Stationary points and the second derivative
At the top of a hill or the bottom of a valley a curve is momentarily flat: its gradient is zero. These points are called stationary points, and finding them and deciding whether each is a maximum or a minimum is one of the most frequent tasks on Paper 1. It appears as a standalone question, inside curve sketching, and as the engine of every practical maximum and minimum problem. The syllabus asks you to locate stationary points, determine their nature (including by the second derivative test) and use them in sketching graphs.
What a stationary point is
Picture walking left to right along the curve . You climb, reach a peak, descend into a dip, then climb again. At the peak and the bottom of the dip the tangent is horizontal, so the gradient is .
The horizontal tangents touch the curve at and . These are the stationary points of this curve.
A stationary point of the curve is a point where . The tangent there is horizontal.
A maximum point (local maximum) is a stationary point where the curve changes from increasing to decreasing. A minimum point (local minimum) is a stationary point where the curve changes from decreasing to increasing. Maximum and minimum points are together called turning points.
The word "local" matters. A maximum point is the highest point in its neighbourhood, not necessarily the highest point on the whole curve. The cubic above goes off to on the right, so is not the greatest value of overall; it is just the top of a hill. Example 2 below has a maximum point that is lower than its minimum point.
Finding stationary points
The condition is a single equation, so finding stationary points is mostly algebra.
- Rewrite the equation as powers of if needed, then find (using the chain rule for brackets raised to a power).
- Set and solve for . Factorise wherever you can.
- Reject any solutions outside the domain given in the question.
- Substitute each into the equation of the curve to find the -coordinate.
- Write each stationary point as a coordinate pair.
For :
so at and , giving and .
Deciding the nature: the second derivative test
The second derivative is the rate of change of the gradient. It tells you which way the gradient is moving as you pass through the stationary point.
- At a minimum, the gradient goes from negative, through zero, to positive. The gradient is increasing, so . The curve is shaped like a cup, .
- At a maximum, the gradient goes from positive, through zero, to negative. The gradient is decreasing, so . The curve is shaped like a cap, .
A memory aid: a positive second derivative is a smile (minimum); a negative one is a frown (maximum).
At a stationary point, where :
| Nature | |
|---|---|
| minimum point | |
| maximum point | |
| test fails: use the first derivative test |
For the cubic, . At it is , so is a minimum. At it is , so is a maximum. This agrees with the graph.
The first derivative test
When the second derivative is zero, or is awkward to find, look at the sign of just either side of the stationary point. A sign change from to means the curve rises then falls: a maximum. From to it falls then rises: a minimum. This uses the same idea as increasing and decreasing functions.
Evaluate at a value of slightly less than and slightly greater than the stationary value (with no other stationary point in between).
| Sign of : left, then right | Shape | Nature |
|---|---|---|
| then | maximum point | |
| then | minimum point | |
| same sign both sides | or | neither |
The third row is the case at the origin: the gradient is zero there, but positive on both sides, so the curve flattens for an instant and carries on rising. Such a point is neither a maximum nor a minimum.
A stationary point that is neither a maximum nor a minimum is called a stationary point of inflexion. The syllabus states that knowledge of points of inflexion is not included in Paper 1, so you will not be asked to find them. You must still not misclassify one: if the second derivative is zero, you cannot conclude anything until you check the signs either side.
A second derivative of zero does not mean "neither". The curve has at its stationary point , yet the point is clearly a minimum, since . Only the first derivative test settles it.
Using stationary points to sketch a curve
Stationary points, intercepts and end behaviour together pin down the shape of a graph. For a cubic or a quartic, the stationary points tell you exactly where the hills and dips are, and the discriminant of a quadratic tells you how many there are.
- Find and classify the stationary points.
- Find the -intercept (put ) and, if they factorise easily, the -intercepts.
- Decide the behaviour for large positive and large negative from the leading term.
- Plot the stationary points, join them smoothly with the correct turning shapes, and label every key point with its coordinates.
A cubic has , a quadratic. It has two stationary points when this quadratic has discriminant , one stationary point (neither max nor min) when the discriminant is , and none when it is . Example 5 uses this.
Worked examples
Find the coordinates of the stationary points on the curve and determine their nature.
Solution
Setting gives or .
Second derivative: .
At : , so is a minimum point.
At : , so is a maximum point.
A curve has equation for . Find the coordinates of the stationary points and determine their nature.
Solution
Write . By the chain rule,
Set this equal to zero:
so or .
Differentiate again: .
At : , so is a minimum point.
At : , so is a maximum point.
The maximum point has a smaller -coordinate than the minimum point. That is no contradiction: they are on different branches of the curve, separated by the asymptote , and each is only a local maximum or minimum.
Find the stationary points of and determine their nature.
Solution
So or , with and .
.
At : , so is a minimum point.
At : , so the test fails. Use the first derivative either side:
| sign |
The gradient is negative on both sides, so is neither a maximum nor a minimum. (The factor is never negative, so the sign of near is the sign of , which is negative.)
The curve has a stationary point at .
(a) Find the values of and .
(b) Find the coordinates of the other stationary point and determine the nature of both.
Solution
(a) Two facts give two equations. The point is on the curve:
The gradient is zero at . Since ,
Subtract (1) from (2): , so and .
(b) Now and
The factor must be there, which is a useful check. The other root is :
. At it is , so is a minimum. At it is , so is a maximum.
Find the set of values of the constant for which the curve has two stationary points.
Solution
Two stationary points means has two distinct real roots, so its discriminant is positive:
So or .
At the derivative is , which has a repeated root: one stationary point, which is neither a maximum nor a minimum. For there are none.
Using to find the -coordinate. Once you have , substitute into the equation of the curve. Substituting into just gives .
Reading as "neither". It means only that the test has failed. Use the signs of either side, as with , which has a minimum.
Swapping the signs. Positive second derivative means minimum. Students who think "positive means top" lose the classification mark.
Losing a root when dividing. In , dividing by throws away . Factorise instead: .
Forgetting the negative root. gives , two stationary points, not one.
Ignoring the domain. If the question says , discard any stationary value with and say so.
- "Find the coordinates" needs both and . Leaving only costs the final accuracy mark.
- "Determine the nature" needs a reason. Write the value of (or its sign) and the conclusion: ", so minimum." A bare "minimum" with no evidence usually scores nothing.
- Use the first derivative test if the question allows "any method" and the second derivative is messy. Show the actual values of at your chosen points, not just signs.
- Exact values: give , not , unless decimals are asked for.
- Check by factor: if you know one stationary value (as in Example 4), the derivative must have that factor. If it does not factorise with it, recheck your constants.
- "Hence sketch" expects the stationary points, intercepts and correct end behaviour, each labelled. You do not need to plot to scale.
- A stationary point is where ; the tangent there is horizontal.
- Solve for by factorising, then use the curve for .
- Second derivative test: gives a minimum, a maximum, is inconclusive.
- First derivative test: to is a maximum, to a minimum, no change is neither.
- Maximum and minimum points are local: a maximum can be lower than a minimum on another branch.
- A cubic's derivative is a quadratic, so its discriminant decides whether there are two, one or no stationary points.
- Unknown constants: "stationary point at " gives two equations, and .
- Knowledge of points of inflexion is not required, but never classify from alone.
Practice questions
- Find the stationary points of and determine their nature.
- Find the coordinates of the stationary point on the curve and determine its nature.
- The curve is defined for . Find the stationary point and show that it is a maximum.
- Find the stationary point of and determine its nature. Explain why the second derivative test does not help.
- Find the stationary points of and determine their nature.
- The curve has a stationary point where . Find , the -coordinate of the stationary point and its nature.
- The curve has stationary points where and . Find and , the coordinates of both stationary points, and their nature.
- (a) Find the stationary points of and determine their nature. (b) Sketch the curve. (c) Hence find the set of values of for which the line meets the curve at exactly four points.
- The curve , where is a positive constant, has a stationary point with -coordinate . Find , and show that the stationary point is a maximum.
Answers
-
, so or . At : ; at : . : at it is , so is a minimum; at it is , so is a maximum.
-
, gives , , . at , so is a minimum.
-
, gives , , . , so is a maximum.
-
only at , giving . at , so the test is inconclusive. At , ; at , . The sign goes to , so is a minimum.
-
, so , , or . At : . At : . : at it is , minimum ; at it is , maximum .
-
. At : , so . Then . , so is a minimum.
-
has roots and , so it equals . Comparing, and , so , . The curve is , so (as in question 1) is a maximum and is a minimum.
-
(a) By the chain rule , which is zero at and . Points: , , . Expanding, , so . At : , maximum. At : , minima. (b) A "W" shape, symmetrical about the -axis, touching the -axis at with a maximum at , rising steeply for large . (c) A horizontal line cuts the W four times when it lies strictly between the minimum value and the maximum value: .
-
gives , so . Then , so and (as ). The point is . , so it is a maximum.