Critical Points, Increasing/Decreasing and Extrema
To find extrema, solve f'(x)=0 and find where f' does not exist; these critical points are the only interior candidates for local maxima and minima. Then check the sign of f' on each side: a change from positive to negative marks a maximum, from negative to positive a minimum, and no change means no extremum.
Increasing and decreasing
The sign of the derivative is the sign of the slope. On an interval where f'(x)>0 at every point, f is increasing. Where f'(x)<0 at every point, f is decreasing.
A sign chart turns this into a routine:
- Compute f' and factor it if possible.
- Mark on a number line every point where f'=0 or f' is undefined, and every point where f itself is undefined.
- Pick one test number in each interval between the marked points and find the sign of f' there.
- Read off the behavior: plus means increasing, minus means decreasing.
One test number per interval is enough because, for the functions in this article, f' keeps a single sign between consecutive marked points.
Critical points
A critical point of f is a number c in the domain of f where f'(c)=0 or f'(c) does not exist. There are two kinds:
- Stationary points, where f'(c)=0. The graph has a horizontal tangent there.
- Points where f' is undefined, such as a corner, a cusp or a vertical tangent, provided f(c) is defined.
A number outside the domain of f is never a critical point, even though it still belongs on the sign chart.
Critical points are candidates, nothing more. A critical point may be a local maximum, a local minimum, or neither.
Running example: f(x)=x^3-6x^2+9x+1
Differentiate and factor:
f'(x)=3x^2-12x+9=3(x^2-4x+3)=3(x-1)(x-3).
The derivative exists everywhere, so the critical points come from f'(x)=0: they are x=1 and x=3. They split the number line into three intervals.
| Interval | Test number | f' at the test number | Sign | Behavior |
|---|---|---|---|---|
| (-\infty,1) | 0 | f'(0)=9 | + | increasing |
| (1,3) | 2 | f'(2)=-3 | - | decreasing |
| (3,\infty) | 4 | f'(4)=9 | + | increasing |
The function rises up to x=1, falls from x=1 to x=3, and rises again after x=3.
First derivative test
At a critical point c where f is continuous, look at how the sign of f' changes:
- From + to -: f has a local maximum at c.
- From - to +: f has a local minimum at c.
- No sign change: there is no extremum at c.
In the example, f' goes from + to - at x=1, so there is a local maximum there with value f(1)=1-6+9+1=5. At x=3 it goes from - to +, so there is a local minimum with value f(3)=27-54+27+1=1.
Second derivative test
When f'(c)=0, the second derivative can settle the question without a sign chart. The idea: if f''(c)<0, then f' is decreasing as it passes through 0, so it goes from positive to negative, which is a maximum. If f''(c)>0, then f' goes from negative to positive, which is a minimum.
| At a point with f'(c)=0 | Conclusion |
|---|---|
| f''(c)<0 | local maximum |
| f''(c)>0 | local minimum |
| f''(c)=0 | inconclusive |
If the notation f'' is new, see the second derivative. For the example, f''(x)=6x-12. Then
f''(1)=-6<0\ \Rightarrow\ \text{local maximum},\qquad f''(3)=6>0\ \Rightarrow\ \text{local minimum}.
Both match the first derivative test.
When the test fails. If f''(c)=0, the point can be a maximum, a minimum, or neither. Take f(x)=x^3 at c=0. Here f'(x)=3x^2 and f''(x)=6x, so f'(0)=0 and f''(0)=0. The test says nothing, so go back to the first derivative test: 3x^2 is positive on both sides of 0, so the sign does not change and x^3 has no extremum at 0. The function keeps rising straight through the point. How the graph bends at such points is the subject of concavity.
The second derivative test also cannot be used where f'(c) does not exist.
Critical points where f' is undefined
Take f(x)=x^{2/3}, with the real cube root so that the function is defined for negative x as well. For x\ne 0,
f'(x)=\frac{2}{3}x^{-1/3}=\frac{2}{3x^{1/3}}.
This is undefined at x=0, but f(0)=0 is defined, so 0 is a critical point. There is no horizontal tangent; the graph has a sharp point, a cusp, described in corners and cusps.
Now use the sign of f'. For x<0 the cube root is negative, so f'(x)<0. For x>0, f'(x)>0. The sign goes from - to +, so there is a local minimum with value f(0)=0. Since x^{2/3}=\left(\sqrt[3]{x}\right)^2\ge 0, this is in fact the lowest value of the function.
Absolute extrema on a closed interval [a,b]
A continuous function on a closed interval [a,b] always has an absolute maximum and an absolute minimum. Finding them takes three steps:
- Find the critical points of f inside (a,b).
- Evaluate f at each of those points and at both endpoints a and b.
- The largest value is the absolute maximum and the smallest is the absolute minimum.
Apply this to the running example on [0,4]. The critical points x=1 and x=3 both lie inside the interval.
| x | f(x) | Role |
|---|---|---|
| 0 | 1 | endpoint |
| 1 | 5 | critical point |
| 3 | 1 | critical point |
| 4 | 64-96+36+1=5 | endpoint |
The absolute maximum is 5, attained at x=1 and at x=4. The absolute minimum is 1, attained at x=0 and at x=3.
Ties are allowed. A function has one largest value and one smallest value, but it can reach each at more than one point. Note also that the endpoint x=4 reaches the same height as the local maximum, while the local minimum at x=3 ties with the endpoint x=0.
If f is not continuous on the interval, or is undefined somewhere in it, the guarantee fails, and an absolute extremum may not exist.
Mistakes to avoid
- Treating every critical point as an extremum. For x^3, the point x=0 is critical but there is no maximum or minimum. Always check the sign change.
- Forgetting the endpoints. On a closed interval, the absolute maximum or minimum can sit at a or b, as it does at x=0 and x=4 in the example.
- Using the second derivative test when f''=0. The result is inconclusive, not “no extremum”. Fall back on the first derivative test.
Try it: Open
x^3-6*x^2+9*x+1with the interval [0,4]. Under the graph, “Local extrema” lists (1,5) as a local max and (3,1) as a local min, and ticking “shade increasing/decreasing” colours the intervals of your sign chart. The panel “Absolute max/min on [a, b]” is filled in with 0 and 4, and its table shows four rows: “left endpoint (a)” with f=1 tagged “absolute min”, “critical point” x=1 with 5 tagged “absolute max”, “critical point” x=3 with 1 tagged “absolute min”, and “right endpoint (b)” with 5 tagged “absolute max”. Both ties are tagged, so the answers are 5 and 1. The panel adds a warning when the function is discontinuous or undefined on the interval, or when a critical point was only found numerically.The “Local extrema” list also marks corners and cusps where f' does not exist: for
abs(x)it lists (0,0) as a local min. On an interval,abs(x)on [-1,2] shows x=0 as a “critical point” tagged “absolute min”.
Frequently asked questions
Is a critical point the same as a stationary point?
Not quite. A stationary point is one where f'(c)=0. A critical point is either a stationary point or a point in the domain where f' does not exist. In the examples above, x=1 and x=3 for the cubic are stationary, and x=0 for x^{2/3} is critical without being stationary.
Can a function have a maximum where f' doesn’t exist?
Yes. Reflect the graph of x^{2/3} in the x-axis and the minimum at the sharp point becomes an absolute maximum at 0. The derivative does not exist there, so a search that only solves f'(x)=0 would miss it. The first derivative test still works, because it needs only the sign of f' on each side.
What if there is no closed interval?
Then absolute extrema may not exist. The running cubic has a local maximum 5 and a local minimum 1, but on the whole real line it has no absolute extremum, since f(x)\to\infty as x\to\infty and f(x)\to-\infty as x\to-\infty. Compare the local values with the behavior of f at the ends of the domain. The function x^{2/3} does have an absolute minimum, 0, on the whole line, because it is never negative.
Where to go next
- Concavity and inflection points: what f'' says about the bend of a graph, and where the bend changes.
- Tangent and normal lines: the equation of the horizontal tangent at a stationary point, and of tangents elsewhere.
- Higher-order derivatives: notation and patterns for f'' and beyond.
To test your own results, find extrema with the calculator and compare its “Check your answer” verdict with your hand-computed derivative, which can be in any equivalent form.