A product such as x²(1−x) looks simple, but its two positive factors compete. Increasing x enlarges x² while shrinking 1−x. Differentiation identifies the balance that makes the product largest over 0 ≤ x ≤ 1.
For f(x) = x²(1−x), the derivative is f′(x) = 2x−3x² = x(2−3x). The interior stationary point is x = 2/3, where f(2/3) = 4/27, the global maximum on the interval. More generally, for positive p and q, the product x^p(1−x)^q has its unique interior maximum at x = p/(p+q). This guide derives the pattern, explains the endpoint checks, develops an inequality proof and shows how the result changes on another interval or with whole-number restrictions.
Why the two factors compete
On 0 ≤ x ≤ 1, both x and 1−x are non-negative. But as x increases, x² becomes larger while 1−x becomes smaller.
At x = 0 the first factor is zero, so the product is zero. At x = 1 the second factor is zero, so the product is also zero.
Between the endpoints, the product is positive. Therefore a positive maximum must occur somewhere in the interior rather than at either extreme.
The problem is an example of constrained optimisation: one quantity grows at the expense of another. The correct result comes from balancing their rates of change, not from maximising either factor independently.
State the domain before using calculus
Let f(x) = x²(1−x), with 0 ≤ x ≤ 1. The interval is part of the mathematical question.
On this domain, x² ≥ 0 and 1−x ≥ 0, so f(x) ≥ 0. Both endpoints are included and produce value zero.
The same formula, if considered over every real x, has quite different behaviour. For large negative x, x²(1−x) becomes arbitrarily large, so there is no global maximum over all real numbers.
Thus the finite interval is essential. Differentiating a function without preserving its specified domain can lead to a stationary value being called a global maximum when the unrestricted function has no such bound.
Expand and differentiate the opening product
Expand x²(1−x) = x²−x³. Differentiate term by term to obtain f′(x) = 2x−3x².
Factor the derivative as f′(x) = x(2−3x). The stationary inputs are x = 0 and x = 2/3.
The value x = 0 lies at the domain boundary, while x = 2/3 lies strictly inside the interval. Both may be examined, but only the second can be an interior positive maximum.
The factorisation reveals the sign of the derivative directly. A lengthy quadratic formula calculation is unnecessary because the stationary equation already has simple linear factors.
Use the derivative sign chart
For 0 < x < 2/3, both x and 2−3x are positive, so f′(x) > 0. Therefore the function increases on this interval.
For 2/3 < x < 1, x is positive while 2−3x is negative, so f′(x) < 0. The function decreases after x = 2/3.
The derivative thus changes from positive to negative at x = 2/3. This proves that the interior stationary point is a local maximum.
Because the function rises from its left endpoint, then falls toward its right endpoint, this point also gives the global maximum over the full permitted interval.
Evaluate the exact maximum
Substitute x = 2/3 into f(x) = x²(1−x).
The squared factor is x² = (2/3)² = 4/9. The remaining factor is 1−2/3 = 1/3.
Multiply: f(2/3) = (4/9)(1/3) = 4/27.
At x = 0 or 1 the function equals zero, so 4/27 is strictly larger than both endpoint values. The global maximum is therefore 4/27, achieved at x = 2/3.
Check with the second derivative
Differentiate f′(x) = 2x−3x² again to obtain f″(x) = 2−6x.
At x = 2/3, f″(2/3) = 2−4 = −2, which is negative. Therefore the stationary point is a local maximum.
At x = 0, f″(0) = 2, suggesting a local minimum for the unrestricted polynomial, while on [0,1] it is an endpoint with value zero.
The second-derivative test agrees with the first-derivative sign chart. The endpoint comparison then completes the global maximum argument. Both steps matter in optimisation over a closed interval.
Why x = one-half is not the optimum
A student might guess x = 1/2 because it balances x and 1−x equally. But the first factor is x² rather than x, so the two contributions are not symmetric.
At x = 1/2, the product is (1/4)(1/2) = 1/8, or 0.125.
At x = 2/3, the product is 4/27, approximately 0.148148, which is larger.
The optimum shifts toward the factor carrying exponent two. The higher power makes the product more sensitive to x being small, so increasing x beyond one-half is worthwhile until the loss from 1−x becomes too large.
A short value table confirms the curve shape
Evaluate f(x) at selected values: f(0) = 0, f(1/3) = (1/9)(2/3) = 2/27, f(1/2) = 1/8 and f(2/3) = 4/27.
At x = 3/4, the value is (9/16)(1/4) = 9/64, about 0.140625. At x = 1 the value returns to zero.
The table supports the rise and fall predicted by differentiation. But it cannot prove that no untested real input gives a larger value.
The derivative sign chart is the general proof. Numerical samples are useful as checks or learning aids, not as substitutes for an optimisation argument over infinitely many possible inputs.
A direct inequality proof without differentiation
There is a second exact method using the arithmetic–geometric mean inequality. For 0 ≤ x ≤ 1, the three non-negative numbers x/2, x/2 and 1−x add to one.
Their product is (x/2)(x/2)(1−x) = x²(1−x)/4.
By AM–GM, the product of three non-negative numbers with sum one is at most (1/3)³ = 1/27.
Therefore x²(1−x)/4 ≤ 1/27, giving x²(1−x) ≤ 4/27. Equality holds when x/2 = 1−x = 1/3, so x = 2/3. This proves the same maximum without differentiation.
What the equality condition teaches us
The AM–GM method does more than provide an upper bound. Its equality condition identifies exactly when the maximum is attained.
All three constructed numbers must be equal: x/2 = x/2 = 1−x. Solving x/2 = 1−x gives 3x/2 = 1 and x = 2/3.
Thus the balance is not x = 1−x, but rather each half of x matching the remaining factor 1−x.
This is consistent with the squared power of x: the product contains two copies of x, so the optimum balances three equal pieces after the appropriate normalisation.
Generalise to x to the power p times one minus x to the power q
Let F(x) = x^p(1−x)^q on 0 ≤ x ≤ 1, where p and q are positive real numbers. The function is positive inside the interval and zero at the endpoints.
On 0 < x < 1, differentiate using the product rule. The derivative is p x^(p−1)(1−x)^q − q x^p(1−x)^(q−1).
Factor the common positive terms x^(p−1)(1−x)^(q−1). This gives F′(x) = x^(p−1)(1−x)^(q−1)[p(1−x)−qx].
The bracket simplifies to p−(p+q)x. It is this simple linear factor that determines where the product stops increasing and begins decreasing.
Find the general optimal input
For positive p and q and 0 < x < 1, the factors x^(p−1) and (1−x)^(q−1) are positive wherever the derivative is evaluated.
Therefore the derivative’s sign is determined by p−(p+q)x. It is positive for x < p/(p+q) and negative for x > p/(p+q).
The unique interior stationary point is x = p/(p+q). Because both exponents are positive, this number lies strictly between zero and one.
The function increases up to that point and decreases afterwards. Hence the unique global maximum over [0,1] occurs at x = p/(p+q). This includes the opening example p = 2, q = 1, giving 2/3.
Calculate the general maximum value
At x = p/(p+q), the complementary factor is 1−x = q/(p+q).
Substitute both into F(x) = x^p(1−x)^q. The maximum becomes [p/(p+q)]^p [q/(p+q)]^q.
Combine the denominators to obtain F_max = p^p q^q/(p+q)^(p+q).
This exact expression is valid for p,q > 0 under the stated closed-interval problem. It is a family of optimisations rather than a number to memorise without understanding the derivative sign change.
A symmetric example with equal exponents
Let F(x) = x²(1−x)² on [0,1]. Here p = q = 2.
The general optimum occurs at x = 2/(2+2) = 1/2, reflecting the function’s symmetry when x and 1−x exchange roles.
At x = 1/2, the maximum is (1/2)²(1/2)² = 1/16.
This differs from the opening example where the exponents were unequal. The midpoint x = 1/2 is optimal when the power weights balance, not simply because every expression contains x and 1−x.
An asymmetric fifth-degree example
Consider F(x) = x³(1−x)² on 0 ≤ x ≤ 1. The powers are p = 3 and q = 2.
The optimal input is x = 3/(3+2) = 3/5. Then 1−x = 2/5.
The maximum is (3/5)³(2/5)² = (27/125)(4/25) = 108/3125.
Even though the function has degree five when expanded, the factored product form makes differentiation and sign analysis straightforward. Expanding every polynomial term first would increase the working without adding useful insight.
A first-degree product gives the familiar quarter
For F(x) = x(1−x), take p = q = 1. The optimum is x = 1/2.
The value is (1/2)(1/2) = 1/4.
Algebraically, x(1−x) = 1/4 − (x−1/2)², so the same result follows from completing the square.
This familiar quadratic is a base case for the general product pattern. It also explains why the optimum shifts only when the powers are unequal: x and 1−x have equal weight when both exponents are one.
The logarithmic-derivative method
For positive F(x) = x^p(1−x)^q on 0 < x < 1, taking natural logarithms gives lnF = p ln x + q ln(1−x).
Differentiate to obtain F′/F = p/x − q/(1−x). A stationary input satisfies p/x = q/(1−x).
Cross-multiply: p(1−x) = qx, so p = (p+q)x and x = p/(p+q).
This route is often efficient when p and q are large or not integers. It uses the same derivative logic and is optional enrichment when logarithmic differentiation has been taught.
Changing the interval changes the optimal coordinate
Suppose the variable is restricted to a general interval a ≤ x ≤ b with a < b, and the product is G(x) = (x−a)^p(b−x)^q, for p,q > 0.
Define t = (x−a)/(b−a), so 0 ≤ t ≤ 1. Then x−a = (b−a)t and b−x = (b−a)(1−t).
The product becomes (b−a)^(p+q)t^p(1−t)^q. The positive factor (b−a)^(p+q) does not change the optimal t.
Therefore t = p/(p+q), so x = a+(b−a)p/(p+q) = (qa+pb)/(p+q). This is the weighted position between the interval endpoints.
A concrete example on a different interval
Maximise G(x) = (x−2)²(8−x) for 2 ≤ x ≤ 8.
Here a = 2, b = 8, p = 2 and q = 1. The optimal input is x = 2+(8−2)(2/3) = 2+4 = 6.
At x = 6, the factors are (6−2)² = 16 and 8−6 = 2, giving a maximum of 32.
The general scaled formula confirms this: interval width is 6, so the maximum is 6³(4/27) = 216·4/27 = 32. The optimum is two-thirds of the way from the left endpoint to the right endpoint.
A positive constant multiplier does not change the maximiser
Consider H(x) = 7x²(1−x) over [0,1]. It is seven times the opening function.
Since multiplying by a positive constant preserves the ordering of outputs, the optimum remains at x = 2/3.
The maximum value becomes 7(4/27) = 28/27.
This is useful when a practical objective includes a price, material coefficient or unit-conversion factor independent of x. A constant scale can change the best output while leaving the best choice of input unchanged.
A negative multiplier changes maximum into minimum
Now consider J(x) = −7x²(1−x) on [0,1]. Because x²(1−x) is non-negative, J(x) is non-positive.
At x = 2/3, where the positive product is largest, J takes its minimum value −28/27.
Its maximum over [0,1] is zero, attained at both endpoints x = 0 and x = 1.
Thus one cannot say that a constant multiplier never changes the optimisation problem. Positive scaling preserves maxima and minima, while negative scaling reverses their roles.
An integer-grid restriction requires checking neighbours
Suppose x can only take values 0, 1/5, 2/5, 3/5, 4/5 or 1. The continuous maximum of x²(1−x) still lies at 2/3, which is unavailable.
Because the function increases before 2/3 and decreases afterwards, compare the adjacent permitted inputs 3/5 and 4/5.
At 3/5, the value is (9/25)(2/5) = 18/125 = 0.144. At 4/5, it is (16/25)(1/5) = 16/125 = 0.128.
Therefore the best permitted grid input is 3/5, not 4/5. The correct decision comes from evaluating both neighbours, not blindly rounding 2/3 to the nearest fifth.
Why choosing the geometrically nearest grid point can fail
The function x²(1−x) is not symmetric about its maximum x = 2/3. Its rate of decrease to the right may differ from its rate of increase to the left.
Therefore for a coarse discrete grid, the closest candidate in horizontal distance need not always give the larger output across different constrained product functions.
The safe method is to exploit monotonicity. Any permitted point farther left than the nearest left neighbour has a smaller output, and likewise for the right side. So the best candidate must be one of the two allowed neighbours bracketing the continuous optimum.
Then evaluate both exactly. This reasoning transfers to other unimodal objective functions without incorrectly assuming symmetry about their peak.
A probability-related interpretation
Expressions of the form x^p(1−x)^q arise in simple probability models where x represents a success probability and repeated outcomes are treated as independent under a fixed probability.
For instance, the probability of a particular pattern with two successes and one failure contains the factor x²(1−x), with 0 ≤ x ≤ 1.
Its largest possible value over choices of x occurs at x = 2/3. This corresponds to the fraction two successes out of three trials in the exponent pattern.
The mathematical optimisation is exact within the model, but real-world inferences depend on the independence and constant-probability assumptions. A fitted probability is not automatically a causal explanation or a guaranteed prediction.
Do not confuse a product maximum with a sum maximum
For x²(1−x), the product is zero at both endpoints and positive inside. Its maximum occurs at a balance point.
The sum x²+(1−x) instead has a different expression and different optimisation behaviour. Maximising each factor separately also gives incompatible choices: x² favours x = 1, while 1−x favours x = 0.
The product rule in differentiation is exactly what combines their competing effects. A student who differentiates only x² or only 1−x is not optimising the stated objective.
Write the complete function before differentiating or applying an inequality. The arithmetic operation between factors fundamentally affects where the optimum occurs.
A compact exam solution
First state the permitted interval. Identify the full product to be maximised and note the endpoint outputs.
Second differentiate, preferably keeping the product form when it simplifies factorisation. Solve f′(x) = 0 and retain only stationary candidates in the interval.
Third use derivative signs or the second derivative to classify the interior candidate. Compare its output with the endpoint values to establish the global maximum.
For x^p(1−x)^q with positive p,q, the stationary balance p(1−x) = qx gives x = p/(p+q). If inputs are restricted to a discrete grid, evaluate the permitted neighbours instead of reporting the continuous optimum directly.
Practice questions with checked answers
Question one: maximise x²(1−x) on [0,1]. Answer: 4/27 at x = 2/3.
Question two: maximise x(1−x) on [0,1]. Answer: 1/4 at x = 1/2.
Question three: maximise x²(1−x)² on [0,1]. Answer: 1/16 at x = 1/2.
Question four: maximise x³(1−x)² on [0,1]. Answer: 108/3125 at x = 3/5.
Question five: maximise (x−2)²(8−x) on [2,8]. Answer: 32 at x = 6.
Question six: maximise 7x²(1−x) on [0,1]. Answer: 28/27 at x = 2/3.
Question seven: among x ∈ {0,1/5,2/5,3/5,4/5,1}, maximise x²(1−x). Answer: 18/125 at x = 3/5.
For each, state why the chosen input is admissible and why an endpoint or competing candidate cannot do better.
Learning transfer and related guidance
Products of powers connect algebraic factorisation, product-rule differentiation, inequalities, constrained optimisation and mathematical modelling. The general power pattern and logarithmic-derivative proof can be treated as enrichment according to the student’s school programme.
For Singapore, refer to the applicable SEAB lists for 2026 O-Level Additional Mathematics, 2027 SEC G3 or 2027 SEC G2, as appropriate.
Related guides include How to Find the Maximum Area of a Rectangle Under a Parabola, How to Maximise a Circular Sector’s Area with Fixed Perimeter and How to Maximise a Quadratic When Only Whole-Number Inputs Are Allowed. Visit the Additional Mathematics archive for further resources.
Parents can enquire about suitable eduKate Singapore maximum three-student Mathematics lessons near Sixth Avenue MRT. The lasting lesson is to express the trade-off precisely, differentiate the whole product and prove the maximum over the domain that the question allows. Properly taught kids shine a bright light into the future.
