=2c2b^2 = 2c^2 even.

The shape of every proof by contradiction

1. Assume the negation, in words, with every condition in it (integers, lowest terms, the interval for xx).

2. Work from the assumption with correct algebra, writing odd and even numbers or fractions with new letters.

3. Reach a statement that is impossible, and say why it is impossible (it contradicts the lowest-terms condition; an even number cannot equal an odd one; a square cannot be negative).

4. Conclude: "this contradicts the assumption, so [the original statement] is true."

The four parts match the four marks a P4 proof usually carries.

How the marks work
  • M1: a correct method applied to this question. dM1, ddM1: method marks that need the M marks before them.
  • A1: a correct line or answer, which needs its M mark. A1*: the last mark of a "show that" or a proof, reached with no errors; in a proof by contradiction it needs the reason, the contradiction and the conclusion. A1cso: correct solution only.
  • B1: an independent mark, often the assumption in a proof or a stated value. ft: follow through from your own earlier value.
  • awrt: any answer which rounds to the value given. isw: ignore subsequent working. oe: or equivalent.
  • Misread: copy a number wrongly and keep the question's character, and you lose two A or B marks in that part, not everything.
  • Where a question says solutions relying entirely on calculator technology are not acceptable, a bare answer scores nothing.
Key terms
  • Proof by contradiction: a proof that assumes the statement is false and shows that this assumption leads to a contradiction, so the statement must be true.
  • Rational number: a number that can be written as ab\dfrac{a}{b}, where aa and bb are integers and b≠0b \neq 0.
  • Irrational number: a real number that cannot be written as ab\dfrac{a}{b} with aa and bb integers and b≠0b \neq 0.
  • Lowest terms: a fraction ab\dfrac{a}{b} is in lowest terms when aa and bb have no common factor other than 1.
  • Negation: the statement that is true exactly when the original statement is false.
  • The wording to learn: "Proof by contradiction", with the guidance "Including proof of the irrationality of 2\sqrt{2} and the infinity of primes, and application to unfamiliar proofs."
Worked example 1: a parity result

Use proof by contradiction to prove that, for every integer nn, if n3n^3 is even then nn is even.

(4)

1. Assume there is an integer nn such that n3n^3 is even and nn is odd. B1 (assumption in words, both parts of the claim negated)

2. nn odd, so n=2k+1n = 2k + 1 for some integer kk, and n3=(2k+1)3=8k3+12k2+6k+1n^3 = (2k + 1)^3 = 8k^3 + 12k^2 + 6k + 1 M1 (expands with a new letter)

3. n3=2(4k3+6k2+3k)+1n^3 = 2(4k^3 + 6k^2 + 3k) + 1, which is odd since 4k3+6k2+3k4k^3 + 6k^2 + 3k is an integer A1

4. This contradicts n3n^3 being even, so for every integer nn, if n3n^3 is even then nn is even. A1*

Answer: proved.

The check on the algebra: at k=1k = 1, n=3n = 3 and n3=27=2(4+6+3)+1n^3 = 27 = 2(4 + 6 + 3) + 1.

Worked example 2: hence an irrational cube root

Hence, using the result of Worked example 1, prove by contradiction that 23\sqrt[3]{2} is irrational.

(4)

1. Assume 23\sqrt[3]{2} is rational, so 23=ab\sqrt[3]{2} = \dfrac{a}{b} with a,ba, b integers, b≠0b \neq 0, and aa and bb having no common factor other than 1 B1

2. Cubing: a3=2b3a^3 = 2b^3, so a3a^3 is even and, by Worked example 1, aa is even; let a=2ca = 2c M1

3. 8c3=2b38c^3 = 2b^3, so b3=4c3=2(2c3)b^3 = 4c^3 = 2(2c^3), which is even, so bb is even A1

4. Then aa and bb have the common factor 2, which contradicts the fraction being in lowest terms. So 23\sqrt[3]{2} is irrational. A1*

Answer: proved.

The "hence" is used in two places: once for aa and once for bb.

Without the earlier result you would have to prove "a3a^3 even means aa even" again inside this proof.

Worked example 3: a "there are no integers" proof (exam length)

In this question you must show all stages of your working. Solutions relying entirely on calculator technology are not acceptable.

Use proof by contradiction to prove that there are no positive integers xx and yy such that x2−y2=10x^2 - y^2 = 10.

(4)

1. Assume there are positive integers xx and yy with x2−y2=10x^2 - y^2 = 10. B1

2. Then (x−y)(x+y)=10(x - y)(x + y) = 10. Since x+y>0x + y > 0 and the product is positive, x−y>0x - y > 0, so x−yx - y and x+yx + y are positive integers with x−y<x+yx - y < x + y whose product is 10 M1

3. So (x−y=1x - y = 1 and x+y=10x + y = 10) or (x−y=2x - y = 2 and x+y=5x + y = 5). In each pair one number is odd and one is even. But (x+y)−(x−y)=2y(x + y) - (x - y) = 2y is even, so x−yx - y and x+yx + y are both even or both odd A1

4. Neither pair has the same parity, which is a contradiction. So there are no positive integers xx and yy with x2−y2=10x^2 - y^2 = 10. A1*

Answer: proved.

The key fact is the parity link: two integers that differ by an even number 2y2y are both odd or both even.

A list of trial values (x=4,y=…x = 4, y = \ldots) is not a proof, because it can never cover every integer.

Exam tip

Write the assumption as a full sentence that starts "Assume" (or "Suppose") and contains every condition in the claim: the type of number, "in lowest terms", and both halves of an "if ...

then ..."

statement.

The first mark goes on that sentence, and without it the rest of a P4 proof scores nothing.

End with a sentence that names what is contradicted and states the original claim as proved.

Common mistakes
  • Negating the wrong statement. For "if n2n^2 is even then nn is even", assume "n2n^2 is even and nn is odd". Assuming "both are odd" or just "nn is odd" proves nothing about the claim. The fix: write the claim, then change it to "the hypothesis holds and the conclusion fails".
  • Reusing a letter. "nn is odd, so n=2n+1n = 2n + 1" says n=−1n = -1. Use a new letter: n=2k+1n = 2k + 1.
  • Forgetting lowest terms. Without "no common factor other than 1" in the assumption, finding both aa and bb even is not a contradiction.
  • No reason and no conclusion. "Contradiction." on its own loses the A1*. Say what is contradicted and why, then state the result.

Check your understanding

1. Write down the negation of each statement.
(a) For every positive integer nn, n2+nn^2 + n is even. (1)
(b) There is no largest even number. (1) Total [2]

2. (a) Use proof by contradiction to prove that, for every integer nn, if n2n^2 is a multiple of 3 then nn is a multiple of 3. (4)
(b) Hence use proof by contradiction to prove that 3\sqrt{3} is irrational. (4) Total [8]

3. Use proof by contradiction to prove that if xx is irrational then 1x\dfrac{1}{x} is irrational. (4) Total [4]

4. A student tries to prove that 4\sqrt{4} is irrational by copying the proof for 2\sqrt{2}. They assume 4=ab\sqrt{4} = \dfrac{a}{b} in lowest terms and reach b2=c2b^2 = c^2. State, with a reason, why the student cannot reach a contradiction. (2) Total [2]

5. Use proof by contradiction to prove that there are no integers aa and bb such that 6a+15b=16a + 15b = 1. (4) Total [4]

6. Use proof by contradiction to show that the sum of a rational number and an irrational number is irrational. (4) Total [4]

Answers

1. (a) There is a positive integer nn for which n2+nn^2 + n is odd. B1
(b) There is a largest even number. B1

2. (a) Assume there is an integer nn such that n2n^2 is a multiple of 3 and nn is not a multiple of 3 B1; then n=3k+1n = 3k + 1 or n=3k+2n = 3k + 2 for an integer kk, giving n2=9k2+6k+1=3(3k2+2k)+1n^2 = 9k^2 + 6k + 1 = 3(3k^2 + 2k) + 1 or n2=9k2+12k+4=3(3k2+4k+1)+1n^2 = 9k^2 + 12k + 4 = 3(3k^2 + 4k + 1) + 1 M1; in both cases n2n^2 leaves remainder 1 on division by 3, so is not a multiple of 3 A1; this contradicts the assumption, so if n2n^2 is a multiple of 3 then nn is a multiple of 3 A1*
(b) Assume 3=ab\sqrt{3} = \dfrac{a}{b}, a,ba, b integers, b≠0b \neq 0, no common factor other than 1 B1; a2=3b2a^2 = 3b^2, so a2a^2 is a multiple of 3 and by (a) a=3ca = 3c M1; 9c2=3b2⇒b2=3c29c^2 = 3b^2 \Rightarrow b^2 = 3c^2, so bb is a multiple of 3 A1; aa and bb share the factor 3, contradicting lowest terms, so 3\sqrt{3} is irrational A1*

3. Assume xx is irrational and 1x\dfrac{1}{x} is rational, so 1x=pq\dfrac{1}{x} = \dfrac{p}{q} with p,qp, q integers, q≠0q \neq 0 B1; 1x≠0\frac{1}{x} \neq 0, so p≠0p \neq 0, and x=qpx = \dfrac{q}{p} M1; qq and pp are integers with p≠0p \neq 0, so xx is rational A1; this contradicts xx being irrational, so 1x\dfrac{1}{x} is irrational A1*

4. b2=c2b^2 = c^2 says nothing about whether bb is even (for example b=c=1b = c = 1) B1; so aa and bb need not share a factor, there is no contradiction, and indeed 4=21\sqrt{4} = \frac{2}{1} is rational B1

5. Assume there are integers a,ba, b with 6a+15b=16a + 15b = 1 B1; 6a+15b=3(2a+5b)6a + 15b = 3(2a + 5b), so 2a+5b=132a + 5b = \frac{1}{3} M1; 2a+5b2a + 5b is an integer because aa and bb are, but 13\frac{1}{3} is not an integer A1; contradiction, so no such integers exist A1*

6. Assume rr is rational, xx is irrational and r+xr + x is rational B1; write r=abr = \dfrac{a}{b} and r+x=cdr + x = \dfrac{c}{d} with integers and b,d≠0b, d \neq 0 M1; then x=cd−ab=bc−adbdx = \dfrac{c}{d} - \dfrac{a}{b} = \dfrac{bc - ad}{bd}, a quotient of integers with bd≠0bd \neq 0, so xx is rational A1; this contradicts xx being irrational, so the sum of a rational number and an irrational number is irrational A1*

Ready for the next section?

Create a free account to read the rest of this free unit, save highlights, track your confidence and plan your revision.

Your next step

Put this topic into practice.

These tools belong to this free unit. Create an account to open them and save your progress.

Practice paper · Sign up to open

Mathematics WMA Topic P4.1-4: Proof; algebra and functions; coordinate geometry in the (x, y) plane; binomial expansion. Structured practice

6 questions · WMA14/01

Attempt the questions, check the mark scheme and record your own marks.

Create free account to open this practice

Other units and paper downloads depend on your plan. This unit has no Section A quiz; use its other practice and checklist.

Already have an account? Sign in →

Give your revision a routine.

Build a timetable, tick off daily tasks, catch up on missed work and opt in to weekly email reviews. Connect with your teacher when you want to share your progress.