Loav · VCE Specialist Mathematics · Units 3 & 4
S34-P02 · Discrete mathematics · Induction
1. Why induction proves an infinite family
A statement P(n) may depend on an integer index n: a sum formula, a divisibility claim or an inequality. Mathematical induction proves P(n) for every integer n from a specified starting value n₀. It needs two logically different ingredients. First, establish P(n₀) directly. Second, prove that whenever P(k) is true for an arbitrary integer k ≥ n₀, P(k + 1) must also be true. The first ingredient starts the argument; the second carries it forward without an upper limit.
The inductive assumption is conditional. You are not claiming the entire theorem is already true. You temporarily assume just P(k) and use that assumption to derive the next case. The implication P(k) ⇒ P(k + 1) is proved for arbitrary k. Starting from the independently verified base case, repeated use establishes every later integer case. A step that jumps from k to k + 2 would leave every second case uncovered unless another starting case were supplied.
Both pieces are indispensable. A correct base case with no step proves only one case. A valid step without a true base case proves only that truth would propagate if it ever began. The range also matters. If the first true case is n = 4, do not conclude the result holds for every positive integer. State “for every integer n ≥ 4” at the start and at the end.
2. Read a partial sum or product precisely
The notation Σr=1n ar means a₁ + a₂ + ⋯ + an. The index r runs through the terms; n determines where the finite sum stops. To move from n = k to n = k + 1, keep the first k terms and add ak+1. Do not add ak again. Writing the new final term explicitly is often the most important line in a sum proof.
Similarly, Πr=1n br means b₁b₂⋯bn. The next product is the old product multiplied by bk+1. State any values that would make a denominator zero. Factorials are an example: (k + 1)! = (k + 1)k!, because factorial notation denotes a finite product. It does not distribute over addition.
When searching for a formula, compute several exact terms and look for structure. A table might suggest a quadratic expression for a sum or a rational expression for a product. This is conjecture formation, not proof. Once a candidate F(n) is found, the induction step often reduces to checking that F(k) + ak+1 = F(k + 1), or its product counterpart. These identities explain why the formula matches the recurrence.
Worked example 1: an arithmetic sum
Prove that Σr=1n(4r − 1) = n(2n + 1) for every integer n ≥ 1. For n = 1, the left side is 3 and the right side is 1 × 3 = 3, so the base case holds. Suppose, for an arbitrary integer k ≥ 1, that Σr=1k(4r − 1) = k(2k + 1).
The next term is 4(k + 1) − 1 = 4k + 3. Hence the sum through k + 1 equals k(2k + 1) + 4k + 3 = 2k² + 5k + 3 = (k + 1)(2k + 3). Since 2k + 3 = 2(k + 1) + 1, this is exactly the required formula with n replaced by k + 1. Therefore, by induction, the identity holds for every integer n ≥ 1.
Notice where the hypothesis entered: it replaced the sum of the first k terms. The rest was algebra. Simply writing down the target formula for k + 1 would assume the result rather than derive it.
3. Telescoping explains cancellation directly
A telescoping sum has terms that cancel after each summand is expressed as a difference. For example, 1/[(r + 1)(r + 2)] = 1/(r + 1) − 1/(r + 2). This identity follows by putting the two fractions over a common denominator. Summing from r = 1 to n leaves 1/2 − 1/(n + 2), because every middle reciprocal appears once positively and once negatively.
= 1/2 − 1/(n + 2) = n/[2(n + 2)].
The endpoints depend on the exact starting and finishing indices. Write the first two terms and the final term before cancelling. A change in lower limit changes the surviving first contribution. Telescoping is a direct proof for arbitrary n; induction can verify the same identity through its recurrence. If a question specifies induction, presenting only cancellation does not meet the requested method.
Products can telescope too. In Πr=1n(r + 2)/(r + 1), factors 3 through n + 1 cancel, leaving (n + 2)/2. Cancellation here removes common nonzero factors in a product. It does not license cancelling separate terms across an addition sign. Expanding several factors makes the structure and the nonzero condition visible.
Worked example 2: divisibility by induction
Prove that 8n − 1 is divisible by 7 for every integer n ≥ 1. At n = 1, the expression is 7, so the base case holds. Suppose that 8k − 1 = 7m for some integer m, with k ≥ 1. Rewrite the next expression to expose the assumed one:
= 8 × 7m + 7 = 7(8m + 1).
Since 8m + 1 is an integer, the expression at k + 1 is divisible by 7. The base case and inductive step establish the claim. The added 7 is essential: 8(8k − 1) equals 8k+1 − 8, not 8k+1 − 1. Expanding a proposed rearrangement is a quick way to detect a missing adjustment.
Worked example 3: preserve the inequality direction
Prove that 3n ≥ 2n + 1 for every integer n ≥ 0. At n = 0, both sides equal 1. Suppose 3k ≥ 2k + 1 for an arbitrary integer k ≥ 0. Multiplying by the positive number 3 preserves the inequality, giving 3k+1 ≥ 6k + 3.
The target is 2(k + 1) + 1 = 2k + 3. Since (6k + 3) − (2k + 3) = 4k ≥ 0, we have 6k + 3 ≥ 2k + 3. Chaining the inequalities proves the next case. Thus the result holds for every integer n ≥ 0.
The induction step needs a comparison with the exact target, not just a larger-looking expression. Equality occurs for n = 0 and n = 1. For n ≥ 2, the step uses k ≥ 1, so the second comparison is strict and the result is strict. The proof therefore explains where equality can survive.
4. Use an algorithm to investigate, then prove
A finite loop can compare a recurrence with a candidate closed form. Here a running total tests the arithmetic-sum identity from the first example. The total is initialised once, and the loop adds exactly one new term on each pass. Equality is checked with integer arithmetic, avoiding a false mismatch caused by decimal rounding. The output records the first failed case if one is found.
INPUT positive integer N
S ← 0
FOR n FROM 1 TO N
S ← S + 4 × n − 1
IF S ≠ n × (2 × n + 1) THEN
OUTPUT n, S, "counterexample"
STOP
END IF
END FOR
OUTPUT "No counterexample among the tested cases"
The assignment arrows, indented selection and inclusive counting loop follow the VCAA pseudocode conventions. FROM 1 TO N includes both endpoints. STOP means exit the entire algorithm, so the final message is reached only after every requested case passes.
The loop is useful for checking a conjecture or finding an implementation error. It is not an induction proof because it stops at N. To trace it, record n, the newly added term, S and the proposed formula. After the first three passes, S is 3, 10 and 21. If S were reset to zero inside the loop, the program would test individual terms instead of partial sums. Algorithm design therefore requires the same attention to quantities and indices as the proof.
When technology is allowed, a symbolic system can simplify the difference between the derived k + 1 expression and the target. A zero result checks an algebraic identity, but the mathematical write-up must still state the inductive assumption and explain why it applies. For inequalities, a system may need assumptions such as k ≥ 0; omitting them can make a correct comparison undecidable or invalid over a larger domain.
5. Avoid the three most common gaps
First, verify the base case numerically on both sides. “Clearly true” can hide a wrong starting index. Second, state the hypothesis for k, rather than writing the whole theorem for all n as an assumption. Third, end the step with the expression required at k + 1. A page of algebra that stops one line early may not show the necessary connection.
For a divisibility claim, show the final multiplier is an integer. For an inequality, name the sign condition that preserves or reverses its direction. For a product, include the new factor rather than a new summand. If a conjecture fails at one permitted value, stop trying to prove it and revise the statement. A valid counterexample is decisive even when a thousand earlier tests succeeded.
Practise, then check
Where induction is requested, include the base case, an arbitrary-index hypothesis, the next-case derivation and the final domain.
- Use induction to prove Σr=1n(6r − 2) = n(3n + 1) for integers n ≥ 1.
- Find Σr=1n2/[(r + 2)(r + 3)] by telescoping.
- Use induction to prove that 5n − 1 is divisible by 4 for every integer n ≥ 1.
- Prove by induction that 2n ≥ n + 1 for integers n ≥ 0. When is equality possible?
- Evaluate Πr=1n(r + 3)/(r + 2), then verify the formula by induction.
- A proposed proof assumes P(k + 1), rearranges it to P(k), and cites the inductive hypothesis. Explain what must be checked, and give the safer presentation.
- A program tests a claim for n = 1 through 10 000 without failure. What has it established? For the sum-testing algorithm above, give S after the fourth pass.
Reasoned answers
- At n = 1, both sides are 4. Assume the sum through k is k(3k + 1). Adding 6(k + 1) − 2 = 6k + 4 gives 3k² + 7k + 4 = (k + 1)(3k + 4), which equals (k + 1)[3(k + 1) + 1]. This proves the step and hence the identity for every integer n ≥ 1.
- Each summand is 2/(r + 2) − 2/(r + 3). Thus the middle terms cancel, leaving 2/3 − 2/(n + 3) = 2n/[3(n + 3)]. The lower endpoint supplies 2/3, not 2 or 1. Every denominator is nonzero for the stated positive indices.
- The base case is 5 − 1 = 4. If 5k − 1 = 4m for an integer m, then 5k+1 − 1 = 5(5k − 1) + 4 = 4(5m + 1). The multiplier is an integer. This establishes the next case and therefore all integers n ≥ 1.
- At n = 0, equality holds. Assuming 2k ≥ k + 1, multiplication by 2 gives 2k+1 ≥ 2k + 2 ≥ k + 2 because k ≥ 0. This is the target at k + 1. Equality holds at n = 0 and 1; for n ≥ 2 the second comparison is strict, so equality is impossible.
- The product is (n + 3)/3. For n = 1 this gives 4/3, matching the single factor. If the product through k equals (k + 3)/3, multiply by the next factor (k + 4)/(k + 3), obtaining (k + 4)/3. This is the formula at k + 1. The cancelled factor is positive for k ≥ 1.
- Deriving an established statement from the target does not usually prove the target. The reasoning is valid only if every step is reversible under the stated conditions. A safer final proof begins with P(k), performs the justified transformations in the forward direction and derives P(k + 1). This also makes any division or inequality restriction visible.
- It has established agreement only for the tested 10 000 values, assuming the implementation is correct. It has not proved the universal statement. In the sum algorithm, the fourth added term is 15, so S = 21 + 15 = 36. The candidate gives 4(2 × 4 + 1) = 36, confirming that test case.
Curriculum and sources
VCE Specialist Mathematics Units 3 and 4. Scope follows the VCAA Mathematics Study Design 2023–2027, Specialist Mathematics Areas of Study 1 and 3 as applicable, and the current Specialist Mathematics FAQ. Unit 2 complex-number foundations are retained where required. Curriculum checked 15 October 2026. Teaching reference: Cambridge Senior Mathematics VCE Specialist Mathematics 3&4, second edition (2023), sections 2D–2E, pp. 82–95; 1C, pp. 19–28. All explanations, examples, practice questions and diagrams are original. Independent teaching material; not a VCAA publication.