Proof by Contraposition

Learn how to construct a proof by contraposition.

The structure of proof by contraposition

We can consider this an indirect method. We know the following logical equivalence:

(PQ)(¬Q¬P).\left(P\Rightarrow Q\right)\Leftrightarrow \left(\neg Q \Rightarrow \neg P\right) .

Starting the proof with PP and trying to conclude QQ is sometimes harder than starting with ¬Q\neg Q and concluding ¬P\neg P. If we want to prove the following statement ss:

s=x(P(x)Q(x)).s=\forall x \left(P(x)\Rightarrow Q(x)\right).

We instead prove the following logically equivalent statement tt:

t=x(¬Q(x)¬P(x)).t=\forall x \left(\neg Q(x)\Rightarrow \neg P(x)\right).

As we know that both ss and tt are logically equivalent, showing that tt is true. This means ss is also true and vice versa. Let’s look at a few examples to explore this method further.

Examples

Let’s assume xx and yy are two real numbers. We will prove the following theorem:

Theorem 1

For any real numbers xx and yy, if x+y2x+y \ge 2 then x1y1x\ge 1 \lor y\ge 1.

Proof

We have to show that ((x+y)2)(x1y1)\left((x+y)\ge 2\right) \Rightarrow \left( x\ge 1 \lor y\ge 1 \right), which is PQP\Rightarrow Q, where, P:=((x+y)2)P:= \left((x+y)\ge 2\right), and Q:=(x1y1)Q:=\left( x\ge 1 \lor y\ge 1 \right). We instead prove that ¬Q¬P\neg Q \Rightarrow \neg P. We start with ¬Q\neg Q, which is the hypothesis. We use DeMorgan’s law to achieve the following equivalence:

\neg Q \equiv \left( x<1 \land y<1 \right).

As, x<1, and y<1, we can add both the inequalities to conclude \left(x+y\right)<2. That is exactly ¬P\neg P. Therefore, ¬Q¬P\neg Q \Rightarrow \neg P is true. We have the following tautology:

(¬Q¬P)(PQ).\left(\neg Q \Rightarrow \neg P\right)\Leftrightarrow (P\Rightarrow Q).

Therefore, PQP\Rightarrow Q is true.

Theorem 2

For any real numbers x>0 and y>0, if m=xym=xy, then (mxmy)\left(|\sqrt{m}|\ge x \lor |\sqrt{m}|\ge y\right).

Proof

We are given that both xx and yy are positive. We have to show that PQP\Rightarrow Q is true, where, P:=(m=xy)P:= \left(m=xy\right), and Q:=(mxmy)Q:=\left(|\sqrt{m}|\ge x \lor |\sqrt{m}|\ge y\right). Because, (PQ)(¬Q¬P)(P\Rightarrow Q)\Leftrightarrow \left(\neg Q \Rightarrow \neg P\right) is a tautology, we will instead show that (¬Q¬P)\left(\neg Q \Rightarrow \neg P\right) is true. We use DeMorgan’s law to achieve the following equivalence:

\neg Q \equiv \left(|\sqrt{m}| < x \land |\sqrt{m}| < y\right).

As we have, \left(|\sqrt{m}| < x\right), we can multiply it by m|\sqrt{m}|. Therefore, we get \left(m < x|\sqrt{m}|\right). Similarly, we have \left(|\sqrt{m}| < y\right) and we get \left(x|\sqrt{m}| < xy\right), because x>0. We can use the transitive property of inequalities to get $m