P2 1.1–P2 1.3 - Proof and counterexamples

P2 1.1–P2 1.3 - Proof and counterexamples

Learn how assumptions support a logical conclusion, how exhaustion covers every possible case, and how one counterexample can overturn a universal claim. P2 1.1 supplies the structure of an argument, P2 1.2 gives a proof method, and P2 1.3 explains what is enough to disprove a statement.

Assumptions and logical conclusions

If a claim works for the first hundred integers, have you proved it for every integer? A hundred checks leave infinitely many unchecked cases. A proof must explain why the conclusion follows for every value allowed by the assumptions.

For example, the sum of 33 and 77 is even. To cover any two odd integers, replace those particular numbers by expressions that represent all odd integers.

Mathematical proof

A proof is a sequence of justified logical steps from stated assumptions to a conclusion. An assumption is information you may use; the conclusion is what you must establish. An even integer has the form 2k2k, and an odd integer has the form 2k+12k+1, where kk is an integer.

Different integers need different variables unless the question says they are equal. Writing both odd integers as 2k+12k+1 would silently assume they are the same integer.

Worked example 1: sum of two odd integers

Question. Prove that the sum of any two odd integers is even.

Plan. Represent the integers independently and show that their sum is twice an integer.

Let the integers be 2k+12k+1 and 2l+12l+1, where k,lk,l are integers. Then

(2k+1)+(2l+1)=2k+2l+2collect terms=2(k+l+1)factor out 2\begin{aligned} (2k+1)+(2l+1)&=2k+2l+2 &&\text{collect terms}\\ &=2(k+l+1) &&\text{factor out 2} \end{aligned}

Since k+l+1k+l+1 is an integer, this is even.

Answer. The sum of any two odd integers is even.

Check. The independent choice k=−2,l=3k=-2,l=3 gives −3+7=4-3+7=4, consistent with the proof. The algebra, rather than this check, covers all choices.

Change only the operation from addition to multiplication. The same assumptions now lead to a different conclusion.

Worked example 2: product of two odd integers

Question. Prove that the product of any two odd integers is odd.

Plan. Expand the product and arrange it as twice an integer plus one.

For integers k,lk,l,

(2k+1)(2l+1)=4kl+2k+2l+1expand=2(2kl+k+l)+1separate the final 1\begin{aligned} (2k+1)(2l+1)&=4kl+2k+2l+1 &&\text{expand}\\ &=2(2kl+k+l)+1 &&\text{separate the final 1} \end{aligned}

The expression 2kl+k+l2kl+k+l is an integer.

Answer. The product is odd for every pair of odd integers.

Check. With −3-3 and 77, the product is −21=2(−11)+1-21=2(-11)+1, which is odd.

Writing an expression in the defining form explains the conclusion; just saying “obviously even” does not.

Watch out

Do not begin by assuming the result you are trying to prove. “The sum is 2m2m, so it is even” has not shown why the given odd integers have that sum. Start from their forms and derive 2m2m.

Change one of the two factors from odd to even in this next argument.

Exhaustive finite cases

Exhaustion proves a claim by covering every possible case, rather than by choosing a large sample.

If the allowed set contains four numbers, four checks can be enough. If the set is infinite, it may still fall into finitely many algebraic cases: every integer has remainder 00, 11 or 22 on division by 33.

Proof by exhaustion

A proof by exhaustion identifies a complete list of cases, establishes the claim in each case, and concludes that the claim holds for all allowed values. Cases may be individual values or classes represented algebraically.

The restrictions in the question determine what “all” means.

Worked example 3: a finite list

Question. Prove by exhaustion that n2−1n^2-1 is a multiple of 33 for every n∈{1,2,4,5}n\in\{1,2,4,5\}.

Plan. Evaluate the expression for all four allowed values.

n1245n2−1031524(n2−1)/30158\begin{array}{c|rrrr} n&1&2&4&5\\\hline n^2-1&0&3&15&24\\ (n^2-1)/3&0&1&5&8 \end{array}

Every quotient is an integer, including 00.

Answer. All four allowed values have been checked, so the statement is true on the given set.

Check. The list contains exactly the four members of the stated set. In particular, 0=3×00=3\times0 is a multiple of 33.

Now change only the domain: allow every positive integer that is not a multiple of 33. We cannot list them individually, but two remainder classes exhaust them.

Worked example 4: an infinite set, two cases

Question. Prove by exhaustion that n2−1n^2-1 is divisible by 33 for every positive integer nn that is not a multiple of 33.

Plan. Use n=3k+1n=3k+1 and n=3k+2n=3k+2, where kk is a nonnegative integer.

Case 1:

(3k+1)2−1=9k2+6k=3(3k2+2k).(3k+1)^2-1=9k^2+6k=3(3k^2+2k).

Case 2:

(3k+2)2−1=9k2+12k+3=3(3k2+4k+1).(3k+2)^2-1=9k^2+12k+3=3(3k^2+4k+1).

In each case the bracket is an integer, so the expression is divisible by 33.

Answer. These two cases cover every positive integer not divisible by 33; hence the statement is true for all such integers.

Check. Division by 33 permits only remainders 0,1,20,1,2; the assumption excludes 00. Both remaining cases appear, including n=1,2n=1,2 when k=0k=0.

The completeness of the cases is as important as the algebra inside them.

Exam tip

Use a final sentence connecting the cases to the whole allowed set. The forms 3k−13k-1 and 3k+23k+2 describe the same remainder class, so using both would miss the remainder-11 case.

Keep the same allowed values and change the expression.

One counterexample disproves a universal claim

A universal claim says that something holds for every allowed value; one allowed failure is enough to disprove it.

The expression n2+n+11n^2+n+11 gives primes for n=1,2,3n=1,2,3: 13,17,2313,17,23. That pattern invites a conjecture, but it supplies no guarantee for later values.

Counterexample

A counterexample satisfies every assumption of a claim but makes its conclusion false. It disproves a universal statement; an example that satisfies the conclusion cannot establish that universal statement.

When a formula contains a fixed prime, try values that make its other terms a multiple of that prime.

Worked example 5: a composite value

Question. Disprove the claim “n2+n+11n^2+n+11 is prime for every positive integer nn”.

Plan. Try n=10n=10, so that n+1=11n+1=11 and n(n+1)n(n+1) contains a factor of 1111.

102+10+11=121=11×11.10^2+10+11=121=11\times11.

Since 121121 has factors other than 11 and itself, it is not prime.

Answer. The allowed value n=10n=10 is a counterexample, so the claim is false.

Check. 1010 is a positive integer; both factors 1111 are strictly between 11 and 121121. Both the assumption and the failure have been established.

Change just the constant from 1111 to 1717. The same search strategy still works, even though the first few outputs again look promising.

Worked example 6: the same search strategy

Question. Disprove the claim “n2+n+17n^2+n+17 is prime for every positive integer nn”.

Plan. Choose n=16n=16, making n+1=17n+1=17.

162+16+17=289=17×17.16^2+16+17=289=17\times17.

Answer. n=16n=16 is permitted and produces the composite number 289289, so the universal claim is false.

Check. Independently, n(n+1)+17=16×17+17=17(16+1)=172n(n+1)+17=16\times17+17=17(16+1)=17^2, agreeing with the expanded calculation.

Assumptions matter in non-integer examples too. To disprove a claim about products of distinct irrational numbers, both inputs must be distinct and irrational. For example, 3×12=6\sqrt3\times\sqrt{12}=6 meets those assumptions and has a rational product; 12×2=1\frac12\times2=1 does not meet them.

Watch out

A counterexample does not prove the opposite statement for every value. Finding one composite output disproves “every output is prime”; it does not show “every output is composite”.

Apply the same strategy with a new constant.

Quick Recap

  • A proof derives a conclusion from the assumptions using justified steps.
  • Exhaustion must cover the whole allowed set, possibly through finitely many algebraic classes.
  • A counterexample must obey every assumption and break the conclusion.

A hundred successful checks do not prove a claim about all integers. One permitted failure can disprove it.