JC2 H3 Mathematics Tutor Bukit Timah

Did you know? Turning on or off exactly two lamps can never change whether the total number of illuminated lamps is odd or even. It does not matter which pair you choose or how many moves you make. That one observation can settle a puzzle involving thousands of possible arrangements without listing any of them.

This guide is for JC2 students and families exploring a JC2 H3 Mathematics tutor in Bukit Timah and wanting a practical way to understand non-routine problems. The useful word is invariant: a quantity or property preserved by every allowed move. Its close relative, a monovariant, moves only one way—usually increasing or decreasing—and can prove that a process must terminate.

Quick answer: To use an invariant, define the state of the problem, list every legal move, find a property unchanged by each move, then compare the starting and proposed final states. If their invariant values differ, the target is impossible. To prove termination, find a nonnegative integer measure that strictly decreases at each legal move. The process cannot continue forever because such a measure cannot decrease indefinitely.

Invariants and monovariants are useful applications of H3’s prescribed mathematical proof and problem-solving heuristics, not separately named additional-content headings in syllabus 9820. The published syllabus includes uncovering structure, considering cases, combinatorial arguments, modular arithmetic, proof by contradiction and non-routine problem solving.

The one-minute learning diagnosis

Put five imagined switches on the board, all initially off. Every move must change the states of exactly two different switches. Ask a learner: can we reach a state in which exactly one switch is on? A student might start drawing many sequences; the mathematically decisive observation is that two switches change together, so the parity of the on-count does not change.

The initial on-count is zero, which is even. The target on-count is one, which is odd. Therefore the target is impossible. The answer does not depend on the number of available switches, as long as moves toggle two distinct ones. This is a complete impossibility argument once the parity change for every allowed move has been justified.

Next ask a different question: can we reach a state with exactly two switches on? Yes. Toggle those two once. The invariant rules out odd targets, but does not prohibit all even targets. This distinction between a necessary condition for reachability and a sufficient construction is an H3-level logic habit.

A well-prepared student can state the invariant, verify it under every move, identify a forbidden target and provide a permitted construction in complete mathematical sentences. If they only say ‘the answer looks even’, ask them to explain why no long sequence of moves changes the conclusion.

What exactly is an invariant?

First describe a state. For a lamp puzzle the state might be a row of zeros and ones. For coins in boxes it might be a list of nonnegative integer counts. For a board game it could be the entire placement of pieces. A state description should include the information needed to decide which moves are legal.

Next describe a move. The allowed operation must be precise: change exactly two different lamps, move one coin without creating a negative box, or replace a number according to a stated rule. If the move definition is vague, the proof can accidentally ignore a possibility.

Now choose a function I(state), or even a property such as ‘the sum is even’. A genuine invariant satisfies I(new state)=I(old state) for every legal move. If a move can change I in one case, it is not an invariant for the stated process.

Finally, an invariant propagates through any finite sequence of moves. You can see this directly: if I remains the same after move one and after move two, it is still the same after both. Repeating the argument gives equality with the starting invariant after any finite number of moves.

The strongest explanations say both what is preserved and why every type of move preserves it. Merely reporting that a numerical pattern remained the same in three examples is evidence, not the general proof.

Invariance is especially powerful for an impossibility claim. If the target has a different invariant value, no permitted sequence can reach it. Conversely, if the invariant values agree, that alone may not prove a path exists. Additional restrictions can still block a move.

One way to discover an invariant is to ask which quantities changed by each move. The raw total may change, but perhaps its parity does not; the exact sum may move, but its remainder modulo a fixed integer remains unchanged; colours may rearrange while a weighted difference is preserved.

These questions connect directly to the syllabus’s emphasis on conditions, quantifiers, mathematical proofs and uncovering hidden structure. The topic is a problem-solving lens rather than a promise that every H3 problem features an invariant.

Full worked investigation: toggling two lamps

Problem. There are seven lamps, each either off or on. Initially all lamps are off. A legal move toggles any two distinct lamps. Can you reach a state with exactly three lamps on? Can you reach every state with an even number of lamps on?

Step 1 — define a measure. Let N be the number of lamps on. The parity N mod 2 is the candidate invariant. We must analyse all ways the two toggled lamps can currently be configured.

Step 2 — check every case. If both selected lamps are off, N increases by two. If both are on, N decreases by two. If one is on and one is off, N stays unchanged. In every case N changes by an even integer, so N mod 2 is preserved.

Step 3 — reject the odd target. Initially N=0, which is even. A state with N=3 would have odd parity. The invariant proves that such a state is unreachable, no matter how long or cleverly the moves are chosen.

Step 4 — prove a constructive converse in this setting. Suppose a desired final arrangement has an even number of lamps on. Starting from all-off, pair its desired on-lamps arbitrarily and toggle each pair once. Different pairs use disjoint lamps, so no earlier desired lamp is reversed. This constructs the target. Therefore, starting from all-off, the reachable arrangements are exactly those with even on-count.

Step 5 — extend the question. What if each move toggles exactly three lamps? The parity of the on-count is no longer invariant: toggling three off lamps changes N by three. The proof above must be rebuilt. A rule change that looks small can destroy an entire invariant.

This exercise illustrates two kinds of reasoning. The invariant gives a necessary condition and an impossibility proof. The explicit pairing construction gives sufficiency. Students who distinguish those jobs avoid the common error of declaring an arrangement reachable merely because it passes one modular test.

Number moves: discovering a congruence obstruction

Suppose a number starts at 11. At each move we may add four or subtract six, provided the resulting number stays nonnegative. Could we reach 20? Both allowed changes are even. Therefore every reachable number has the same parity as 11—odd. Twenty is even, so reaching it is impossible.

This argument remains valid if the process permits any number of moves or arbitrarily large intermediate values, because no move changes parity. It is much stronger than trying ten sequences and observing that none reaches twenty.

Could we reach 17? The parity condition allows it, but we still need a construction. From 11 subtract six to get five, then add four three times: 5→9→13→17. The entire path is nonnegative. This example shows the two stages: screen a target by an invariant, then construct a valid path.

Instead of parity, consider the greatest common divisor of allowed increments. If moves change a number by multiples of g, its residue modulo g is invariant. With moves +8 and −12, g=4, so the residue modulo four cannot change. A target with a different remainder is impossible.

A matching residue still does not settle reachability when moves have direction, boundary or scheduling restrictions. For instance, if only positive increments are permitted, a smaller target might have the same residue but be inaccessible. A nonnegative-state rule may also prevent temporarily needed steps.

To analyse a new move game, calculate the change Δ in each candidate invariant for every move. If all Δ are zero modulo g, you have a modular invariant. If some legal move gives a different residue, revise the candidate rather than ignoring that move.

This method develops the same skills as congruence and modular arithmetic, but the reader job here is different: deciding whether a state transition is possible.

Conservation laws: totals, differences and weighted sums

Imagine three labelled jars containing 3, 5 and 7 counters. A legal move transfers one counter from one nonempty jar to another. The total number of counters is initially fifteen. Every move subtracts one somewhere and adds one elsewhere, so the total remains fifteen.

Thus a target distribution with sixteen counters is impossible. A target such as (0,0,15) passes the total test, and it is reachable: repeatedly transfer all counters from the first two jars into the third. The construction handles the remaining question.

Not every conserved quantity is the plain sum. Suppose a move replaces a pair (a,b) by (a+2,b−1), with integer entries. The sum a+b increases by one, but the weighted combination a+2b remains constant: (a+2)+2(b−1)=a+2b. The right coefficients matter.

This is a useful discovery exercise: write an unknown linear combination pa+qb and require that its change be zero. Under (+2,−1), the change is 2p−q. Any choice with q=2p works; the simplest is a+2b. The invariant was found by solving a small equation.

Of course, if the move is allowed only when b≥1, the invariant proof still holds for every legal move. But the restriction may influence which states are actually reachable. State validity and arithmetic invariance are related but distinct claims.

Geometric colourings serve a similar purpose. When every legal tile covers one black and one white square, the difference between the covered black and white counts stays zero for any collection of tiles. A proposed target with unequal colour counts cannot be tiled by those pieces.

The important lesson is not to memorise ‘use parity’ in every puzzle. It is to compare the way legal moves transform the information in the state and find something they cannot change.

Monovariants: quantities that move only one way

An invariant stays equal. A monovariant changes in a single direction: it is always nonincreasing or nondecreasing under legal moves, and often strictly so. A familiar example is a process that repeatedly removes one object. The object count is not constant, but it decreases.

A decreasing quantity is especially useful for proving termination. If it is a nonnegative integer and drops by at least one after every move, only finitely many moves are possible from any fixed starting state. The process cannot loop forever without violating the bound.

Monotonicity alone is not enough. The real sequence 1, 1/2, 1/4, 1/8,… strictly decreases forever while staying positive. To prove finite termination from a decreasing measure, use a well-founded set such as the nonnegative integers, or prove an appropriate minimum decrement and lower bound.

An invariant often rules out a destination, while a monovariant proves progress or blocks a cycle. Some problems need both: one conserved gcd and one decreasing sum, or one preserved parity and one increasing number of correctly placed pieces.

Students should write the proposed potential Φ(state), check exactly how it changes and specify its range. For a sorting algorithm, Φ might count inverted pairs; for repeated subtraction, it may be a sum of positive integers; for a game, it could be the number of misplaced tokens.

When a proof says a process ‘obviously finishes’, ask for the exact bounded measure that decreases. This replaces intuition with a checkable argument and often reveals that a supposedly guaranteed algorithm can actually loop.

A full example combining an invariant and a monovariant

Problem. Start with positive integers a and b. If a>b replace (a,b) with (a−b,b). If b>a replace it with (a,b−a). Stop when a=b. Prove that the process terminates and identify the final common value.

Find an invariant. The greatest common divisor is unchanged by subtraction: gcd(a,b)=gcd(a−b,b) for a>b, and likewise gcd(a,b)=gcd(a,b−a) for b>a. Any common divisor of a and b divides their difference; any common divisor of a−b and b also divides a. So the sets of common divisors coincide.

Find a monovariant. Let Φ(a,b)=a+b. When a>b, replacing a by a−b reduces the sum by b, a positive integer. When b>a, replacing b by b−a reduces the sum by a, also positive. Both entries remain positive because we subtract the smaller from the larger.

Establish termination. The sum is a positive integer and strictly decreases at each nonterminal step. Therefore it cannot decrease forever. The process must eventually reach a state where neither a>b nor b>a holds, so a=b.

Identify the final value. At termination the pair is (d,d), whose gcd is d. Because the gcd remained invariant from the start, d=gcd(original a,original b). The process therefore ends at the pair consisting of the original gcd twice.

Numerical example. Start with (14,8). The states are (6,8), (6,2), (4,2), (2,2). The sums 22,14,8,6,4 decrease; the gcd remains two. The final value is two, as the proof predicts.

This is a particularly useful H3 reading exercise because each claim does a different job. Without gcd invariance we might show termination but not identify the answer. Without the decreasing sum we might preserve the gcd but never justify that the sequence of moves actually stops.

Sorting by adjacent swaps: a rigorous progress measure

A list is sorted into increasing order when there is no pair of positions i<j with the earlier value exceeding the later value. Define an inversion to be such a pair. The number of inversions is a nonnegative integer bounded by n(n−1)/2 for a list of n distinct numbers.

Consider a legal sorting move that swaps two adjacent entries a,b only when a>b. Their order changes from incorrect to correct. The relative order of every other element with respect to a and b stays unchanged because those two entries were adjacent. Thus the inversion count decreases by exactly one.

If an unsorted list contains an adjacent inversion, swap it. Each swap decreases the inversion count by one, so there can be at most the original number of inversions many swaps. At zero inversions, the list is sorted.

Take the list (3,1,4,2). Its inverted pairs are (3,1), (3,2) and (4,2), so Φ=3. Swap adjacent 3 and 1 to obtain (1,3,4,2), leaving Φ=2. Swap 4 and 2 to obtain (1,3,2,4), leaving Φ=1. Swap 3 and 2 to obtain (1,2,3,4), leaving Φ=0.

This proves a termination property of a particular adjacent-swap procedure. It does not imply that any arbitrary sequence of adjacent swaps terminates; swapping already-correct neighbours could undo progress and create a loop. A monovariant depends on the specified legal move.

The example connects discrete mathematics with mathematical algorithms. An invariant tells us what cannot change; a strictly decreasing potential tells us that a chosen procedure makes measurable progress.

Thirty original investigation clinics

1. 1. Two toggled lamps

Try it: Start with four lamps off and toggle exactly two different lamps per move. Can exactly one be on?

Reasoned answer: No. The number on changes by +2, 0 or −2, so its parity stays even. A one-lamp state has odd count and is impossible.

What it teaches: Always verify the full list of move outcomes, not only the convenient case of two off lamps.

2. 2. Reach an even lamp pattern

Try it: Start with six lamps off; the target has lamps 1, 3, 4 and 6 on. Give a legal sequence.

Reasoned answer: Toggle (1,3), then (4,6). The result has exactly the four target lamps on. Each move changes two distinct switches.

What it teaches: Passing the parity test suggested possibility; the explicit construction proves reachability.

3. 3. Toggling three lamps

Try it: Every move toggles three distinct lamps. Is on-count parity invariant?

Reasoned answer: No. From all-off, toggling three lamps produces an odd on-count. This one legal counterexample disproves parity invariance.

What it teaches: The number of toggles matters: odd-size flips can change parity, while even-size flips preserve it.

4. 4. Moving coins between boxes

Try it: Jars start with counts (2,4,6). Transfers move one coin from one jar to another. Can total 13 be reached?

Reasoned answer: No. The sum remains 12 under every transfer. A proposed total of 13 differs from the invariant.

What it teaches: Be clear that the total is not merely the same in a few examples—it is preserved by the algebra of every transfer.

5. 5. Construct the target distribution

Try it: From (2,4,6), can the jars reach (0,0,12) if any legal transfer is allowed?

Reasoned answer: Yes. Transfer two coins from the first jar to the third, then four from the second to the third. All intermediate counts stay nonnegative.

What it teaches: An invariant may be necessary without being a complete reachability test; here a direct construction supplies sufficiency.

6. 6. Modular move obstruction

Try it: Start at 5 and allow only moves +8 or −12, keeping states nonnegative. Can 10 be reached?

Reasoned answer: No. Both changes are multiples of four, so the current number stays congruent to 1 modulo four. Ten is congruent to 2 modulo four.

What it teaches: Use the greatest common divisor of increments to suggest a modulus, then verify every allowed move preserves it.

7. 7. A valid modular move path

Try it: Starting at 5 with +8 and −12, can 13 be reached?

Reasoned answer: Yes: add eight once. The target has the same residue modulo four, and this valid path is a complete witness.

What it teaches: An allowed residue is not automatically a reachable state; finding an explicit permitted path removes that uncertainty.

8. 8. A constrained move sequence

Try it: Start at 1 and only add four. Can the value −3 be reached even though the two differ by four?

Reasoned answer: No. The sole legal move increases the value. Residue modulo four matches, but direction makes all smaller targets unreachable.

What it teaches: This is a useful case where a modular invariant alone misses an essential monotonicity restriction.

9. 9. Weighted-sum invariant

Try it: Pairs (a,b) can change by (+2,−1) when b≥1. Find a preserved linear combination.

Reasoned answer: The quantity a+2b is unchanged, because its change is 2+2(−1)=0. The ordinary sum a+b changes by one.

What it teaches: A short unknown-coefficient equation often discovers useful weighted invariants more efficiently than trial and error.

10. 10. Reachability of two weighted states

Try it: From (2,5), can (+2,−1) moves reach (8,2)?

Reasoned answer: Yes. Three moves give (4,4), (6,3), (8,2). Both states have the same invariant a+2b=12.

What it teaches: The sequence also respects b≥1 until the final state, so arithmetic conservation and legal state constraints are both checked.

11. 11. An incompatible weighted target

Try it: Start from (2,5) under (+2,−1). Is (7,2) possible?

Reasoned answer: No. Initially a+2b=12; the suggested target has value 11. The invariant rules it out.

What it teaches: A single numerical invariant comparison may replace a long search through possible moves.

12. 12. Colour-balance tiling

Try it: A domino covers two orthogonally adjacent squares on a chequerboard. What colour counts must every complete tiling have?

Reasoned answer: Every domino covers one black and one white square, so the tiled region must contain equal numbers of each colour.

What it teaches: This is a necessary condition for such a tiling; other geometric constraints may still make a balanced region impossible to tile.

13. 13. A potential that declines

Try it: A game starts with 25 tokens and each legal move removes at least one. Can it have infinitely many moves?

Reasoned answer: No. The nonnegative token count strictly decreases by at least one after each move, so there are at most 25 moves.

What it teaches: State the lower bound and the smallest decrement. ‘It keeps getting smaller’ alone is not a fully general termination argument.

14. 14. Decreasing positive reals

Try it: A process replaces a positive real x by x/2 forever. Does strict decrease prove finite termination?

Reasoned answer: No. Starting from one, the positive values 1,1/2,1/4,… strictly decrease but the process can continue indefinitely.

What it teaches: Monovariant termination claims need an appropriate discrete or well-founded range, not merely monotonicity.

15. 15. Euclid’s invariant

Try it: For positive a>b, replace (a,b) with (a−b,b). What number-theory quantity is unchanged?

Reasoned answer: The gcd remains unchanged because gcd(a,b)=gcd(a−b,b); common divisors of the two pairs coincide.

What it teaches: This argument applies to every positive pair satisfying the move condition, not only a numerical demonstration.

16. 16. Euclid’s progress measure

Try it: For positive unequal a,b under repeated subtraction, what decreases at every legal move?

Reasoned answer: The positive integer a+b drops by the smaller entry, so it decreases strictly until the entries become equal.

What it teaches: Combine this with gcd invariance to explain both why the procedure stops and which value it reaches.

17. 17. Euclid’s numerical trace

Try it: Start the subtraction gcd process at (21,15). What is the terminal pair?

Reasoned answer: A valid sequence is (6,15)→(6,9)→(6,3)→(3,3). The final common value is three, the initial gcd.

What it teaches: Tracing one case tests arithmetic; the general proof remains the gcd invariant and decreasing-sum argument.

18. 18. Adjacent inversion count

Try it: For the list (3,2,1), how many inversions are present?

Reasoned answer: All three pairs are inverted: (3,2), (3,1), (2,1). The inversion count equals three.

What it teaches: The inversion count is defined over pairs of positions, not merely adjacent pairs, even though each legal correction swaps adjacent elements.

19. 19. A correcting swap

Try it: Why does swapping an adjacent inverted pair decrease the inversion count by exactly one?

Reasoned answer: The two swapped values reverse their own incorrect order. Their order relative to any third value remains unchanged because no third value lies between them.

What it teaches: The proof explains a global count using a local move; it is stronger than checking a short sample list.

20. 20. A sorting loop with a bad rule

Try it: Suppose arbitrary adjacent swaps are allowed, including swapping a sorted pair. Must the process terminate?

Reasoned answer: No. Two elements may be swapped back and forth forever. The inversion count is not strictly decreasing for all such moves.

What it teaches: A termination claim depends on a move rule that guarantees progress, not on the existence of some successful sequence.

21. 21. Parity of a sum

Try it: Each move replaces (x,y) by (x+3,y−3). What is invariant?

Reasoned answer: The sum x+y is exactly constant. Therefore both its value and its parity are invariant.

What it teaches: Prefer the strongest simple invariant: the exact sum conveys more information than parity alone.

22. 22. Parity of each coordinate

Try it: Under the move (x,y)→(x+2,y−2), what happens to individual coordinate parities?

Reasoned answer: Both x and y retain their respective parity because their changes are even. The sum also remains constant.

What it teaches: Several invariants may coexist. Choose the one that distinguishes the particular start and target.

23. 23. A non-invariant candidate

Try it: Under (x,y)→(x+1,y−1), is x−y invariant?

Reasoned answer: No. The difference becomes (x+1)−(y−1)=x−y+2. Its exact value changes even though its parity is preserved.

What it teaches: When a proposed invariant fails, sometimes a coarser version such as a remainder class still works.

24. 24. A reduced residue

Try it: A number can increase by six or decrease by nine. Which modulus is automatically preserved?

Reasoned answer: Every change is divisible by three, so the residue modulo three is unchanged. Neither operation necessarily preserves the residue modulo six.

What it teaches: Do not choose a modulus simply because one move uses it; verify it against all possible moves.

25. 25. Maximum move count

Try it: A list of five distinct entries is sorted using adjacent inverted-pair swaps. Give a universal upper bound.

Reasoned answer: There are at most C(5,2)=10 inversions at the start. Every legal swap reduces that count by one, so at most ten swaps are needed.

What it teaches: The extremal reversed list reaches the upper bound; the bound was proved without listing all possible orderings.

26. 26. A move with optional inaction

Try it: A process may either remove one token or make a move that changes nothing. Does token count prove finite termination?

Reasoned answer: Not if the idle move can be repeated indefinitely. The token count is only nonincreasing, not strictly decreasing on every legal move.

What it teaches: For guaranteed termination, rule out infinite nonprogress steps or use a stronger progress measure.

27. 27. Distinguishing necessity and sufficiency

Try it: A parity invariant says odd states cannot be reached from even ones. Does matching even parity prove all even states reachable?

Reasoned answer: Not in general. A separate construction or reachability proof is required; the two-lamp complete-pair puzzle happens to permit one.

What it teaches: The logical difference between a necessary invariant condition and sufficient construction should be stated explicitly.

28. 28. A state-space loop

Try it: A deterministic game has finitely many states and runs for more steps than there are states. What must happen?

Reasoned answer: By pigeonhole, some state repeats. Because the update is deterministic, repetition produces a subsequent cycle.

What it teaches: A repeated state need not mean the player has reached the target; it can diagnose nontermination.

29. 29. Building a potential

Try it: A state (a,b) changes to (a−1,b+1), with a≥1. Suggest a decreasing integer measure.

Reasoned answer: The first coordinate a is a nonnegative integer and decreases by one on each move. Therefore no more than the initial a moves can occur.

What it teaches: A weighted total can be constant while an individual coordinate supplies the progress measure.

30. 30. Designing a move invariant

Try it: A move changes three coordinates by (+1,+1,−2). What simple quantity stays constant?

Reasoned answer: The sum x+y+z changes by 1+1−2=0, so it is conserved. Individual coordinates do not remain fixed.

What it teaches: This generalises conservation to more variables and invites the student to search for additional weighted invariants.

Mistakes worth finding before a timed paper

The first mistake is writing down an alleged invariant after testing only two or three moves. A rigorous proof must cover every possible legal move, including mixed cases. With two toggled switches, for example, both-on, both-off and mixed are different cases and each must be checked.

The second mistake is using conservation to claim reachability. Matching totals or remainders is not enough to build a permitted sequence. Counterexamples appear whenever there are one-way moves, blocked intermediate states or geometry that restricts placement.

The third is forgetting the state domain. If jars cannot have negative counters, a purely algebraic path passing through a negative entry is not a legal construction. A proof about positive integer moves must keep the entries positive in every intermediate state.

The fourth is claiming termination from a quantity that sometimes stays constant. A nonincreasing integer potential can support a termination proof only when every nonterminal move makes strict progress, or when a separate bound prevents indefinitely many neutral moves.

The fifth is using a real-valued decreasing potential without a discrete lower-bound mechanism. The sequence x→x/2 demonstrates why positive and decreasing need not mean finitely many steps.

The sixth is blurring proof and illustration. A table of moves can suggest a pattern, but the general invariant equation or general monovariant inequality is what justifies a claim for arbitrarily long legal sequences.

Finally, a good written solution ends at the right logical strength. If you prove impossibility, say what invariant values differ. If you prove reachability, show an actual path or general construction. If you prove termination, state both the strict change and the well-founded bound.

How an H3 tutor can develop this thinking

A purposeful lesson begins by asking the student to describe the state, not to guess the result. The tutor can ask: what information do you need to know before each move, which quantities change and which restrictions apply? This verbal diagnosis reveals whether the learner is reasoning about the actual problem or an imagined easier one.

Next, the learner predicts a possible invariant and deliberately attempts to break it with a legal move. If it survives, the student writes a general proof. If it fails, the tutor asks whether an exact value, a parity or a remainder class might still be preserved.

Then ask for a counterexample to an overstrong conclusion. In the lamp puzzle, equal parity is sufficient because an explicit pairing construction exists. In a one-direction move game, equal parity alone may be insufficient. Comparing the two conditions helps the learner separate arithmetic obstruction from geometric or procedural reachability.

A second session introduces the decreasing-potential method. Sorting by adjacent inverted swaps and the subtraction gcd algorithm make useful paired examples because the potential has a clear nonnegative integer range and a visible strict decrement.

After the proof is comfortable, vary one rule: permit an idle move, allow three lamps to toggle rather than two, or change the allowed number increments. The student should predict which previously valid proof breaks and explain exactly where.

For home practice, ask a JC2 learner to explain why a long sequence cannot change a certain property, rather than demanding dozens of numerical outcomes. A parent can assess clarity by checking that the student can name the state, move, preserved quantity and target contradiction.

A well-paced practice week might use day one for parity games, day two for modular moves, day three for conservation, day four for weighted invariants, day five for Euclid’s gcd process, day six for sorting potentials and day seven for an unfamiliar mixed problem. The objective is improved problem selection and complete proof writing, not a claim of guaranteed examination marks.

The eduKateSG immutable small-group Mathematics reference describes a three-learner Secondary 1 model near Sixth Avenue MRT. This provides a model of close tutor feedback, not confirmation of a current JC2 H3 class. For a current programme, parents should enquire directly with eduKateSG about level coverage, assessment needs and availability.

A productive tutorial should also assess H2 fluency. If algebraic factors, divisibility or basic functions are unstable, the student may fail to find the invariant even when they recognise the game idea. Repairing that foundation is different from merely asking another harder puzzle.

At the end of a session, have the student rewrite one proof in five parts: state, move, candidate invariant, verification for all moves and conclusion. For a termination proof, replace the fourth part with a strict change and explicit lower bound. The final script should read as an argument rather than a diary of guesses.

Frequently asked questions

Are invariants explicitly listed as a separate H3 Mathematics topic?

The 2027 Singapore–Cambridge H3 Mathematics 9820 syllabus does not list ‘invariants’ as a separately named additional content topic. They are an enriching approach to listed proof principles, uncovering structure, congruence and non-routine mathematical problems.

What is the difference between an invariant and a monovariant?

An invariant stays unchanged under every allowed move. A monovariant moves consistently in one direction and can support a termination argument when its range and amount of progress are appropriately controlled.

Can the same problem have more than one invariant?

Yes. A coin-transfer game may preserve both the total count and its parity. A number game can preserve several remainder properties. Choose the strongest invariant that separates the target from the initial state.

Does matching parity guarantee a target can be reached?

No. Matching parity is only a necessary condition when parity is invariant. Reachability also depends on the legal moves and intermediate-state constraints; a valid construction may still be required.

Why does a decreasing nonnegative integer prove termination?

If it strictly decreases at every move, it can decrease only finitely many times from a fixed initial value before reaching zero or a state where no legal move remains. An infinite strictly decreasing sequence of nonnegative integers is impossible.

What if my potential function is a positive real number?

Strict decrease and a lower bound are not always sufficient. For example, 1, 1/2, 1/4,… decreases forever. Additional properties such as a uniform minimum decrease or a well-founded discrete range are needed.

Are the worked games actual past-year SEAB questions?

No. The worked exercises here are original learning examples illustrating proof techniques. For precise examination requirements, use the official SEAB syllabus and materials for the student’s examination year.

How can parents check progress without knowing H3 mathematics?

Ask whether the learner can name what changes, what remains fixed, why that holds for every legal move and what the conclusion proves. An explanation that covers both the possible and impossible cases is more informative than a guessed answer.

Official source and connected lessons

The 2027 Singapore–Cambridge H3 Mathematics (9820) syllabus identifies proof, combinatorial arguments, the pigeonhole principle, mathematical investigation and uncovering patterns as appropriate capabilities. Its note on problems outside specified content is a reason to practise unfamiliar constructions, not an assertion that this particular collection of games is a prescribed chapter.

For more on modular obstructions, read congruences and divisibility. To understand when a counterexample disposes of a claim, see counterexamples and proof by contradiction. For a different approach to counting outcomes, compare induction and the pigeonhole principle.

Read the official SEAB A-Level syllabus index to confirm the H3 subject code and examination-year document. The aim of this article is to develop reasoning that transfers across mathematical topics, not to promise which enrichment example will appear in a particular paper.

JC2 H3 Mathematics Tutor Bukit Timah | Invariants and Monovariants in Mathematical Games