Did you know? You can sometimes prove that a mathematical object exists by imagining the smallest possible counterexample. If you can always turn that counterexample into an even smaller one, you have reached a contradiction: the supposedly smallest counterexample could not have been smallest after all.
For students and parents researching a JC2 H3 Mathematics tutor in Bukit Timah, this is a particularly useful non-routine problem-solving pattern. It is often called the extremal principle: choose a largest, smallest, longest, shortest or otherwise extreme object, and use that choice to force structure. A related technique, infinite descent, rules out positive-integer solutions by proving that any such solution would generate an endless chain of smaller positive-integer solutions.
Quick answer: To use the extremal principle, define a finite or well-ordered collection, justify the existence of an extreme member, and show what its extremality implies. To use a minimal counterexample, assume the claim is false, select a counterexample with smallest positive integer parameter, and derive an even smaller counterexample. The final contradiction establishes the theorem.
The 2027 H3 Mathematics 9820 syllabus explicitly includes contradiction, induction, existence, combinatorial proofs and uncovering structure. ‘Extremal principle’ and ‘infinite descent’ are useful enrichment names for proof techniques, not a claim that they form separately listed examination chapters. The worked examples here develop precisely the rigorous reasoning and transfer that the syllabus describes.
Find the part that helps
- A one-minute reasoning check
- What an extreme choice adds
- Every integer has prime factors: full proof
- Infinite descent in a Diophantine equation
- Longest paths and cycles
- An unexpected maximum-degree theorem
- Extremal problems in inequalities
- Compare contradiction, induction and descent
- Thirty original learning clinics
- Logical pitfalls
- Tutor and home practice
- Frequently asked questions
- Official references
A one-minute diagnostic: ‘Choose the smallest one’
Ask a JC2 learner to justify why a nonempty finite list of real numbers has a smallest entry. If the list is finite, comparing its finitely many entries selects a minimum. But the set of all positive real numbers has no smallest member: for any x>0, the number x/2 is another positive real smaller than x.
This matters because an extremal proof cannot simply begin ‘choose a smallest object’ whenever the writer wishes. The set being minimised must genuinely possess that minimum. Nonempty sets of positive integers do possess a least element; general infinite sets of real numbers may not.
Next ask: what would prove the claim ‘every integer n≥2 is prime or can be factored into primes’? A few examples—6=2×3, 12=2×2×3—cannot justify every n. A minimal-counterexample argument can: if the least counterexample is composite, both its proper factors are smaller, and the factorizations of those smaller factors combine to factor the original.
The diagnosis separates proof architecture from arithmetic fluency. A student who recognises the intended contradiction is ready to write a complete general proof; a student who only keeps trying values needs to learn how a well-founded choice replaces enumeration.
A strong explanation explicitly names the collection, its domain, why a minimum or maximum exists, and which smaller or larger object contradicts the choice. This four-part structure is reusable across number theory, counting, algorithms and graph puzzles.
The extremal principle is a choice plus a consequence
An extremal argument begins by choosing an object at one end of an ordering. In a finite collection of numbers you can choose the smallest or largest. In a finite graph you can choose a longest simple path. Among finitely many arrangements you can select one with the fewest conflicts. The choice is guaranteed because the collection is finite.
The second step is to exploit the extreme condition. If your construction creates an object in the same allowed collection with a smaller score, you contradict minimality. If it creates a valid longer path, you contradict maximality. The new object must satisfy the original restrictions: an illegal improvement proves nothing.
Suppose we have a finite group of students with a set of scores. Choosing a highest-scoring student does not mean the highest score is unique. Another student might share it. ‘Choose a maximum’ means the score is at least as high as every other score, not that the maximiser is necessarily the only one.
Now suppose an argument seeks to replace a candidate with a better one. If the replacement is outside the feasible set, the proposed contradiction fails. In geometry, a shorter path might leave the permitted region; in an integer equation, a smaller real solution is not necessarily an integer solution.
The extremal viewpoint is a guide to the right representation. A problem may appear to have too many possibilities, but selecting one with minimum size, minimum counterexamples, maximum degree or shortest length can impose an equation or inequality that reduces the problem.
One must also distinguish a maximum from a supremum. The open interval (0,1) has no largest member, even though its least upper bound is one. Proofs involving real-valued functions need continuity and compactness or some other justification when asserting that an optimum is attained.
In H3 Mathematics, precise qualification often determines the validity of the argument. State whether the domain is finite, positive integers, closed intervals or something else. Once the selection is legitimate, explain what property it forces.
Full proof: every integer n≥2 is a product of primes
The statement. For every integer n≥2, there exists a finite list of prime numbers whose product equals n. A prime number itself counts as a product of one prime.
Assume the contrary. Suppose at least one integer n≥2 cannot be expressed as a product of primes. Let C be the collection of all such integers. This is a nonempty set of positive integers, so it has a least member m by the well-ordering principle.
Use minimality. The number m cannot be prime, because a prime is already a product of primes consisting of itself. Therefore m is composite: m=ab for integers a,b satisfying 2≤a<m and 2≤b<m.
Apply the smaller-case information. Since m was the least integer without a prime factorisation, both smaller integers a and b have prime factorizations. Multiply those two lists of prime factors. Their product is ab=m, giving a prime factorisation of m.
Contradiction and conclusion. We constructed a prime factorisation of an integer chosen precisely because it had none. Therefore the set C is empty, and every integer n≥2 is a product of primes.
The proof is short because the minimal counterexample supplies information about all smaller eligible integers. This is closely related to strong induction: the induction hypothesis would assert the theorem for every integer below m, and the composite factorisation would establish it for m.
For a tutoring diagnostic, ask a student why m=ab has factors smaller than m. Because a,b≥2 and their product is m, each is at most m/2 and therefore strictly below m. Also ask why ‘m is prime’ was handled before the composite case. Without that split, the argument would not cover every integer n≥2.
This theorem establishes existence of prime factorisation, not uniqueness. Unique factorisation requires additional reasoning, commonly through a divisibility lemma about primes. Conflating existence and uniqueness is an avoidable advanced Mathematics mistake.
Infinite descent: a positive-integer solution would create a smaller one
Now investigate the equation x²+y²=3z² in positive integers x,y,z. We will show that no such positive-integer triple exists. The method is infinite descent: if one solution existed, we would construct a strictly smaller positive solution, contradicting a minimal choice.
Suppose at least one positive-integer solution exists. Choose one with the smallest possible positive value of x+y+z. Such a minimum exists because the sums of positive-integer triples are positive integers.
Every integer square is congruent to zero or one modulo three. The equation gives x²+y²≡0 (mod 3). The only way two residues from {0,1} can add to zero modulo three is 0+0. Hence three divides both x and y.
Write x=3a and y=3b, where a and b are positive integers. Substitution gives 9(a²+b²)=3z², or z²=3(a²+b²). Thus z² is divisible by three, and because a square has residue zero modulo three exactly when its integer base is divisible by three, three divides z as well.
Write z=3c. Substitution back into the original equation yields 9a²+9b²=27c², so a²+b²=3c². Thus (a,b,c) is another positive-integer solution of exactly the same equation.
But a+b+c=(x+y+z)/3, strictly smaller than x+y+z. This contradicts the choice of a solution with smallest coordinate sum. Hence the assumed solution cannot exist.
Notice the crucial structure: divisibility of a solution forces every coordinate to share a factor of three, so dividing produces a strictly smaller solution of the same kind. This is stronger than simply observing that several tried triples failed.
The proof also handles the correct domain. The triple (0,0,0) does satisfy the equation if zero is allowed, so the statement cannot be ‘no integer solution whatsoever’. We proved there is no positive-integer solution, which is exactly what the descent assumption used.
If a purported descent only decreases a real-valued coordinate but does not ensure a new positive integer solution, the well-ordering contradiction may fail. The output has to remain inside the same permitted collection and genuinely become smaller.
Graph theory application: a longest path reveals a cycle
Consider a finite undirected simple graph in which every vertex has degree at least two. We will prove that the graph contains a cycle. A graph is a collection of vertices connected by edges; the degree of a vertex counts its neighbours.
Choose a simple path with the greatest possible number of vertices, say v₀,v₁,…,vₖ. ‘Simple’ means no vertex repeats. A longest simple path exists because the graph has finitely many vertices.
Look at the final vertex vₖ. If it had a neighbour outside the chosen path, we could append that neighbour and obtain a longer simple path, contradicting maximality. Therefore every neighbour of vₖ already lies on the path.
By assumption vₖ has at least two distinct neighbours. One is vₖ₋₁, the preceding vertex in the path. Another must be some earlier vⱼ with j≤k−2. Follow the path from vⱼ to vₖ and then take the edge from vₖ back to vⱼ.
This produces a cycle with at least three distinct vertices. The proof did not need to inspect every possible graph: a maximal path creates enough structure at its endpoint to force a cycle.
The finite hypothesis matters. Choosing a longest simple path in an arbitrary infinite graph is not always possible; there may be paths of unbounded finite length with no longest one. A rigorous H3 solution states the assumption that guarantees the extreme object exists.
This graph argument is enrichment illustrating the syllabus’s proof and problem-solving capabilities, not an assertion that formal graph theory is separately prescribed H3 content. It is particularly useful for learning to justify ‘choose a longest path’ instead of treating it as an unquestioned shortcut.
A parent or tutor can adapt the argument using dots and lines on paper. Ask the learner why a neighbour outside the path would contradict maximality, and why degree at least two supplies an earlier neighbour. These two questions pinpoint the actual logical bridge.
A surprising maximum-degree argument: a king in a tournament
A tournament is a finite set of vertices where each pair of distinct vertices is joined by exactly one directed edge, pointing in one direction. Think of a competition in which every pair of players plays once and one beats the other.
Choose a vertex v with the greatest number of outgoing edges; write that number as d. We claim that every other vertex w can be reached from v by a directed path of length at most two. Such a vertex is sometimes called a ‘king’.
If v beats w directly, the claim holds with a one-edge path. So consider a vertex w that v does not beat. The tournament must have an edge w→v.
Suppose, seeking a contradiction, there were no two-edge path v→u→w. Then for every vertex u that v beats, the edge between u and w would have to point w→u; otherwise v→u→w would provide the desired path.
Thus w would beat v and every one of the d vertices beaten by v. It would have at least d+1 outgoing edges, contradicting the choice of v as a vertex of maximum out-degree d.
Therefore some two-edge path from v to w must exist. Since w was arbitrary, every other vertex is reachable from v in at most two directed moves. The proof hinges on the extremal degree choice and a contradiction if a required short path is absent.
This is a valuable pattern for non-routine problems: pick an object that is best according to an obvious score, then show that an apparent failure of the desired property would produce a different object with a strictly better score.
For parents assessing an explanation, ask the student to identify the pairwise direction argument. The mathematical mechanism is not that maximum-degree vertices are ‘powerful’ in an informal sense; it is the specific count d+1 that violates maximality.
Extremal inequalities and the equality condition
Consider positive real numbers a and b with a+b=20. We wish to maximise the product ab. The AM–GM inequality gives √(ab)≤(a+b)/2=10, hence ab≤100. Equality occurs when a=b=10.
This calculation identifies the maximiser because it checks equality as well as the bound. Merely reporting ab≤100 would establish an upper bound without proving that some permitted pair attains it.
Now consider a finite collection of points on a circle and choose two with the greatest distance. Their distance cannot exceed the diameter because every chord is at most the circle’s diameter. Equality is possible when the pair consists of opposite points, if both are in the given collection. If not, the maximum distance among the provided points is smaller. The feasible set matters.
A function can have a supremum that is not attained. For instance, the function f(x)=x on 0<x<1 has values arbitrarily close to one, but no permitted x gives f(x)=1. Therefore one cannot start a proof by choosing the ‘largest x’ from that domain.
Contrast this with f continuous on a closed and bounded interval [a,b], where the extreme value theorem guarantees a maximum and a minimum. If a calculus optimisation problem uses that theorem, say which property justifies the extreme point.
A common proof technique selects a minimiser among finitely many candidate arrangements, then performs a local exchange. If the exchange strictly improves the objective without violating constraints, the supposedly optimal arrangement was not optimal. This exchange argument powers many counting and optimisation proofs.
Be careful with local versus global improvement. Showing that one swap improves a particular arrangement does not automatically locate the global optimum. An extremal proof works when it applies to an actual global minimiser from a suitable feasible set and establishes the promised structural contradiction.
Minimal counterexample, induction and contradiction: related but different
Ordinary induction proves a base case P(n₀) and the implication P(k)⇒P(k+1) for every permitted k. Strong induction allows the inductive step to use all earlier cases P(n₀),…,P(k). A minimal-counterexample proof assumes the theorem fails and chooses the least failing integer.
For statements about all integers above a starting value, strong induction and minimal-counterexample reasoning are closely related through the well-ordering principle. Often one style is clearer than another depending on whether the proof naturally uses one earlier case or several smaller cases.
Proof by contradiction is broader. You assume the negation and derive an impossibility; it need not involve a minimum. Infinite descent is a particular contradiction structure where each alleged positive-integer solution forces a smaller positive-integer solution.
When a student chooses a minimal counterexample, the crucial question is whether the constructed smaller object is also a counterexample. Simply finding a smaller number that is not relevant to the claim does not challenge minimality.
In an extremal graph proof, by contrast, the extreme object may be a longest path rather than a counterexample to a theorem. The contradiction comes from extending that path, not necessarily from creating a smaller false case.
All these approaches demand precise domains, justified existence of the chosen extreme object and an explicit contradiction. The techniques are not magic words: their validity rests on the details of the construction.
Thirty original proof and investigation clinics
1. 1. The least counterexample
Question: Suppose P(n) is claimed for all integers n≥2 and the claim is false. What can be chosen?
Worked reasoning: The set of failing integers is nonempty and contained in the positive integers, so it has a least member m. Any eligible integer smaller than m must satisfy P.
Takeaway: This is the justification for minimal-counterexample reasoning; it is not a license to choose a minimum from arbitrary real sets.
2. 2. A false minimum claim
Question: Does the set {1/2,1/3,1/4,…} have a smallest element?
Worked reasoning: No. For each positive integer n, 1/(n+1)<1/n. Every candidate has a smaller member, though zero is a lower bound.
Takeaway: Distinguish lower bounds and infima from minima that are attained inside the set.
3. 3. A finite maximum
Question: A finite nonempty list has entries 4,−2,9,9,1. What is its maximum?
Worked reasoning: The maximum value is nine, attained twice. A maximum element need not be unique even though the maximum numerical value is well defined.
Takeaway: A proof may choose one maximiser without assuming all maximisers have identical identities.
4. 4. Prime decomposition
Question: Why is the least integer without a prime product necessarily composite?
Worked reasoning: If it were prime, it would itself be a product of one prime. That contradicts membership in the counterexample set, so it must be composite.
Takeaway: Handle mutually exclusive cases before invoking factorisation.
5. 5. Smaller proper factors
Question: If m=ab for integers a,b≥2, why are a,b<m?
Worked reasoning: Since the other factor is at least two, each factor is at most m/2. Thus both factors are strictly smaller than m.
Takeaway: A minimal-counterexample proof needs an explicit smaller parameter, not a vague word such as ‘simpler’.
6. 6. Strong induction reformulation
Question: What strong-induction statement matches the prime-factorisation proof?
Worked reasoning: Assume every integer 2≤j<m has a prime factorisation. For composite m=ab, apply the hypothesis to both factors and combine them.
Takeaway: Recognise when a minimal-counterexample proof is a concise form of strong induction.
7. 7. A coin representation
Question: Prove every integer n≥2 can be written as 2a+3b with nonnegative integers a,b.
Worked reasoning: Check n=2 and n=3. For n≥4, the integer n−2≥2 has such a representation by strong induction; adding one to a represents n.
Takeaway: This has separate base cases because the inductive construction reduces the target by two.
8. 8. Restricted domain
Question: Why is ‘every positive integer is 2a+3b’ false?
Worked reasoning: For n=1 there are no nonnegative integers a,b with 2a+3b=1. The corrected statement begins at n≥2.
Takeaway: An extreme proof must honour the smallest permissible input and the exact theorem being claimed.
9. 9. Descent mod three
Question: For integer t, what are the possible residues of t² modulo three?
Worked reasoning: Only zero and one. The residues 0,1,2 square to 0,1,1 modulo three.
Takeaway: A complete residue list is often the key structural step in an integer descent argument.
10. 10. Divisibility of two squares
Question: If x²+y²≡0 (mod 3), what follows about x and y?
Worked reasoning: Each squared residue is zero or one. Their sum can be zero modulo three only if both are zero, so both x and y are multiples of three.
Takeaway: The argument covers all residue cases and does not simply infer a factor from one observed example.
11. 11. A descent proof condition
Question: Why does the equation x²+y²=3z² have no positive-integer solution?
Worked reasoning: Any such triple has all coordinates divisible by three. Dividing the coordinates by three produces another positive solution with a smaller coordinate sum, contradicting minimality.
Takeaway: The new triple stays in the same positive-integer domain and satisfies the same equation.
12. 12. Zero as an exception
Question: Does (0,0,0) satisfy x²+y²=3z²?
Worked reasoning: Yes, both sides are zero. Therefore ‘there are no integer solutions’ would be false; the valid claim rules out positive-integer solutions.
Takeaway: An explicit domain is indispensable in contradiction and descent proofs.
13. 13. Longest simple path
Question: In a finite graph with minimum degree two, what prevents a longest simple path’s final vertex having a new neighbour?
Worked reasoning: Such an outside neighbour would extend the path by one vertex, contradicting the chosen maximum length.
Takeaway: The extension must remain simple and use a legitimate graph edge.
14. 14. Force a cycle
Question: Why does a longest path in a finite graph of minimum degree two reveal a cycle?
Worked reasoning: Its endpoint has at least two neighbours already on the path, one of which is earlier than the preceding vertex. That earlier edge closes a cycle.
Takeaway: An extreme endpoint becomes informative only after applying the degree constraint.
15. 15. Finite graph hypothesis
Question: Why does the ‘longest path’ argument need care in an infinite graph?
Worked reasoning: A graph with arbitrarily long finite simple paths need not possess a longest finite simple path. The maximiser used in the proof may not exist.
Takeaway: Never claim an extreme exists without a finiteness or attainment argument.
16. 16. Tournament choice
Question: Which vertex should be selected in the directed tournament proof?
Worked reasoning: Choose a vertex with maximum out-degree. Such a vertex exists because the tournament has finitely many vertices.
Takeaway: The chosen score predicts what a counterexample must violate.
17. 17. Tournament contradiction
Question: If a vertex w beats the chosen maximum-outdegree vertex v and every vertex beaten by v, what follows?
Worked reasoning: Then w beats at least d(v)+1 vertices, contradicting maximality of d(v). Thus there must be a two-edge path from v to w.
Takeaway: Count directed edges precisely rather than relying on informal dominance language.
18. 18. A max-degree tie
Question: Does the tournament proof require the maximum-outdegree vertex to be unique?
Worked reasoning: No. Any vertex whose out-degree is maximal suffices. A tie causes no difficulty because the contradiction would produce strictly more than that maximum.
Takeaway: Optimality and uniqueness are separate properties.
19. 19. Rectangle area
Question: For positive a,b with a+b=20, what is the greatest possible ab?
Worked reasoning: AM–GM gives ab≤100, and equality at a=b=10 shows that 100 is attained.
Takeaway: A bound becomes an actual maximum only when a permitted equality case exists.
20. 20. An open-interval extremum
Question: Does f(x)=x on the interval 0<x<1 have a maximum?
Worked reasoning: No. For every permitted x, the number (x+1)/2 is also permitted and larger. One is the supremum but not a value attained.
Takeaway: A supposed greatest element can be contradicted by a valid improvement inside the same domain.
21. 21. Maximising a product
Question: For nonnegative integers a+b=8, which pair maximises ab?
Worked reasoning: The integer pairs closest to equality give a=b=4 and product sixteen. Algebraically ab=16−(a−4)²≤16.
Takeaway: Rewriting around the expected extremum supplies both a global bound and its equality case.
22. 22. Minimal unsorted arrangement
Question: Suppose an arrangement has an out-of-order adjacent pair and swapping it lowers an inversion score. What could a minimiser look like?
Worked reasoning: A true minimum-inversion arrangement cannot contain such a pair, since swapping would create a feasible arrangement with lower score.
Takeaway: This is an exchange argument: optimal arrangements cannot admit a legal strictly improving local move.
23. 23. A bad exchange
Question: Why does a move that shortens a path but leaves the permitted region fail as an extremal contradiction?
Worked reasoning: The improved path is not an element of the feasible set whose minimum was chosen. It therefore says nothing about that minimum.
Takeaway: Always check that the replacement object satisfies all constraints.
24. 24. A positive-integer descent
Question: A supposed smallest positive integer m has a required property forcing m=2k with k also possessing it. What would contradict minimality?
Worked reasoning: If k is a positive integer and k<m, then k is a smaller member of the same solution set, impossible by the choice of m.
Takeaway: The smaller object must satisfy the same defining condition; being merely smaller is insufficient.
25. 25. Descent on a sum
Question: Why choose x+y+z rather than x alone for a triple solution?
Worked reasoning: If all three coordinates are reduced by a common factor, the positive integer sum definitely decreases. It supplies a simple well-ordered measure.
Takeaway: A good descent measure should respond predictably to the proposed transformation.
26. 26. Compare two proofs
Question: Can prime factorisation be shown by ordinary induction using only the case n−1?
Worked reasoning: Not directly through the usual composite-factor split; the factors of n may be many smaller integers. Strong induction or minimal-counterexample logic matches the argument naturally.
Takeaway: Choose a proof structure that matches what the next step actually uses.
27. 27. An invariant and an extreme
Question: In repeated subtraction gcd, what stays fixed and what decreases?
Worked reasoning: The greatest common divisor is invariant, while the positive integer sum of the pair strictly decreases until equality.
Takeaway: Many sophisticated proofs combine a preserved quantity with an extremal or termination argument.
28. 28. A nonunique shortest route
Question: Could a finite network have two different shortest paths with equal length?
Worked reasoning: Yes. A square network has two equal-length ways to travel between opposite corners. Choosing a shortest path does not imply it is unique.
Takeaway: Extremal selection does not automatically establish uniqueness.
29. 29. A bound not attained
Question: The inequality x²>0 holds on 0<x<1. Does x² have a positive minimum there?
Worked reasoning: No. Values x² approach zero as x approaches zero from within the interval, but x=0 is excluded.
Takeaway: An infimum may exist without a minimiser, even for a simple continuous formula on an open domain.
30. 30. Fix the first broken claim
Question: A student’s proof says, ‘Choose the smallest positive real solution.’ What must be checked first?
Worked reasoning: Determine whether the solution set is nonempty and has an attained minimum. Positive reals alone are not well ordered, so the selection may be unjustified.
Takeaway: Proof criticism should locate the earliest unsupported extremal choice rather than merely rejecting the final answer.
Six traps in extremal arguments
First, failing to prove that a minimum or maximum exists. A finite nonempty collection gives an attained extreme. A nonempty set of positive integers has a least member. An arbitrary nonempty subset of the positive reals need not have a minimum.
Second, selecting an extreme from the wrong collection. If the theorem is about simple paths, the replacement must remain simple. If the theorem concerns positive integers, the new counterexample must remain a positive integer. If the problem restricts a variable to an open interval, a proposed maximiser at its endpoint may not be admissible.
Third, creating an object that is smaller but not a counterexample. Minimal-counterexample reasoning does not rule out smaller ordinary examples; it rules out smaller members of the counterexample set. Restate which property the replacement satisfies before claiming a contradiction.
Fourth, proving a number is an upper bound but not that it is a maximum. To identify the maximum, show that a legal object attains the bound. The same caution applies to minima and equality conditions.
Fifth, confusing the direction of induction. An ordinary induction step from P(k) to P(k+1) can use only the stated hypothesis. If the argument depends on all smaller cases, identify it as strong induction or present a minimal-counterexample proof with well-ordering.
Sixth, misusing infinite descent over real numbers. A sequence of positive real solutions can decrease indefinitely without contradicting any minimum unless such a minimum is known. The positive-integer domain, or another proven well-founded measure, is what powers the standard descent contradiction.
A final self-check is to read your opening and closing sentences together. The opening chooses a specified extreme; the closing should identify a legal object that strictly improves it, or explain how all legal objects meet the target theorem. If the two are not directly connected, a logical step may be missing.
How to teach this to a JC2 learner
Begin with the simplest extremal claim: among a finite list, some value is largest. Then use the claim in a short exchange argument: if an alleged best arrangement can be improved without violating any condition, it was not best. The tutor should explicitly distinguish discovery of the improvement from proof that the improvement is legal.
Next introduce the well-ordering principle for positive integers. Students often know ordinary induction but have not used a least counterexample. Ask them to write both versions of the prime-factorisation proof and compare what information about smaller integers each relies on.
A third lesson can practise modular residue analysis and infinite descent. The key questions are: which coordinate or measure becomes smaller, is it still positive, and does the new tuple solve exactly the same equation? Each missing answer invalidates the argument.
A fourth lesson might use the longest-path graph proof. Students draw a small finite network, choose a longest path and explain why any neighbour outside it would contradict the selection. The example is useful even for a learner who has never taken a formal graph-theory course because the definitions can be supplied within the lesson.
For extension, analyse a tournament king. This teaches a distinct way to use a maximum: the vertex with largest out-degree must have a two-step reachability property. A student who can explain the d+1 contradiction understands more than a remembered slogan about extreme objects.
Parents can ask three questions without solving the mathematics themselves: ‘Why does the best object exist?’, ‘What would make it even better?’, and ‘Would that improved object still meet the rules?’ If the student answers all three precisely, the proof architecture is becoming stable.
A practical seven-session sequence is: finite maxima; positive-integer minima; prime-factorisation descent; modular descent; longest paths; tournament out-degrees; and a mixed timed problem. Ask the learner to rewrite one solution with the assumptions, extreme choice, constructed alternative and contradiction labelled.
Written progress is visible when students stop saying ‘obviously there is a smallest’ without checking the domain, stop confusing counterexamples with ordinary smaller examples, and begin justifying why the replacement really belongs to the same problem.
For Bukit Timah families, the immutable eduKateSG Mathematics small-group reference describes three-student Secondary 1 tutorials near Sixth Avenue MRT. It serves as a qualitative teaching reference and is not evidence of current JC2 H3 timetable or capacity. Enquire with eduKateSG directly about advanced Mathematics programme coverage and availability.
One productive tutoring approach is to combine the extremal method with the student’s H2 foundations. Algebraic factorisation, modular arithmetic and elementary inequalities supply the actual deductions. Strong problem-solving heuristics become valuable when those tools are used accurately and in the right logical order.
Do not spend the entire lesson on sophisticated examples if the student cannot state quantifiers and domains. A two-line false proof can reveal the most important gap. Repair the first invalid line, then build towards unfamiliar problems under time pressure.
Frequently asked questions
Is the extremal principle an official separate H3 Mathematics topic?
The official 2027 H3 Mathematics syllabus lists proof, induction, contradiction, existence, combinatorial arguments and problem-solving heuristics. The extremal principle is a useful way to apply those expectations rather than a separately named additional-content section.
Why can I choose a least counterexample among positive integers?
Every nonempty subset of positive integers has a least element. This well-ordering fact justifies the selection; a corresponding statement is not true for arbitrary positive real subsets.
Does a minimum have to be unique?
No. Several objects may attain the same minimum value. A valid extremal proof can select any one of them unless uniqueness is separately required.
How is infinite descent different from ordinary contradiction?
Infinite descent is a particular contradiction argument: the existence of one permitted solution forces a strictly smaller permitted solution, contradicting a least-solution choice or well-foundedness.
Does one graph example prove all finite graphs have cycles?
No. The general proof assumes minimum degree at least two and chooses a longest simple path. The hypotheses and argument, not a particular drawing, establish the universal claim.
Can I use a decreasing real number to prove termination?
Not by decrease alone. A real sequence can decrease indefinitely while staying positive. Termination requires an appropriate well-founded measure or other strong progress condition.
Why do equality conditions matter in optimisation?
An inequality may show only that a value is a bound. Checking equality verifies whether a feasible object attains it, allowing a maximum or minimum claim.
Do these original examples appear on the SEAB H3 examination?
These are original learning problems illustrating transferable proof principles, not claims about a particular past or future question. The official syllabus and specimen materials control examination expectations.
Official sources and related H3 guides
Consult the official 2027 Singapore–Cambridge H3 Mathematics (9820) syllabus. Its content outline includes proofs by contradiction and induction, combinatorial arguments, solving non-routine problems and using supplied definitions or results outside the named areas. The SEAB syllabus directory identifies the official examination-year materials.
For related skills, use counterexamples and contradiction, mathematical induction and pigeonhole reasoning, and invariants and monovariants. They share a common discipline: clearly state the domain, verify every inference and conclude no more than has actually been proved.
The central lesson for a JC2 reader is that mathematical difficulty is not measured by the length of a calculation. A well-chosen extreme can turn a seemingly endless search into a short argument that no permitted counterexample can survive.