A1 - Proof by induction

A1 - Proof by induction

Checking a formula for ten, a thousand or a million cases still leaves infinitely many cases unchecked. Proof by induction replaces that endless task with a starting case and a rule that carries truth to the next integer. You will use it to prove finite sums and recurrence formulae, and learn to recognise a gap in a proposed proof.

The structure of an induction proof

How can a finite argument prove infinitely many statements? The first few odd-number sums suggest

1=12,1+3=22,1+3+5=321=1^2,\qquad 1+3=2^2,\qquad 1+3+5=3^2

There is a reason the pattern continues: adding the next odd number, 2n+12n+1, changes n2n^2 into (n+1)2(n+1)^2. But this link needs a verified starting point.

Proof by induction

Let P(n)P(n) be a statement about integers n≥n0n\ge n_0. A proof by induction establishes:

  1. Basis: P(n0)P(n_0) is true.
  2. Inductive step: for an arbitrary integer k≥n0k\ge n_0, assume P(k)P(k), then deduce P(k+1)P(k+1).
  3. Conclusion: since the basis holds and truth passes from each integer to the next, P(n)P(n) holds for every integer n≥n0n\ge n_0.

The assumption is called the induction hypothesis. You temporarily assume only the kk case in order to prove the implication P(k)⇒P(k+1)P(k)\Rightarrow P(k+1); you do not assume that both cases are true.

[DIAGRAM: asset_name: A starting case and a repeatable implication; asset_slug: aqa7367-induction-chain; caption: The basis starts the chain and the inductive implication extends it; description: A chain of boxes labelled P of one, P of two, P of three and P of four followed by dots. P of one has a checked basis label. Every arrow represents the same proved implication from P of k to P of k plus one. A second line states that the basis and the implication together give every positive integer case.]
The basis starts the chain and the inductive implication extends it

In proof notation, p⇒qp\Rightarrow q means “if pp, then qq”; p⇐qp\Leftarrow q means “if qq, then pp”. The symbol p⇔qp\Leftrightarrow q claims both directions. Induction needs the forward implication. You may also see ∴\therefore for “therefore” and ∵\because for “because”; words are equally clear.

Worked example 1: the sum of odd numbers

Question. Prove by induction that ∑r=1n(2r−1)=n2\sum_{r=1}^{n}(2r-1)=n^2 for all positive integers nn.

Plan. Verify the first sum and formula separately, then add the next odd number to the assumed sum.

For n=1n=1, the sum is 2×1−1=12\times1-1=1, and the formula is 12=11^2=1. The basis holds.

Assume, for an arbitrary integer k≥1k\ge1,

∑r=1k(2r−1)=k2\sum_{r=1}^{k}(2r-1)=k^2

Then

∑r=1k+1(2r−1)=∑r=1k(2r−1)+(2(k+1)−1)separate the new term=k2+2k+1use the hypothesis=(k+1)2factorise\begin{aligned} \sum_{r=1}^{k+1}(2r-1) &=\sum_{r=1}^{k}(2r-1)+\bigl(2(k+1)-1\bigr) && \text{separate the new term}\\ &=k^2+2k+1 && \text{use the hypothesis}\\ &=(k+1)^2 && \text{factorise} \end{aligned}

Answer. The result holds for n=1n=1, and if it holds for n=kn=k, it holds for n=k+1n=k+1. Therefore, by induction, it holds for all positive integers nn.

Check. The new term is 2(k+1)−1=2k+12(k+1)-1=2k+1, not 2k−12k-1. Independently, (k+1)2−k2=2k+1(k+1)^2-k^2=2k+1, exactly the term added.

Now change every summand by adding 22: 2r−12r-1 becomes 2r+12r+1. There are nn terms, so the proposed total changes from n2n^2 to n2+2nn^2+2n.

Worked example 2: shifting the odd numbers

Question. Prove by induction that ∑r=1n(2r+1)=n(n+2)\sum_{r=1}^{n}(2r+1)=n(n+2) for all positive integers nn.

Plan. Keep the same proof structure, changing the added term and the target formula.

For n=1n=1, the sum is 33, and the formula is 1(1+2)=31(1+2)=3.

Assume ∑r=1k(2r+1)=k(k+2)\sum_{r=1}^{k}(2r+1)=k(k+2) for an arbitrary integer k≥1k\ge1. Then

∑r=1k+1(2r+1)=k(k+2)+(2(k+1)+1)hypothesis plus new term=k2+4k+3expand=(k+1)(k+3)factorise=(k+1)((k+1)+2)identify the next formula\begin{aligned} \sum_{r=1}^{k+1}(2r+1) &=k(k+2)+\bigl(2(k+1)+1\bigr) && \text{hypothesis plus new term}\\ &=k^2+4k+3 && \text{expand}\\ &=(k+1)(k+3) && \text{factorise}\\ &=(k+1)\bigl((k+1)+2\bigr) && \text{identify the next formula} \end{aligned}

Answer. The basis is true and P(k)⇒P(k+1)P(k)\Rightarrow P(k+1). Hence, by induction, ∑r=1n(2r+1)=n(n+2)\sum_{r=1}^{n}(2r+1)=n(n+2) for every positive integer nn.

Check. Finite-sum formulae independently give 2∑r+∑1=n(n+1)+n=n(n+2)2\sum r+\sum1=n(n+1)+n=n(n+2).

The words in the conclusion describe the logical work you have done.

Watch out

“Assume true for kk, therefore true for k+1k+1” skips the proof. Show where the hypothesis is used and reach the formula with every nn replaced by k+1k+1.

Try another constant shift of the same sum.

Proving recurrence formulae

A recurrence relation defines each term from earlier terms. Induction can prove a proposed explicit formula because the recurrence itself tells you how to move from term uku_k to the next.

For example, if u1=1u_1=1 and un+1=2un+1u_{n+1}=2u_n+1, the first few terms are 1,3,7,151,3,7,15. These are each one less than a power of two, suggesting un=2n−1u_n=2^n-1. The initial terms suggest the formula; the proof establishes it.

Worked example 3: a recurrence with a constant addition

Question. The sequence is defined by u1=1u_1=1 and un+1=2un+1u_{n+1}=2u_n+1 for integers n≥1n\ge1. Prove that un=2n−1u_n=2^n-1 for all integers n≥1n\ge1.

Plan. Check the initial term, then substitute the assumed explicit formula into the recurrence.

At n=1n=1, the recurrence's initial value is u1=1u_1=1, and 21−1=12^1-1=1.

Assume uk=2k−1u_k=2^k-1 for an arbitrary integer k≥1k\ge1. Then

uk+1=2uk+1given recurrence=2(2k−1)+1use the hypothesis=2k+1−1simplify\begin{aligned} u_{k+1} &=2u_k+1 && \text{given recurrence}\\ &=2(2^k-1)+1 && \text{use the hypothesis}\\ &=2^{k+1}-1 && \text{simplify} \end{aligned}

Answer. The formula holds at n=1n=1, and its truth at kk implies its truth at k+1k+1. Thus un=2n−1u_n=2^n-1 for all integers n≥1n\ge1, by induction.

Check. Add 11 to both sides of the recurrence: un+1+1=2(un+1)u_{n+1}+1=2(u_n+1). Thus the shifted sequence doubles from u1+1=2u_1+1=2, independently explaining the powers of two.

Change only the constant added in the recurrence, from 11 to 33, while keeping u1=1u_1=1. The explicit formula must change too.

Worked example 4: changing the recurrence constant

Question. Now u1=1u_1=1 and un+1=2un+3u_{n+1}=2u_n+3 for integers n≥1n\ge1. Prove that un=2n+1−3u_n=2^{n+1}-3 for all integers n≥1n\ge1.

Plan. Verify the same initial value, then use the changed recurrence in the inductive step.

At n=1n=1, the given term is 11, and 21+1−3=12^{1+1}-3=1.

Assume uk=2k+1−3u_k=2^{k+1}-3 for an arbitrary integer k≥1k\ge1. Then

uk+1=2uk+3given recurrence=2(2k+1−3)+3use the hypothesis=2k+2−3simplify=2(k+1)+1−3the formula at k+1\begin{aligned} u_{k+1} &=2u_k+3 && \text{given recurrence}\\ &=2(2^{k+1}-3)+3 && \text{use the hypothesis}\\ &=2^{k+2}-3 && \text{simplify}\\ &=2^{(k+1)+1}-3 && \text{the formula at }k+1 \end{aligned}

Answer. The basis and inductive implication hold. Therefore un=2n+1−3u_n=2^{n+1}-3 for every integer n≥1n\ge1, by induction.

Check. The first terms from the recurrence are 1,5,131,5,13; the formula gives the same. More structurally, un+1+3=2(un+3)u_{n+1}+3=2(u_n+3), so the shifted sequence doubles from 44.

In both examples, the given recurrence supplies a justified first line for uk+1u_{k+1}. Starting with the formula you want for uk+1u_{k+1} and merely rearranging it would not show that the defined sequence satisfies it.

Problem-solving

For a recurrence, write the definition of the next term first. Substitute the induction hypothesis only after that line, then compare with the explicit formula at k+1k+1.

Keep the initial value and change the constant again.

Why examples are not a proof

A numerical check can reveal an error, but even many successful checks do not establish a claim about every positive integer. There is always another integer beyond the last one checked.

For a concrete warning, consider the claim

n+(n−1)(n−2)(n−3)(n−4)=nfor every positive integer nn+(n-1)(n-2)(n-3)(n-4)=n \qquad\text{for every positive integer }n

At n=1,2,3,4n=1,2,3,4, one factor in the product is zero, so every check succeeds. At n=5n=5, the left side is 5+4×3×2×1=295+4\times3\times2\times1=29, while the right side is 55. The formula was designed to agree at the checked values, then fail.

Counterexample

A counterexample is one allowed case in which a universal claim is false. One counterexample disproves “for every”; a finite collection of successful examples does not prove it.

An induction argument avoids the gap because kk is arbitrary. Its algebra proves the implication for every allowed step, rather than checking just one transition such as 33 to 44. With the basis, this reaches any positive integer after finitely many steps.

When checking a proposed proof, ask whether the basis was actually evaluated, whether only the current case was assumed, and whether the next case was deduced. The sentence “it is true for kk and for k+1k+1” states two truths without explaining their dependence. The needed statement is “if it is true for kk, then it is true for k+1k+1”.

Key idea

The basis supplies truth; the inductive step passes it on. Neither a collection of examples nor an implication without a starting case supplies both.

This answers the opening question: a finite proof can cover infinitely many integer cases because it proves a repeatable implication, not because it checks a very large sample.

Quick Recap

  • Verify the starting case on both sides of the proposed equality.
  • Assume the statement at an arbitrary allowed integer kk.
  • Use that hypothesis to deduce the statement at k+1k+1.
  • Conclude for the stated integer domain using both the basis and the implication.

Induction proves infinitely many cases through a starting point and a repeatable logical step.