Proof by Contrapositive

Lay out an even-odd table and set a statement beside its contrapositive. This is the detour for when the direct road is blocked.

Some statements resist a frontal attack. Then flip and negate: prove the contrapositive instead. It always shares the original's truth value. Settle one side and the other comes free.

For example, if a square is even, the number itself is even. Instead of studying squares, look only at odd numbers. Squaring an odd number always gives an odd number. So an even square never came from an odd number.

n2 even⇒n even  ⟺  n odd⇒n2 oddn^2 \text{ even} \Rightarrow n \text{ even} \iff n \text{ odd} \Rightarrow n^2 \text{ odd}
(2k+1)2=2(2k2+2k)+1(2k+1)^2 = 2(2k^2+2k)+1

Examples & Exercises

Prove completely, by contrapositive, that for an integer nn, "if n2n^2 is even then nn is even".

Reduce the claim to its contrapositive, "if nn is odd then n2n^2 is odd", and prove that by the odd-square expansion.

n2 even⇒n even⟺n odd⇒n2 oddn^2 \text{ even} \Rightarrow n \text{ even} \quad\Longleftrightarrow\quad n \text{ odd} \Rightarrow n^2 \text{ odd}
n=7n = 7 (odd): n=2⋅3+1n = 2 \cdot 3 + 1, n2=49=2⋅24+1n^2 = 49 = 2 \cdot 24 + 1 — odd. The contrapositive computation is on display.
Show proof
  1. The claim is P⇒QP \Rightarrow Q with PP: "n2n^2 is even" and QQ: "nn is even". An implication is logically equivalent to its contrapositive (both fail exactly when PP is true and QQ false), so we prove ¬Q⇒¬P\lnot Q \Rightarrow \lnot P: "if nn is odd then n2n^2 is odd".

  2. The contrapositive is proved directly: if nn is odd, then n=2k+1n = 2k + 1 for some integer kk, and n2=(2k+1)2=4k2+4k+1=2(2k2+2k)+1n^2 = (2k+1)^2 = 4k^2 + 4k + 1 = 2(2k^2 + 2k) + 1.

    Since 2k2+2k2k^2 + 2k is an integer, n2n^2 is odd.

    Check in the scene: n=7n = 7 gives k=3k = 3 and n2=49=2⋅(2⋅9+6)+1=2⋅24+1n^2 = 49 = 2 \cdot (2 \cdot 9 + 6) + 1 = 2 \cdot 24 + 1.

    The contrapositive is proved, so by equivalence the original claim "if n2n^2 is even then nn is even" is proved.

    (Why the direct attempt stalls: the assumption n2=2mn^2 = 2m gives no handle for computing with nn — the contrapositive swaps it for the workable form n=2k+1n = 2k + 1.)

    (2k+1)2=4k2+4k+1=2(2k2+2k)+1(2k+1)^2 = 4k^2 + 4k + 1 = 2(2k^2 + 2k) + 1

Prove by contrapositive that for an integer nn, "if 3n+23n + 2 is odd then nn is odd".

First explain why the direct route is awkward: "3n+23n + 2 odd" only yields "3n3n odd", with no direct grip on nn itself.

3n+2 odd⇒n odd⟺n even⇒3n+2 even3n + 2 \text{ odd} \Rightarrow n \text{ odd} \quad\Longleftrightarrow\quad n \text{ even} \Rightarrow 3n + 2 \text{ even}
n=8n = 8 (even): n=2⋅4n = 2 \cdot 4, 3n+2=26=2(3⋅4+1)=2⋅133n + 2 = 26 = 2(3 \cdot 4 + 1) = 2 \cdot 13 — even. The contrapositive direction is on display.
Show proof
  1. Why the direct route is awkward: assuming 3n+23n + 2 is odd gives 3n=(3n+2)−23n = (3n + 2) - 2, so 3n3n is odd.

    But passing from "3 times nn is odd" to "nn is odd" needs a separate argument that nn cannot be even — there is no computable form of nn to work with.

    So we turn to the contrapositive: "if nn is even then 3n+23n + 2 is even".

  2. Prove the contrapositive directly: if nn is even, write n=2kn = 2k with kk an integer; then 3n+2=3(2k)+2=6k+2=2(3k+1)3n + 2 = 3(2k) + 2 = 6k + 2 = 2(3k + 1).

    Since 3k+13k + 1 is an integer, 3n+23n + 2 is even.

    The contrapositive is proved, so by equivalence the original claim "if 3n+23n + 2 is odd then nn is odd" is proved.

    Check in the scene: n=8=2⋅4n = 8 = 2 \cdot 4 gives 3n+2=26=2(3⋅4+1)=2⋅133n + 2 = 26 = 2(3 \cdot 4 + 1) = 2 \cdot 13, even.

Prove that for an integer nn, "if n2n^2 is divisible by 3 then nn is divisible by 3".

Take the contrapositive, then split the case "3 does not divide nn" into n=3k+1n = 3k + 1 and n=3k+2n = 3k + 2, showing in each case that 3 does not divide n2n^2.

n=3k+1⇒n2=3(3k2+2k)+1,n=3k+2⇒n2=3(3k2+4k+1)+1n = 3k+1 \Rightarrow n^2 = 3(3k^2 + 2k) + 1, \qquad n = 3k+2 \Rightarrow n^2 = 3(3k^2 + 4k + 1) + 1
n=9n = 9: n2=81=3⋅27n^2 = 81 = 3 \cdot 27 — divisible by 3, and n=9=3⋅3n = 9 = 3 \cdot 3 is too. A check in the statement's own form.
Show proof
  1. We prove the contrapositive: "if 3 does not divide nn then 3 does not divide n2n^2".

    By the division algorithm, if 3 does not divide nn the remainder is 1 or 2, so n=3k+1n = 3k + 1 or n=3k+2n = 3k + 2 for some integer kk.

    Compute each case.

  2. Case 1: if n=3k+1n = 3k + 1 then n2=9k2+6k+1=3(3k2+2k)+1n^2 = 9k^2 + 6k + 1 = 3(3k^2 + 2k) + 1 — remainder 1 on division by 3, so 3 does not divide n2n^2.

    Case 2: if n=3k+2n = 3k + 2 then n2=9k2+12k+4=3(3k2+4k+1)+1n^2 = 9k^2 + 12k + 4 = 3(3k^2 + 4k + 1) + 1 — again remainder 1, so 3 does not divide n2n^2.

    Both cases give the contrapositive's conclusion, so the contrapositive is proved, and by equivalence so is the original claim.

    Check in the scene: n=9=3⋅3n = 9 = 3 \cdot 3 is divisible by 3 and n2=81=3⋅27n^2 = 81 = 3 \cdot 27 is too.

1. Foundations of Proof