Quick Read: The Hardest Step Comes Before Differentiation
Optimisation word problems in Additional Mathematics often look much harder than the derivatives needed to solve them. The real challenge is turning a description into a function of one variable that measures exactly what must be maximised or minimised. A student may know how to differentiate a quadratic perfectly yet fail because the rectangle’s perimeter condition was translated incorrectly or an impossible dimension was accepted as the final answer.
The reliable process is to define the unknown with its unit, write the condition that links all quantities, eliminate extra variables, state the physically allowed domain, differentiate the objective function and locate stationary candidates. Then decide whether the candidate is a maximum or minimum, compare relevant endpoints or boundaries, and return to the original context with the right units. Every stage is necessary; a derivative alone is not a complete modelling solution.
For example, a rectangle with perimeter 40 m and width x metres has length 20−x metres. Its area is A(x)=x(20−x)=20x−x², with 0
The 2027 G3 Additional Mathematics Curriculum Connection
The 2027 Singapore-Cambridge SEC G3 Additional Mathematics K341 syllabus explicitly lists applications of differentiation to maxima and minima problems, stationary points, increasing and decreasing functions and the second derivative test. It also includes quadratic functions as models and finding quadratic maxima or minima through completing the square. This article shows how these skills work together using original illustrative problems, without representing the examples as past examination questions.
Students should consult the official 2027 SEAB G3 syllabus directory and the subject document for their examination year. A particular school may teach optimisation through rectangles, open boxes, cost models or other situations, but the transferable method remains the same. The important distinction is between a given mathematical model and a claim about a real measured business or physical system; the models below state their assumptions explicitly.
First Find the Objective, Not the Derivative
Read the final sentence of the word problem before calculating. If it asks for the largest area, the objective is area. If it asks for the smallest perimeter, the objective is perimeter. If it asks for a maximum volume, multiplying the relevant dimensions is the right starting structure. If it asks for a price that gives the highest model revenue, the objective is price multiplied by the model quantity sold. The derivative comes after the objective has been expressed correctly.
Students sometimes use the constraint as though it were the objective. For a rectangle with fixed perimeter 40 m, maximising the perimeter is meaningless because the perimeter is already fixed. The mathematical task is to use that perimeter to express the length in terms of width, then maximise area. A good tutor asks the student to identify both the constraint and the objective in separate sentences before writing any algebra.
Naming the objective also helps with units. A(x) should have area units such as m², V(x) volume units such as cm³, and a perimeter P(x) length units such as m. If an expression intended to represent area contains only a length added to a number without compatible units, the model may be malformed. Units can reveal errors that a plausible derivative would otherwise hide.
The Six-Stage Optimisation Route
| Stage | Student question | Expected evidence |
|---|---|---|
| 1. Define | What does x represent, and in what units? | A labelled variable |
| 2. Constrain | Which quantities are fixed or linked? | An equation connecting dimensions |
| 3. Model | What quantity must be largest or smallest? | One objective function |
| 4. Restrict | Which x-values make sense? | A domain or inequality |
| 5. Differentiate | Where can the objective stop increasing or decreasing? | Derivative and stationary candidates |
| 6. Verify | Is this a max/min, and is it allowed? | Sign/second derivative, endpoints and units |
This checklist is intentionally visible while the student is learning. Once the first four stages become familiar, differentiation is often straightforward. But skipping a constraint or domain can turn an otherwise accurate derivative calculation into the answer to the wrong problem. A learner should be able to explain why each line follows from the scenario, not just imitate the final algebra from a workbook.
Worked Example 1: Largest Area with Fixed Perimeter
A rectangle has perimeter 40 m. Let x be its width and y its length, both positive. The perimeter condition is 2x+2y=40, so x+y=20 and y=20−x. The area is A=xy=x(20−x)=20x−x². Since both dimensions must be positive, 0
Differentiate: A′(x)=20−2x. Setting the derivative to zero gives x=10. The second derivative is A″(x)=−2, negative everywhere, so this is a maximum. The corresponding length is y=20−10=10. The largest area is 10×10=100 m². The optimal rectangle happens to be a square under these assumptions.
The domain is not decorative. If x=25 were accepted, y=−5 and the model would describe a negative physical dimension rather than a rectangle. Likewise, the endpoints x=0 and x=20 produce degenerate zero-area shapes rather than positive rectangles. The interior stationary candidate x=10 is permitted and provides a genuine maximum because the area tends towards zero near both boundaries.
A second route completes the square: A(x)=20x−x²=100−(x−10)². Since a real square is nonnegative, the area is at most 100, with equality at x=10. This independent algebraic check confirms the calculus result. A strong student recognises that completing the square can be quicker for a quadratic objective, while differentiation remains essential for more complicated functions.
Perimeter, Area and Volume Are Different Objectives
A fixed-perimeter rectangle problem is not the same as a fixed-area rectangle problem. With fixed area, the dimensions multiply to a constant and the perimeter becomes the variable quantity to minimise. Similarly, an open-top box made from a sheet has a volume that depends on both height and base dimensions. Students should avoid applying “the optimal shape is always a square” without checking the particular constraints and objective.
In a short tutorial, ask the learner to create a sentence beginning “I want to maximise...” or “I want to minimise...” and then name the quantity and unit. This verbal step can be more revealing than the derivative. A student who cannot state the objective clearly is not ready to commit to a mathematical expression, no matter how quickly they differentiate a polynomial.
Worked Example 2: Minimise Perimeter for a Given Area
A rectangle must have area 144 m². Let one side be x>0 metres. Then the other side is 144/x metres. The perimeter is P(x)=2x+288/x, with domain x>0. Differentiate to obtain P′(x)=2−288/x². Setting it to zero gives 2x²=288, so x²=144 and x=12 because a side length must be positive.
The other side is 144/12=12 m. The perimeter is 2(12+12)=48 m. The second derivative is P″(x)=576/x³>0 for x>0, so the stationary point is a minimum. As x approaches zero, the term 288/x becomes very large; as x grows without bound, 2x becomes very large. Thus the stationary point is also the global minimum within the positive domain.
If a student solves x²=144 and reports x=−12 as another candidate, the algebraic equation has been solved but the geometry has been ignored. A physical length must be positive. The right result is selected by the original domain, not by the teacher's preference for a positive-looking answer. This example demonstrates how domain restrictions can simplify the answer set before the final conclusion.
Worked Example 3: A Fence Beside a River
A rectangular enclosure uses 60 m of fencing along three sides, with the fourth side bordered by a straight river and needing no fence under the model's assumptions. Let x be the length of each side perpendicular to the river and y the fenced side parallel to the river. The constraint is 2x+y=60, not 2x+2y=60, because only three sides use fencing.
The enclosed area is A=xy=x(60−2x)=60x−2x². For positive dimensions, 015×30=450 m².
The river condition changes the optimum from the fixed-four-sides case. A student who includes fencing for the river side will create the wrong constraint and solve a different problem. This is why drawing a labelled rectangle and marking exactly which sides consume material is the most important first step. Geometry supplies the algebra; calculus optimises only the model it is given.
Why Stationary Does Not Automatically Mean Maximum
A point where the derivative equals zero is a stationary candidate. It might be a maximum, minimum or stationary point of inflexion, depending on the function. A second derivative less than zero at the point is evidence of a local maximum, while greater than zero indicates a local minimum. If the second derivative is zero, the test is inconclusive and derivative sign analysis or another valid method is required.
For example, f(x)=x³ has derivative 3x², zero at x=0, and second derivative 6x, also zero at x=0. Yet the function keeps increasing through zero, so there is no local maximum or minimum there. A student who labels every f′(x)=0 root a maximum has skipped the classification stage. Optimisation requires proof of the nature of the candidate and the relevant domain.
The Domain May Be the Decisive Step
A valid algebraic root can be physically meaningless. In a box made from a sheet 20 cm by 12 cm, cutting squares of side x from every corner leaves dimensions 20−2x and 12−2x. Positive dimensions require x>0, 20−2x>0 and 12−2x>0. Therefore the actual allowed interval is 0
Domain restrictions are especially important with an objective whose derivative has two algebraic roots. One root may lie outside the physical interval. A student must reject it using the original length condition rather than simply call it a local minimum or maximum of an algebraic polynomial considered on all real numbers. The problem asks about a real object described by a particular model.
Worked Example 4: Maximum Volume of an Open-Top Box
A rectangular sheet is 20 cm long and 12 cm wide. Equal squares of side x cm are cut from the four corners, and the remaining edges are folded to form an open-top box. The box's height is x, its length is 20−2x and its width is 12−2x. Each base dimension loses x at both ends, which is why the subtraction contains 2x rather than x. The resulting volume is V(x)=x(20−2x)(12−2x) cubic centimetres.
The physical constraints are x>0, 20−2x>0 and 12−2x>0. Their intersection is 0
Expand the volume carefully: (20−2x)(12−2x)=240−64x+4x². Hence V(x)=240x−64x²+4x³. Differentiate: V′(x)=240−128x+12x². Divide the stationary equation by four to get 3x²−32x+60=0. This quadratic is the mathematical heart of the problem; it appears only after the physical model has been built.
Apply the quadratic formula: The second derivative is This result is a useful illustration of why students should not blindly solve a cubic's derivative and report both roots. The derivative has two real roots, but the original physical domain retains only one. The final answer must say what x represents—a square-cut size in centimetres—and what V represents—a volume in cubic centimetres. Reporting x=2.427 as “the maximum volume” would confuse an optimal input with the value of the objective. A student who calculates the derivative perfectly but models the base as (20−x)(12−x) will get a plausible-looking wrong optimum. The first correction belongs to the geometry of cutting both ends, not to polynomial differentiation. The folded-box sketch should mark the four corner squares and the reduced dimensions. Once the model is correct, a derivative can be trusted to answer the intended physical question. A teacher can also ask whether the final answer is a cut size, a box dimension or the maximum volume. In a multi-part problem, these may all be requested. The permitted x produces box length about 15.145 cm, width about 7.145 cm and height about 2.427 cm. The volume is the product of all three, not equal to any one dimension. Students should connect each number to a labelled quantity. Suppose an illustrative model predicts that at price p currency units, quantity demanded is Differentiate: A student might instead maximise q(p) and report p as close to zero as possible, which does not answer the revenue question. The distinction between demand and revenue matters: one measures quantity and the other price multiplied by quantity. Asking “What is being maximised?” before differentiating catches this error. In a real commercial analysis, costs, capacity, competition and model uncertainty would also matter; the mathematical answer applies only to the stated demand model. If the question had asked for maximum model profit, revenue alone would not suffice. Profit equals revenue minus cost, and the cost function would need to be supplied or defined. Students should not introduce an unsupported cost assumption merely to use a familiar derivative formula. A clear objective function must follow the conditions actually given, not a story invented during the working. The first or second derivative can identify local behaviour near a stationary point. But a word problem may ask for the greatest or least value over an entire allowed domain. A local maximum need not be the global maximum if there are other stationary points or endpoints with larger objective values. For a function continuous on a closed interval, the standard global comparison checks all relevant stationary points and the endpoints. A rectangle with 0 The open-top box has a cubic volume function with two derivative roots, one outside the permitted interval. Its volume tends to zero at each physical boundary and reaches a positive maximum at the allowed interior stationary root. Comparing these behaviours supports a global conclusion rather than merely labelling V″<0 and stopping. This level of reasoning helps students distinguish a calculus calculation from a complete optimisation argument. A first derivative that changes from positive to negative indicates a local maximum; a change from negative to positive indicates a local minimum. The test explains how the objective moves on either side of the candidate. For the rectangle area This reasoning can be helpful when a second derivative calculation is awkward or inconclusive. It also makes the result intuitively meaningful: if the width is too narrow, a small increase increases area, but beyond the optimum the resulting decrease in length dominates. The sign test should be applied to the physically permitted interval, not to irrelevant x-values that make dimensions negative. For a proposed stationary point where f″=0, the second derivative test gives no conclusion. A sign chart around the point, higher-order analysis where appropriate, or direct comparison of objective values is needed. Students should not write “second derivative equals zero, so neither maximum nor minimum” as a universal rule. The conclusion is only that this particular test is inconclusive. A useful algebraic habit is to introduce two variables initially if they help represent the geometry, then use the constraint to eliminate one before differentiating. In the perimeter-40 rectangle, x and y are the sides; x+y=20 allows y=20−x. In the three-fenced-sides enclosure, 2x+y=60 allows y=60−2x. The formulas look similar, but the coefficients encode different physical conditions. Students who try to differentiate an objective with two linked variables as though they were independent may omit the way one dimension changes when the other changes. The one-variable reduction makes the constraint explicit and enables ordinary G3 single-variable differentiation. More advanced methods for constrained optimisation are unnecessary for these examples and should not be confused with the syllabus route. When several quantities are connected by an equation, solve for the variable that gives the simplest objective expression. For a fixed area xy=144, y=144/x leads to a manageable rational perimeter. For an open box, the dimensions are already expressed in one cut-size variable, so the volume can be multiplied directly. Choose the substitution that reduces complexity without losing restrictions. Quadratic objectives can often be optimised exactly without calculus. For For These comparisons teach a broader lesson: optimisation is about a mathematical objective with a domain, not about the exclusive use of one formula. If the question says “use differentiation”, show the derivative route. If no method is specified and a quadratic completes the square neatly, that may be efficient. The desired conclusion must be justified in either case. An improvement plan should follow the earliest wrong decision. If the objective was wrong, more differentiation will not repair the problem. If the function was correct but x was outside the allowed domain, teach physical restrictions and root filtering. If the calculus was correct but the answer used square metres instead of cubic centimetres, reinforce unit interpretation. The same numerical test score can conceal very different learning needs. A final conclusion should name the optimal quantity and its corresponding objective value when required. For the open-top box, write that the corner-square side should be approximately 2.43 cm to maximise volume, and that the maximum model volume is approximately 263 cm³. For the river enclosure, write that the two perpendicular fenced sides are 15 m each and the third fenced side 30 m, enclosing 450 m². Avoid saying only “maximum at x=10” unless the problem already defines x and asks solely for that number. A school mark scheme may give separate credit for defining the variable, obtaining the objective, locating the stationary point, proving its nature and interpreting the final value. Read the question's requested accuracy and label units consistently. A precise final sentence can prevent an otherwise strong solution from appearing unfinished. When a model uses illustrative currency or demand units, retain that context and do not present a hypothetical optimum as real-world pricing advice. Mathematical optimisation tells us what is best within the stated assumptions; it does not prove the assumptions are realistic. This distinction is part of sound mathematical communication and helps students apply school mathematics responsibly. Begin with simple rectangles, enclosures and fixed-sum problems without differentiation. Ask the student to label every relevant dimension on a sketch and write one condition that links the variables. Then have the student say in words what is being maximised or minimised. The goal is a correct model: if the constraint is already wrong, no amount of later calculus can repair the answer. Use the constraint to eliminate a second variable before attempting differentiation. Practise converting fixed perimeter into area and fixed area into perimeter. Require the permitted domain to be written alongside each objective. This makes it clear that a rectangle's side length cannot be negative and that a denominator representing a length cannot be zero. Work with quadratic objectives whose derivatives are simple linear expressions. Solve f′(x)=0, evaluate the second derivative and use a sign chart to explain whether the candidate is a maximum or minimum. Compare with completing the square as an independent check. Do not let the learner stop at the derivative or report the x-value as the objective value. Compare a four-sided fixed-perimeter rectangle with an enclosure that requires fencing on only three sides. Have the student identify which edges use material, then write the appropriate constraint. Mix different perimeter totals and ask for both optimal dimensions and the largest enclosed area. The changing numerical answers should follow from one stable modelling procedure. Use a labelled rectangular sheet and mark square cut-outs at all four corners. Write the box dimensions, the volume function and the strict physical interval before expanding. Differentiate the volume cubic, solve the quadratic stationary equation and reject any roots outside the permitted interval. Calculate the maximum volume only after the valid cut size has been identified. Introduce one hypothetical revenue or cost model, clearly separating the objective from a related quantity such as demand. Ask the student to check units, the allowed domain and any modelling assumptions. For functions with several stationary points or boundary possibilities, compare the relevant objective values rather than assuming the first derivative root is automatically the global optimum. Give an unlabeled mix of area, perimeter, volume and model-revenue problems. The learner should choose a variable and build the objective without being told the formula. Record the earliest incorrect line, assign a changed follow-up and revisit the weak step after a short delay. Correct independent modelling is stronger evidence of progress than reproducing a familiar derivative from the last lesson. 1. A rectangle has perimeter 32 m. What dimensions maximise its area? Answer: 8 m by 8 m, area 64 m². 2. A three-sided enclosure has 48 m of fencing beside a straight boundary. What dimensions maximise area? Answer: two perpendicular sides of 12 m, one parallel side of 24 m, area 288 m². 3. A rectangle has area 100 m². What dimensions minimise perimeter? Answer: 10 m by 10 m, perimeter 40 m. 4. For 8. Differentiate the volume: The answers should be checked for reasons, not just numbers. A student may differentiate correctly but use the wrong constraint, reject the wrong stationary root, omit physical units or confuse the input dimension with maximum volume. Identifying which decision failed tells the tutor what to practise next. A new problem with different measurements tests whether the skill has truly transferred. A parent can ask four practical questions: “What does x stand for?”, “What is fixed?”, “What quantity are you trying to make largest or smallest?”, and “Are the final dimensions physically possible?” A learner who answers these without looking at a worked solution has understood much of the hardest reasoning. The derivative matters, but it should be attached to an objective whose meaning is clear. When a school script shows an incorrect answer, inspect the earliest invalid line. If the volume was modelled with 20−x instead of 20−2x, the repair is geometric. If the model was right but differentiation was wrong, practise the calculus rule. If the calculus was correct but an impossible cut size was accepted, practise the domain. Each issue requires a different corrective action rather than an identical long worksheet. Progress should be measured through fresh problems attempted after a delay, not merely how many similar examples were completed during one lesson. An independent student can draw the geometry, construct a valid one-variable function, test candidates and explain the conclusion with units. This is a stronger indication of learning than copying a maximisation formula from a nearby worked example. In eduKateSG's small three-student tutorial format, the tutor can watch the learner identify the objective and set up the constraint before any differentiation begins. One student may need geometry-to-algebra repair, another may be strong at modelling but weak at derivatives, and a third may need help comparing stationary points with domain boundaries. Close observation allows targeted practice rather than treating every wrong optimum as one generic calculus error. The learning sequence is diagnose, repair, practise, connect, test, correct and retest. The aim is to help a student select a variable, build a meaningful model and verify the result independently. Optimisation is especially valuable because it combines algebra, calculus, units and careful reading in one coherent problem-solving process. The basic differentiation methods in this syllabus operate on a function of one variable. If a constraint links two dimensions, use it to express one in terms of the other. Then the area, perimeter or volume becomes a function that can be differentiated consistently while respecting the original relationship. Not by itself. It identifies a stationary candidate, which could be a maximum, minimum or stationary point of inflexion. Use the second derivative test or a first-derivative sign chart, then consider relevant endpoints and all other allowed candidates when the question asks for a global optimum. A derivative equation can have solutions outside the physical model's domain. In a box built from a 20-by-12 cm sheet, the cut must be less than 6 cm for both base dimensions to stay positive. An algebraic root around 8.239 cm does not describe a box and must be excluded, even though it solves the derivative equation. For a quadratic objective, completing the square can give an exact maximum or minimum efficiently. But more complicated objectives such as cubic box volumes usually require other tools, and a question may expressly request differentiation. Understanding both approaches allows an independent check where they overlap. A negative second derivative at a stationary point establishes a local maximum, while a positive value establishes a local minimum. If the second derivative is zero, the test is inconclusive; examine derivative signs or another valid method. The physical domain and possible boundaries still matter when deciding a global answer. The 2027 SEAB G3 school-candidate syllabus directory provides the official Additional Mathematics K341 content, including maxima and minima applications of differentiation and quadratic modelling. Students following another subject level or examination year should check that cohort's specification rather than assuming identical requirements. For related skills, read Classifying Stationary Points with First and Second Derivatives, Choosing the Right Quadratic Method and Connected Rates of Change Without Unit Errors. These reinforce the derivative, algebra and interpretation skills used above. The broader Additional Mathematics improvement guide and subject knowledge hub organise further topic-specific learning. Parents can arrange an eduKate Singapore parent–student consultation to review actual school working and choose the most useful next step.x=[32±√(32²−4·3·60)]/6=[32±√304]/6=(16±2√19)/3. Numerically these candidates are about 2.427 cm and 8.239 cm. Only x=(16−2√19)/3≈2.427 lies inside 0V″(x)=−128+24x. At x≈2.427 it is negative, so the stationary point is a local maximum. As x approaches either endpoint 0 or 6 from within the permitted interval, the box volume approaches zero. There is only one permitted stationary candidate, so this gives the greatest volume for the model. Substitution yields approximately V=262.682 cm³, or about 263 cm³ to three significant figures.Box Problems: What Could Go Wrong Before Differentiation?
Student working Issue Repair V=x(20−x)(12−x)Only one corner cut subtracted per dimension Draw and label both removed strips Domain written x>0 only Box base might become negative Require 20−2x>0 and 12−2x>0 Both quadratic roots accepted Physical constraints ignored Test each candidate against 0 V″<0 stated without evaluationClassification unsupported Substitute permitted stationary root Answer “2.427 cm³” for cut size Variable and objective units confused Write x in cm, V in cm³ Worked Example 5: A Hypothetical Revenue Model
q(p)=120−3p units, for 0R(p)=p(120−3p)=120p−3p² currency units. This is a stipulated mathematical model, not a measured forecast of consumer behaviour. Its purpose is to show how an algebraic relationship becomes an optimisation objective.R′(p)=120−6p. Setting it to zero yields p=20. The second derivative is −6, so the stationary point is a maximum. The model then predicts q=120−3(20)=60 units and maximum revenue R=20×60=1200 currency units. The domain condition is met because 20 lies between zero and 40.The Difference Between Local and Global Extrema
What the First Derivative Sign Test Contributes
A=20x−x², the derivative 20−2x is positive when x<10 and negative when x>10, within the domain. Area increases before the optimal width and decreases afterwards.Constraint First, Function Second
An Efficiency Check with Completing the Square
R(p)=120p−3p², complete the square: R(p)=1200−3(p−20)². Since a square is nonnegative, R≤1200 and equality holds at p=20. The derivative method reaches the same conclusion. Two independent routes agreeing is a useful check on the algebra and the maximum value.A(x)=60x−2x² in the river-enclosure example, completing the square gives A=450−2(x−15)². The model's maximum is 450 when x=15, exactly as differentiation shows. A student can choose the method requested by the question and use the other as a check. For a cubic volume objective, completing a simple square is not available in the same direct way, making calculus particularly valuable.Common Modelling Errors and What They Reveal
Wrong step Likely weakness Next practice Differentiate the constraint instead of the objective Max/min quantity not identified State “maximise/minimise ...” first Perimeter relation missing a factor of two Geometry translated incorrectly Label every side on a sketch Physical domain omitted Algebra detached from context Write positivity inequalities All stationary roots reported Candidates not filtered Test each root against domain f″=0 called a maximum automatically Derivative test misunderstood Use a first-derivative sign chart Input x reported as objective value Units and question deliverable confused Write both optimum x and f(x) Premature decimal rounding Exact and approximate work mixed Keep radical root until final value How to Write a Complete Optimisation Conclusion
A Fourteen-Day Optimisation Improvement Plan
Days 1–2: Turn Words into Variables and Constraints
Days 3–4: Build One-Variable Objective Functions
Days 5–6: Differentiate and Classify Stationary Candidates
Days 7–8: Add Three-Sided Fencing Models
Days 9–10: Practise Open-Top Boxes
Days 11–12: Interpret and Verify a Contextual Maximum
Days 13–14: Unseen Mixed Questions and Delayed Retesting
Independent Self-Test with Answer Checks
A(x)=20x−x², find the stationary point: x=10. 5. Is this point a maximum or minimum? A maximum, because A″=−2. 6. Explain why the side length domain is 0V′=240−128x+12x². 9. Find the physically valid optimum: x=(16−2√19)/3≈2.427 cm. 10. State the approximate maximum volume: 263 cm³ to three significant figures. 11. For hypothetical R(p)=120p−3p², find the revenue-maximising price p=20 and maximum model revenue 1200 currency units. 12. Explain why a correct stationary x-value is not necessarily the final objective value requested.How Parents Can Recognise Better Modelling
How a Focused 3-Pax Tutorial Can Help
Frequently Asked Questions
Why must I express the objective using one variable?
Does f′(x)=0 prove that I found the greatest value?
Why do I have to reject some quadratic roots?
Can completing the square replace calculus?
What does the second derivative tell me?
Official Curriculum and Related Reading