Skip to content
Family Table Math
Auto

Critical Points and Extrema

Where is a function highest? Where is it lowest? These questions come up everywhere: the top of a ball’s flight, the lowest cost, the largest volume. This page sets up the vocabulary (absolute and relative extrema), tells you when a highest and lowest point are guaranteed to exist, and shows where to look for them: at critical points.

“Extrema” is the plural of “extremum”: a maximum or a minimum.

  • ff has an absolute maximum (or global maximum) at x=cx = c if f(c)≥f(x)f(c) \ge f(x) for every xx in the domain (or interval) you’re looking at. The value f(c)f(c) is the absolute maximum value.
  • ff has a relative maximum (or local maximum) at x=cx = c if f(c)≥f(x)f(c) \ge f(x) for all xx near cc (on both sides of cc): it’s the top of a hill, even if there’s a taller hill somewhere else.

Minimums are defined the same way with ≤\le. An absolute extremum can also be a relative one, and it can happen at an endpoint of an interval.

The graph of f(x) = x cubed minus 6x squared plus 9x plus 1 on the closed interval from 0.5 to 4.2. It has a relative maximum at (1, 5), a relative and absolute minimum at (3, 1), and its absolute maximum at the right endpoint (4.2, 7.048). The left endpoint (0.5, 4.125) is neither. 1 2 3 4 2 4 6 relative max (1, 5) relative and absolute min (3, 1) absolute max (4.2, 7.048) endpoint (0.5, 4.125)
On [0.5,4.2][0.5, 4.2], the top of the hill at x=1x = 1 is only a relative maximum: the right endpoint is higher.

Textbooks differ on whether an endpoint can count as a relative extremum. To be safe, look for relative extrema at interior points, and always check endpoints when you want absolute extrema.

Extreme Value Theorem (EVT). If ff is continuous on a closed interval [a,b][a, b], then ff has both an absolute maximum and an absolute minimum on [a,b][a, b].

Both conditions matter:

  • On an open interval, there may be no maximum. f(x)=x2f(x) = x^2 on (0,2)(0, 2) gets close to 44 but never reaches it, because x=2x = 2 isn’t included.
  • With a discontinuity, there may be no maximum. f(x)=1x−1f(x) = \dfrac{1}{x - 1} on [0,2][0, 2] shoots up to ∞\infty near x=1x = 1.

Like the Mean Value Theorem, the EVT guarantees that the extrema exist; it doesn’t tell you where they are.

A critical point (or critical number) of ff is a number cc in the domain of ff where

f′(c)=0orf′(c) does not existf'(c) = 0 \qquad \text{or} \qquad f'(c) \text{ does not exist}
  • f′(c)=0f'(c) = 0: a horizontal tangent, like the top of a smooth hill.
  • f′(c)f'(c) undefined: a corner, a cusp, or a vertical tangent.

If ff isn’t defined at cc, then cc is not a critical point, even if f′f' is undefined there.

If ff has a relative extremum at an interior point x=cx = c, then cc must be a critical point. (At the top of a smooth hill the tangent is flat; otherwise the hilltop is a sharp point where f′f' doesn’t exist.)

The reverse is not true: a critical point doesn’t have to be an extremum. For f(x)=x3f(x) = x^3, f′(0)=0f'(0) = 0, but the graph keeps rising through x=0x = 0. Critical points are candidates. The first derivative test and the candidates test decide which candidates win.

Example 1: Critical points of a polynomial

Section titled “Example 1: Critical points of a polynomial”

Find the critical points of f(x)=2x3−3x2−12x+4f(x) = 2x^3 - 3x^2 - 12x + 4.

Solution. A polynomial’s derivative exists everywhere, so the only critical points are where f′(x)=0f'(x) = 0:

f′(x)=6x2−6x−12=6(x2−x−2)=6(x−2)(x+1)f'(x) = 6x^2 - 6x - 12 = 6(x^2 - x - 2) = 6(x - 2)(x + 1)

The critical points are x=−1x = -1 and x=2x = 2.

Example 2: Where the derivative doesn’t exist

Section titled “Example 2: Where the derivative doesn’t exist”

Find the critical points of f(x)=x2/3(x−5)f(x) = x^{2/3}(x - 5).

Solution. Expand first, so you can use the power rule:

f(x)=x5/3−5x2/3f(x) = x^{5/3} - 5x^{2/3} f′(x)=53x2/3−103x−1/3=53x−1/3(x−2)factor out 53x−1/3=5(x−2)3x1/3\begin{aligned} f'(x) &= \frac{5}{3}x^{2/3} - \frac{10}{3}x^{-1/3} \\ &= \frac{5}{3}x^{-1/3}(x - 2) && \text{factor out } \tfrac{5}{3}x^{-1/3} \\ &= \frac{5(x - 2)}{3x^{1/3}} \end{aligned}
  • f′(x)=0f'(x) = 0 when the numerator is 00: x=2x = 2.
  • f′(x)f'(x) is undefined when the denominator is 00: x=0x = 0. And f(0)=0f(0) = 0 is defined, so x=0x = 0 is in the domain.

The critical points are x=0x = 0 and x=2x = 2. (The graph has a cusp at x=0x = 0.)

Find the critical points of f(x)=sin⁡x+cos⁡xf(x) = \sin x + \cos x on [0,2π][0, 2\pi]. Calculus always uses radians: π\pi radians is 180∘180^\circ.

Solution. f′(x)=cos⁡x−sin⁡xf'(x) = \cos x - \sin x, which exists everywhere. Set it equal to 00:

cos⁡x=sin⁡x⇒tan⁡x=1\cos x = \sin x \quad\Rightarrow\quad \tan x = 1

(Dividing by cos⁡x\cos x is fine: if cos⁡x=0\cos x = 0, then sin⁡x=±1\sin x = \pm 1, so they can’t be equal.) On [0,2π][0, 2\pi], tan⁡x=1\tan x = 1 at x=π4x = \dfrac{\pi}{4} and x=5π4x = \dfrac{5\pi}{4}.

Example 4: Is an absolute maximum guaranteed?

Section titled “Example 4: Is an absolute maximum guaranteed?”

For each function, does the EVT guarantee an absolute maximum and minimum on the interval?

  • (a) f(x)=x3−4xf(x) = x^3 - 4x on [−1,3][-1, 3]
  • (b) g(x)=1xg(x) = \dfrac{1}{x} on [−1,1][-1, 1]
  • (c) h(x)=xh(x) = \sqrt{x} on (0,4)(0, 4)

Solution.

(a) Yes. ff is a polynomial, so it’s continuous on the closed interval [−1,3][-1, 3].

(b) No. gg is not continuous at x=0x = 0, which is in the interval. (In fact gg has no absolute max or min there: it heads to ±∞\pm\infty near 00.)

(c) No. The interval is open. In fact hh has neither: its values get close to 00 and 22 but never reach them, since x=0x = 0 and x=4x = 4 are left out.

Forgetting the points where f′ is undefined. After solving f′(x)=0f'(x) = 0, look at the denominator of f′f' too. In Example 2, missing x=0x = 0 would miss the cusp.

Calling a point a critical point when it isn’t in the domain. For f(x)=1xf(x) = \dfrac{1}{x}, f′f' is undefined at x=0x = 0, but so is ff. So x=0x = 0 is not a critical point.

Assuming every critical point is a maximum or minimum. f(x)=x3f(x) = x^3 has a critical point at 00 and no extremum. You need a test to decide.

Using the EVT without checking both conditions. Write ”ff is continuous on the closed interval [a,b][a, b]” before using the theorem. Open intervals and discontinuities break it.

Giving the x-value when the question asks for the value. “The absolute maximum value” is f(c)f(c), a yy-value. “Where” or “at what xx” asks for cc. Read the question carefully.

1. (Warm-up) Find the critical point of f(x)=x2−8x+3f(x) = x^2 - 8x + 3.

Solution

f′(x)=2x−8=0f'(x) = 2x - 8 = 0 gives x=4x = 4.

2. (Warm-up) Find the critical points of f(x)=x3−12xf(x) = x^3 - 12x.

Solutionf′(x)=3x2−12=3(x−2)(x+2)f'(x) = 3x^2 - 12 = 3(x - 2)(x + 2)

The critical points are x=−2x = -2 and x=2x = 2.

3. (Warm-up) Does the EVT guarantee that f(x)=1x−3f(x) = \dfrac{1}{x - 3} has an absolute maximum on [0,2][0, 2]? On [2,4][2, 4]?

Solution

On [0,2][0, 2]: yes. The only discontinuity is at x=3x = 3, which is outside the interval, so ff is continuous on the closed interval [0,2][0, 2].

On [2,4][2, 4]: no. x=3x = 3 is inside the interval, so ff is not continuous there. (It actually has no maximum: it goes to ∞\infty as x→3+x \to 3^+.)

4. (Core) Find the critical points of f(x)=x4−4x3f(x) = x^4 - 4x^3.

Solutionf′(x)=4x3−12x2=4x2(x−3)f'(x) = 4x^3 - 12x^2 = 4x^2(x - 3)

The critical points are x=0x = 0 and x=3x = 3.

5. (Core) Find the critical points of f(x)=(x2−4)2/3f(x) = (x^2 - 4)^{2/3}.

Solution

By the chain rule:

f′(x)=23(x2−4)−1/3⋅2x=4x3x2−43f'(x) = \frac{2}{3}(x^2 - 4)^{-1/3} \cdot 2x = \frac{4x}{3\sqrt[3]{x^2 - 4}}
  • f′(x)=0f'(x) = 0 when x=0x = 0.
  • f′(x)f'(x) is undefined when x2−4=0x^2 - 4 = 0, so x=±2x = \pm 2. Since f(±2)=0f(\pm 2) = 0 is defined, these are in the domain.

The critical points are x=−2x = -2, x=0x = 0, and x=2x = 2.

6. (Core) Find the critical points of f(x)=x−2sin⁡xf(x) = x - 2\sin x on [0,2π][0, 2\pi] (radians).

Solution

f′(x)=1−2cos⁡xf'(x) = 1 - 2\cos x, which exists everywhere. Set it to 00:

cos⁡x=12⇒x=π3  or  x=5π3\cos x = \frac{1}{2} \quad\Rightarrow\quad x = \frac{\pi}{3} \ \text{ or } \ x = \frac{5\pi}{3}

7. (Core) Find the critical points of f(x)=xx2+9f(x) = \dfrac{x}{x^2 + 9}.

Solution

By the quotient rule:

f′(x)=(x2+9)(1)−x(2x)(x2+9)2=9−x2(x2+9)2f'(x) = \frac{(x^2 + 9)(1) - x(2x)}{(x^2 + 9)^2} = \frac{9 - x^2}{(x^2 + 9)^2}

The denominator is never 00, so f′f' always exists. f′(x)=0f'(x) = 0 when 9−x2=09 - x^2 = 0: x=−3x = -3 and x=3x = 3.

8. (Challenge) Show that f(x)=x−1x+2f(x) = \dfrac{x - 1}{x + 2} has no critical points, even though f′f' is undefined at x=−2x = -2.

Solutionf′(x)=(x+2)(1)−(x−1)(1)(x+2)2=3(x+2)2f'(x) = \frac{(x + 2)(1) - (x - 1)(1)}{(x + 2)^2} = \frac{3}{(x + 2)^2}

The numerator is 33, so f′(x)f'(x) is never 00. f′f' is undefined only at x=−2x = -2, but f(−2)f(-2) is undefined too (division by zero), so −2-2 is not in the domain and isn’t a critical point. So ff has no critical points.

9. (Challenge) Find constants aa and bb so that f(x)=x3+ax2+bxf(x) = x^3 + ax^2 + bx has critical points at x=−1x = -1 and x=3x = 3.

Solution

f′(x)=3x2+2ax+bf'(x) = 3x^2 + 2ax + b. We need f′(−1)=0f'(-1) = 0 and f′(3)=0f'(3) = 0, so f′f' must be 3(x+1)(x−3)3(x + 1)(x - 3) (the leading coefficient is 33):

3(x+1)(x−3)=3x2−6x−93(x + 1)(x - 3) = 3x^2 - 6x - 9

Matching coefficients: 2a=−62a = -6 and b=−9b = -9. So a=−3a = -3 and b=−9b = -9.

Check: f′(x)=3x2−6x−9f'(x) = 3x^2 - 6x - 9 gives f′(−1)=3+6−9=0f'(-1) = 3 + 6 - 9 = 0 and f′(3)=27−18−9=0f'(3) = 27 - 18 - 9 = 0. ✓