20 Mathematical Induction
Mathematical induction proves a statement for every positive integer \(n\), which no finite amount of checking could do. Consider \[1=1=1^{2},\quad 1+3=4=2^{2},\quad 1+3+5=9=3^{2},\quad 1+3+5+7=16=4^{2}.\] The pattern is unmistakable, and yet checking a hundred more cases would still leave infinitely many untested. Induction settles all of them at once.
Theorem 20.1 (Principle of mathematical induction). Let \(P(n)\) be a statement about the positive integer \(n\). If
- (i).
- \(P(1)\) is true, and
- (ii).
- whenever \(P(k)\) is true, \(P(k+1)\) is also true,
then \(P(n)\) is true for every positive integer \(n\).
Note 20.2. The ladder is the standard image, and it is a good one. Condition (i) says you can reach the first rung. Condition (ii) says that from any rung you can always reach the next. Together they let you climb as high as you like — and note that neither alone would do: being able to step from rung to rung is useless if you cannot get on, and standing on the bottom rung is useless if you cannot step up.
The step most often misunderstood is (ii). You are not asked to show that \(P(k+1)\) is true. You are asked to show that if \(P(k)\) were true then \(P(k+1)\) would follow — an implication, proved without ever knowing whether \(P(k)\) actually holds. That is why assuming \(P(k)\) in the middle of the proof is not circular reasoning: the assumption is the hypothesis of the implication being established, not a claim about the world.
The assumption made in step 2 is called the inductive hypothesis, and every proof below turns on using it. If a proof of step 3 never appeals to step 2, something has gone wrong — either the statement did not need induction, or the argument has a hole.
Example 20.3. Prove by mathematical induction that for all positive integers \(n\):
- (a).
- \(1+3+5+\cdots +(2n-1)=n^{2}\)
- (b).
- \(4+8+12+\cdots +4n=2n(n+1)\)
- (c).
- \(\frac {1}{2}+\frac {1}{2^{2}}+\frac {1}{2^{3}}+\cdots +\frac {1}{2^{n}}=1-\frac {1}{2^{n}}\)
- (d).
- \(x^{n}-y^{n}\) is divisible by \(x-y\)
Proof of (a). Step 1. For \(n=1\) the left side is \(2(1)-1=1\) and the right side is \(1^{2}=1\). They agree, so the statement is true for \(n=1\).
Step 2. Assume it holds for \(n=k\): \[1+3+5+\cdots +(2k-1)=k^{2}.\]
Step 3. We show it then holds for \(n=k+1\), that is \[1+3+5+\cdots +(2k-1)+\left (2(k+1)-1\right )=(k+1)^{2}.\] Taking the left side and replacing the first \(k\) terms using the inductive hypothesis, \begin {align*} k^{2}+\left (2(k+1)-1\right ) &= k^{2}+2k+2-1\\ &= k^{2}+2k+1\\ &= (k+1)^{2}, \end {align*}
which is the right side. So the statement holds for \(n=k+1\), and by induction it holds for every positive integer \(n\). □
Proof of (b). Step 1. For \(n=1\) the left side is \(4\) and the right side is \(2(1)(1+1)=4\). True for \(n=1\).
Step 2. Assume \(4+8+12+\cdots +4k=2k(k+1)\).
Step 3. Adding the next term \(4(k+1)\) to both sides, \begin {align*} \underbrace {4+8+\cdots +4k}_{=\,2k(k+1)}+4(k+1) &= 2k(k+1)+4(k+1)\\ &= 2(k+1)(k+2), \end {align*}
taking out the common factor \(2(k+1)\). This is the required formula with \(k+1\) in place of \(n\), so the result follows by induction. □
Proof of (c). Step 1. For \(n=1\), the left side is \(\frac {1}{2}\) and the right side is \(1-\frac {1}{2}=\frac {1}{2}\). True.
Step 2. Assume \(\frac {1}{2}+\frac {1}{2^{2}}+\cdots +\frac {1}{2^{k}}=1-\frac {1}{2^{k}}\).
Step 3. Adding \(\frac {1}{2^{k+1}}\) to both sides, \begin {align*} 1-\frac {1}{2^{k}}+\frac {1}{2^{k+1}} &= \frac {2\cdot 2^{k}-2+1}{2\cdot 2^{k}}\\ &= \frac {2\cdot 2^{k}-1}{2\cdot 2^{k}}\\ &= 1-\frac {1}{2\cdot 2^{k}} = 1-\frac {1}{2^{k+1}}, \end {align*}
as required, so the result holds for all \(n\) by induction. □
Proof of (d). Step 1. For \(n=1\), \(x^{1}-y^{1}=x-y\), which is divisible by \(x-y\).
Step 2. Assume \(x^{k}-y^{k}\) is divisible by \(x-y\).
Step 3. Write \(x^{k+1}-y^{k+1}\) by adding and subtracting \(xy^{k}\): \[x^{k+1}-y^{k+1}=x^{k}x-xy^{k}+xy^{k}-y^{k}y =x\left (x^{k}-y^{k}\right )+y^{k}(x-y).\] The first term is divisible by \(x-y\) by the inductive hypothesis; the second has \(x-y\) as an explicit factor. A sum of two multiples of \(x-y\) is again a multiple of \(x-y\), so \(x^{k+1}-y^{k+1}\) is divisible by \(x-y\) and the result follows. □
Note 20.4. Part (d) shows the manoeuvre that makes divisibility proofs work: adding and subtracting the same quantity — here \(xy^{k}\) — to manufacture a term the inductive hypothesis can be applied to. Nothing is changed by adding zero, but the expression is regrouped into a piece we already know about plus a piece that is visibly divisible.
Look for the same move in the practice questions below. To show \(6^{n}+4\) is divisible by \(5\), write \(6^{k+1}+4\) as \(6\left (6^{k}+4\right )-20\): the first part is divisible by \(5\) by hypothesis, the second obviously.
Questions on this section
Stuck on something here? Ask below and it stays attached to this topic.