4.4 The Mean Value Theorem

Rolle's theorem and the Mean Value Theorem, the proof by subtracting the chord, the two hypotheses and counterexamples showing both are needed, and the three corollaries — a zero derivative gives a constant, equal derivatives differ by a constant, and the derivative's sign determines increase or decrease.

Subject: Calculus I · 67 slides · symbolic lesson

Open the interactive version of this deck

What this lesson covers

The lesson, slide by slide

1. The Mean Value Theorem

Title

Calculus I · Section 4.4

Every average rate is the actual rate at some instant

2. What this lesson gives you

Objectives

Section 4.3 found the extreme values of a function. This section turns that into the most used theorem of differential calculus: a bridge from what the derivative says at single points to what the function does over a whole interval.

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 329 — learning objectives 4.4.1 to 4.4.3

Since Chapter 3 you have been reading the derivative as a slope or a rate at a single point. You have also been using a fact that sounds obvious: if the derivative is positive, the function goes up. You have never proved it, and it is not as obvious as it sounds, because the derivative only describes the function one instant at a time.

The Mean Value Theorem is the tool that turns information about single points into information about whole intervals. You will meet it first in a special case, Rolle's theorem, then in general, then prove it, then test its hypotheses to destruction.

The last part is the payoff. Three corollaries follow in a few lines each, and Section 4.5 and the whole of integration rest on them. By the end you will know exactly why a positive derivative means an increasing function, and why every antiderivative comes with a plus C.

3. Before anything new: where can an extreme value hide?

Warm-up

Discussion prompt

Section 4.3 gave you two theorems about extreme values. Write down the Extreme Value Theorem and Fermat's theorem, each with its hypotheses, without looking back.

Write both statements before revealing them. You will need both, with their hypotheses, in the next five minutes, and it is much easier to follow a proof when its ingredients are already in your head.

Notice how the two fit together. The Extreme Value Theorem needs continuity on a closed interval and promises that the largest and smallest values are actually attained somewhere. Fermat's theorem needs a derivative at a point where an extremum occurs, and says that derivative is zero. What Fermat does not cover is an extremum at an endpoint, where the derivative need not be zero at all.

So the question Rolle's theorem answers is really this: when can you be sure that at least one extreme value is inside the interval rather than at an end? Keep that question in mind.

4. Rolle's theorem

Section

Part 1

5. Equal heights at the ends force a level tangent

Concept

Figure (svg): Three panels. In each a curve starts and ends at the same height over an interval from a to b. The first is an arch with a level tangent at its top, the second a valley with a level tangent at its bottom, the third a wave with level tangents at both its crest and its trough.

Whatever the curve does in between, if it comes back to its starting height it has to turn around, and where it turns the tangent is level.

Rolle's theorem (Theorem 4.4) — Let f be continuous over the closed interval from a to b and differentiable over the open interval, with f(a) equal to f(b). Then there is at least one c strictly between a and b where the derivative is zero.

\[ f(a) = f(b) \;\Longrightarrow\; f'(c) = 0 \text{ for some } c \in (a,b) \]

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, pp. 329-330 — Theorem 4.4 and Figure 4.21

Look at the three panels. In each, the curve starts and finishes at the same height. The first goes up and has to come back down, so it turns at the top. The second goes down and has to come back up, so it turns at the bottom. The third does both, and turns twice. Wherever a smooth curve turns, its tangent is level.

That is all Rolle's theorem says, and the picture is very nearly the proof. The formal statement adds the conditions that make the picture trustworthy: no breaks in the curve on the closed interval, and no corners or vertical tangents inside it.

Pay attention to what the theorem does not say. It does not say how many level points there are, as the third panel shows, and it does not say where they are. It promises at least one, strictly inside.

6. Reading Theorem 4.4 piece by piece

Notation

Annotate

On: \( f(a) = f(b) \;\Longrightarrow\; \exists\, c \in (a,b):\; f'(c) = 0 \)

  • f continuous on the closed interval [a, b] and differentiable on the open interval (a, b). Both must be checked, every time.
  • The curve ends at the height it started. This is the condition the Mean Value Theorem will drop.
  • There exists at least one such c. Maybe several; the theorem does not say how many or where.
  • Strictly inside. An endpoint never counts, even if the derivative happens to be zero there.
  • A level tangent: the curve is momentarily neither rising nor falling.

Step through the notes one by one. The first note is the one people forget: the compact formula on the slide hides two hypotheses, continuity on the closed interval and differentiability on the open one. A complete answer checks all three conditions in writing, not just the equal end values.

The phrase there exists is doing precise work. It promises at least one point and makes no claim about uniqueness. When you solve for c, you may find one value, several, or, as later in this lesson, you may find the equation has more solutions than belong in the interval.

The last two notes go together. The point c must lie strictly inside, and the tangent there is horizontal. If a zero of the derivative falls exactly at an endpoint or outside the interval, it is not one of Rolle's points.

7. Proving Rolle: three cases

Concept

Call the common end value k. Every function on the interval does one of three things.

\[ k = f(a) = f(b) \]

Case 1: the function never leaves k. A constant function has zero derivative everywhere, so any interior c will do.

\[ f(x) = k \text{ for all } x \in (a,b) \;\Longrightarrow\; f'(x) = 0 \]

Case 2: somewhere the function rises above k. The Extreme Value Theorem gives an absolute maximum, which is bigger than k, so it cannot be at either end. It is interior, and Fermat's theorem finishes the job.

Figure (svg): A curve starting and ending at height k over an interval from a to b, rising above the dashed line y equals k in between. Its highest point sits strictly inside the interval, where a short level tangent is drawn.

Once the curve rises above the endpoint height anywhere, its highest point cannot be at an end. An interior maximum of a differentiable function has a level tangent: that is Fermat's theorem.

\[ f(x) > k \text{ somewhere} \;\Longrightarrow\; \text{max} > k \;\Longrightarrow\; c \in (a,b) \;\Longrightarrow\; f'(c) = 0 \]

Case 3: somewhere the function dips below k. The same argument with the absolute minimum.

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 330 — proof of Theorem 4.4

The proof is short because the hard work was done in Section 4.3. Name the common end value k, and ask what the function does in between. It either stays at k the whole time, or it rises above k somewhere, or it dips below k somewhere. At least one of these must happen.

If it stays at k, the function is constant and its derivative is zero everywhere, so any interior point works. If it rises above k, look at the figure. The Extreme Value Theorem, which needs continuity on the closed interval, says there is a highest point. That highest value is bigger than k, so it cannot sit at either end, where the value is only k. It is inside, and Fermat's theorem, which needs the derivative to exist there, says the derivative is zero.

The dipping case is the same argument with the lowest point. Notice that each hypothesis was used exactly once: continuity for the Extreme Value Theorem, differentiability for Fermat, and equal end values to push the extreme inside.

8. Put the proof of Rolle's theorem in order

Ranking

Put in order

Drag the steps of the proof into the order that makes each one follow from the ones before.

  1. Let k be the common value f(a) = f(b)
  2. If f(x) = k for every x, then f′ = 0 everywhere and any c works
  3. Otherwise f goes above k (or below k) somewhere inside
  4. By the Extreme Value Theorem, f has an absolute max (or min) on [a, b]
  5. That extreme value is not k, so it is not at an endpoint
  6. By Fermat's theorem at that interior point, f′(c) = 0

Why: The constant case has to be dealt with first, because only when f leaves k can you be sure the extreme is interior. Continuity feeds the Extreme Value Theorem; differentiability inside feeds Fermat's theorem.

Arrange the steps before checking. The logic has a definite shape: set up k, dispose of the easy constant case, then use the fact that the function leaves k to force an interior extreme.

The step people most often put in the wrong place is the constant case. It has to come before the extreme-value argument, because that argument only works once you know the function goes above or below k somewhere. Without that, both extremes could sit at the endpoints, and Fermat's theorem would have nothing to say.

Also notice the order of the last three steps: existence of the extreme first, then its location, then its derivative. Each one needs the one before.

9. Example 4.14(a): a parabola from minus 2 to 0

Worked example

Verify that the function satisfies Rolle's hypotheses on the interval, then find every c where the derivative is zero.

\[ f(x) = x^2 + 2x \quad \text{on } [-2, 0] \]

Check continuity and differentiability

Why: A polynomial is continuous and differentiable everywhere, so in particular on this interval.

\[ f \text{ is a polynomial} \]

Check the end values

Why: Substitute both endpoints.

\[ f(-2) = 4 - 4 = 0, \qquad f(0) = 0 \]

Differentiate

Why: Power rule, term by term.

\[ f'(x) = 2x + 2 = 2(x + 1) \]

Set the derivative to zero

Why: Rolle promises at least one solution inside.

\[ 2(c + 1) = 0 \;\Longrightarrow\; c = -1 \]

Figure (svg): The parabola y equals x squared plus 2x from x equals minus 2 to 0, touching the x-axis at both ends, with a level tangent at its lowest point, x equals minus 1, height minus 1.

Equal heights at minus 2 and 0, so Rolle promises a level tangent. It sits at the vertex, x equals minus 1.

Check that c is inside and the slope is zero

Why: Minus 1 lies strictly between minus 2 and 0, and the derivative there is 2 times 0.

\[ -2 < -1 < 0, \qquad f'(-1) = 2(0) = 0 \]

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 331 — Example 4.14a

This is the model for every Rolle problem: check the hypotheses in writing, then solve for c, then confirm c is inside. The hypotheses are quick here because the function is a polynomial, but the end values must actually be computed. Both come out zero.

Setting the derivative to zero gives a single point, minus 1. The figure shows it is the vertex of the parabola, the lowest point on the interval, which is exactly what the proof predicted: Rolle's point is an interior extreme.

The check at the end has two parts, and both matter. Minus 1 must be strictly between minus 2 and 0, and the derivative there must be zero. A value that satisfies the equation but lies outside the interval would not count.

10. Example 4.14(b): a cubic with two level tangents

Worked example

\[ f(x) = x^3 - 4x \quad \text{on } [-2, 2] \]

Check continuity and differentiability

Why: Again a polynomial, so both hold everywhere.

\[ f \text{ is a polynomial} \]

Check the end values

Why: Factor first: the cubic is x times x squared minus 4.

\[ f(-2) = -8 + 8 = 0, \qquad f(2) = 8 - 8 = 0 \]

Differentiate

Why: Power rule.

\[ f'(x) = 3x^2 - 4 \]

Set the derivative to zero

Why: Isolate x squared.

\[ 3c^2 = 4 \;\Longrightarrow\; c^2 = \frac{4}{3} \]

Take both square roots

Why: Two candidates, one on each side of zero.

\[ c = \pm \frac{2}{\sqrt 3} \approx \pm 1.1547 \]

Figure (svg): The cubic y equals x cubed minus 4x from minus 2 to 2, zero at both ends, with a level tangent at its local maximum near x equals minus 1.15 and another at its local minimum near x equals 1.15.

Rolle guarantees at least one level tangent; this cubic has two, one at each turning point, and both lie inside the interval.

Check both are inside and both work

Why: 1.1547 is less than 2, so both points are strictly inside, and three times four thirds minus 4 is zero.

\[ f'\left(\pm \tfrac{2}{\sqrt3}\right) = 3\cdot\tfrac43 - 4 = 0, \qquad |c| \approx 1.155 < 2 \]

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, pp. 331-332 — Example 4.14b and Figure 4.24

Same routine, a richer answer. The cubic is zero at both ends, so Rolle applies, and setting the derivative to zero gives c squared equal to four thirds. That has two solutions, plus and minus two over root 3, about plus and minus 1.155.

Both are inside the interval, so both are answers. The figure shows why: the curve rises to a peak on the left and falls to a valley on the right before coming back to zero, so it turns twice. The theorem promised one level tangent; the function delivers two.

When you take a square root in a problem like this, always write down both signs and then test each against the interval. Here both survive. On the trap slide later in this part, one of them will not.

11. Checkpoint 4.14: a parabola from 1 to 3

Worked example

\[ f(x) = 2x^2 - 8x + 6 \quad \text{on } [1, 3] \]

Check continuity and differentiability

Why: A polynomial.

\[ f \text{ is a polynomial} \]

Check the end values

Why: Both come out zero.

\[ f(1) = 2 - 8 + 6 = 0, \qquad f(3) = 18 - 24 + 6 = 0 \]

Differentiate

Why: Power rule.

\[ f'(x) = 4x - 8 \]

Solve for the level point

Why: Divide by 4.

\[ 4c - 8 = 0 \;\Longrightarrow\; c = 2 \]

Figure (svg): The parabola y equals 2x squared minus 8x plus 6 from x equals 1 to 3, zero at both ends, dipping to minus 2 at x equals 2 where a level tangent is drawn.

The vertex of a parabola is the only point where its tangent is level, and here it lands exactly in the middle of the interval.

Check the answer

Why: Two is strictly between 1 and 3, and the derivative there is zero; the point is the vertex, height minus 2.

\[ f'(2) = 8 - 8 = 0, \qquad f(2) = 8 - 16 + 6 = -2 \]

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 332 — Checkpoint 4.14

Try this one on your own before reading the steps. Every line has the same shape as Example 4.14(a): polynomial, so smooth; end values both zero; derivative four x minus eight; one solution.

The answer, c equal to 2, is the midpoint of the interval and the vertex of the parabola. That is no coincidence. A parabola is symmetric about its vertex, so if it has equal values at two points, the vertex is halfway between them.

The check computes both the derivative and the height at c. The height, minus 2, is not needed for the answer, but it confirms from the figure that you have found the bottom of the dip, the interior minimum the proof guarantees.

12. Trap: reporting a root outside the interval

Trap

The trap

Find the Rolle points of this cubic on the interval from 0 to root 3:

\[ f(x) = x^3 - 3x \]

\[ 3c^2 - 3 = 0 \]

\[ c = 1 \text{ and } c = -1 \quad ? \]

Wrong. One of those is not inside the interval.

The fix

The ends agree, since both values are zero, so Rolle applies. But the theorem only speaks about points strictly inside. Minus 1 is a genuine zero of the derivative that lies outside, so it is not an answer.

\[ -1 \notin (0, \sqrt3) \]

\[ c = 1 \text{ only} \]

The algebra on the left is correct: the derivative three x squared minus three is zero at plus 1 and at minus 1. The mistake is in the last line, which reports both as Rolle points without asking where they are.

The interval runs from 0 to root 3, about 1.73. Plus 1 is inside it. Minus 1 is not, so it has nothing to do with this problem. The theorem only ever speaks about points strictly inside the interval you were given.

This slip is easy to make because solving the equation feels like the end of the problem. Make it a habit that the last line of every Rolle or Mean Value calculation is a check of each candidate against the interval.

13. Does Rolle promise a level tangent for |x| minus 1?

Prediction

\[ f(x) = |x| - 1 \quad \text{on } [-1, 1] \]

Predict first

The ends match: f(−1) = 0 = f(1), and the function is continuous. What does Rolle's theorem say?

  • It guarantees a c with f′(c) = 0, at x = 0
  • It guarantees a c somewhere else
  • Nothing: a hypothesis fails, and in fact no such c exists
  • Nothing, but a c exists anyway

Correct: Nothing: a hypothesis fails, and in fact no such c exists

Why: The absolute value has a corner at 0, so f is not differentiable there, and 0 is inside the interval. The theorem is silent. Checking directly, the derivative is −1 to the left of 0 and +1 to the right, so it is never zero: the conclusion really does fail.

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 330 — Figure 4.22

Commit to an option before revealing. Two of the three hypotheses hold, the ends match and the graph is unbroken, which makes it tempting to apply the theorem anyway.

The third hypothesis fails at a single point. The absolute value has a sharp corner at 0, so there is no derivative there, and 0 is inside the interval. That is enough to switch the theorem off. It then says nothing either way.

In this case you can check directly that the conclusion really fails. To the left of the corner the slope is minus 1, to the right it is plus 1, and there is no point where it is zero. The corner is where the function turns, but it turns abruptly, without a tangent.

14. A single corner is enough to break it

Picture it

Figure (svg): The V-shaped graph of y equals absolute value of x minus 1 on the interval from minus 1 to 1: zero at both ends, a sharp corner at the origin at height minus 1, slope minus 1 on the left and plus 1 on the right. No level tangent exists.

Equal heights at the ends, but the slope jumps from minus 1 to plus 1 without ever passing through zero. One missing derivative is enough to lose the conclusion.

The endpoints are at the same height and the curve is unbroken, yet no tangent is level. The slope never passes through zero: it jumps from minus 1 to plus 1 at the corner, where there is no tangent at all.

This picture is worth remembering as the standard counterexample for differentiability. The function comes back to its starting height, just as Rolle requires, but it turns at a corner instead of along a smooth curve.

Follow the slope from left to right. It is minus 1 all the way to the corner, then suddenly plus 1. A smooth curve would have to pass through every slope in between, including zero, as it turned. The corner lets the function skip that entire range of slopes.

So differentiability is not a technicality. It is precisely the condition that forces the turn to be gradual, and a gradual turn is where the level tangent lives.

15. Does Rolle's theorem apply?

Sorting

Sort into buckets

Check all three hypotheses: continuous on [a, b], differentiable on (a, b), f(a) = f(b).

All hypotheses hold
x² − 4x + 3 on [1, 3]; sin x on [0, π]; cos x on [0, 2π]
A hypothesis fails
|x| on [−1, 1]; x² on [0, 2]; 1/x² on [−1, 1]
yes
x² − 4x + 3 is 0 at both 1 and 3 (level point c = 2); sin x is 0 at 0 and π (c = π/2); cos x is 1 at 0 and 2π (c = π). All are smooth everywhere.
no
|x| has a corner at 0; x² takes the values 0 and 4, which differ; 1/x² is not even defined at 0, so continuity fails inside the interval.

For each function, run through the three checks in order: continuous on the closed interval, differentiable on the open interval, equal end values. A single failure puts it in the second bucket.

The three that pass are worth finding c for, as practice. The quadratic gives c equal to 2, sine on the interval from 0 to pi gives pi over 2, and cosine on the interval from 0 to 2 pi gives pi, the bottom of its trough.

The failures fail for three different reasons, and that is the point of the exercise. The absolute value has a corner. The squaring function is smooth but its end values differ, 0 and 4. And one over x squared is not even defined at 0, so continuity fails inside the interval. Even though its end values match, the theorem cannot be used.

16. The Mean Value Theorem

Section

Part 2

17. Tilt the picture: some tangent is parallel to the chord

Concept

Drop the requirement that the ends match. The chord joining the two end points now has a slope, and the promise becomes a tangent with that same slope.

Figure (svg): The curve y equals x cubed minus 3x from a equals minus 2 to b equals 2.2, with the dashed chord joining its end points and two short tangent lines, at c one near minus 1.22 and c two near 1.22, both parallel to the chord.

Tilt Rolle's picture: the chord is no longer level, and the promise becomes a tangent parallel to the chord. This curve has two such points.

Mean Value Theorem (Theorem 4.5) — Let f be continuous over the closed interval from a to b and differentiable over the open interval. Then there is at least one c strictly between a and b where the derivative equals the slope of the chord.

\[ f'(c) = \frac{f(b) - f(a)}{b - a} \quad \text{for some } c \in (a,b) \]

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, pp. 332-333 — Theorem 4.5 and Figure 4.25

Imagine taking Rolle's picture and tilting the page. The level line joining the end points becomes a sloping chord, and the level tangent becomes a tangent parallel to that chord. That is the Mean Value Theorem: drop the equal-ends condition, and the conclusion becomes a tangent parallel to the chord.

In the figure the chord runs from a to b, and two points, c one and c two, have tangents with exactly the chord's slope. One way to find them in your head is to slide a copy of the chord up or down, keeping it parallel, until it just touches the curve.

The formula says the same thing in symbols. The right side is the slope of the chord; the left is the slope of the tangent at c. The hypotheses are the same as Rolle's, minus the requirement on the end values.

18. Reading Theorem 4.5 piece by piece

Notation

Annotate

On: \( f'(c) = \frac{f(b) - f(a)}{b - a}, \quad c \in (a,b) \)

  • The instantaneous rate of change at one point: the slope of the tangent there.
  • The total change in the output over the interval: the rise of the chord.
  • The length of the interval: the run of the chord. The fraction is the AVERAGE rate of change.
  • Strictly inside, at least one, location unknown. When f(a) = f(b) the right side is 0 and this is exactly Rolle's theorem.

\[ f(b) - f(a) = f'(c)(b - a) \]

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 333 — Theorem 4.5

The numerator and denominator on the right are the rise and run of the chord, so the fraction is the average rate of change of f over the interval. The left side is an instantaneous rate at one point. The theorem says the average is achieved as an instantaneous rate somewhere inside.

Read the last note carefully. When the end values are equal the fraction is zero, and the theorem becomes exactly Rolle's theorem. So Rolle is not a different result; it is the level-chord case of this one.

The line underneath is the same theorem with the fraction cleared. This increment form is how the theorem is most often used, because it expresses the change in f directly: the change equals some derivative value times the length of the interval. Keep it in view; the corollaries are built on it.

19. Average speed is some instant's speed

Intuition

Drive for one hour down a straight road and cover 45 miles. Your position is a differentiable function of time, so the theorem applies on the interval from 0 to 1 hour.

\[ v(c) = s'(c) = \frac{s(1) - s(0)}{1 - 0} = 45 \text{ mph} \]

You may have stopped at lights and sped up on the open road, but at some instant your speedometer read exactly 45. The theorem does not say when.

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 334 — the car example after Example 4.15

This is the rate reading of the theorem, and it is the one that makes it feel inevitable. Position is a function of time, velocity is its derivative, and the chord's slope is the average velocity over the trip.

Think about why it must be true. If your speed had been below 45 the entire hour, you would have covered less than 45 miles. If it had been above 45 the entire time, you would have covered more. Since you covered exactly 45, your speed cannot have stayed on one side of 45, and a speed that changes smoothly must pass through 45.

Notice the assumption that position is differentiable, which is physically reasonable: a car cannot teleport, and it cannot change speed instantaneously in a way that leaves no velocity defined. Those are exactly the two hypotheses.

20. Existence without location

Intuition

This is the third theorem of its kind. Each promises that a point exists, and none says where; finding the point, when you need it, is a separate calculation.

theoremhypothesispromises a point where
Intermediate Value (§2.4)f continuous on [a, b]f takes any value between f(a) and f(b)
Extreme Value (§4.3)f continuous on [a, b]f is largest (and smallest)
Mean Value (§4.4)continuous on [a, b], differentiable on (a, b)the tangent is parallel to the chord

The theorem earns its keep not by locating c but by what you can deduce from knowing c is there. Part 6 of this lesson is built entirely on that.

You have now met three theorems with the same personality. Each takes a hypothesis about continuity, sometimes with differentiability, and promises that some special point exists, without telling you where it is.

That can feel unsatisfying the first time, because you are used to calculations that produce a number. But look at what the Intermediate Value Theorem let you do in Chapter 2: prove that an equation has a root without solving it. The Mean Value Theorem is used the same way. Most of the time nobody needs to know where c is; it is enough that it exists and that whatever is true of the derivative everywhere is true of it.

When a problem does ask you to find c, that is a separate calculation: set the derivative equal to the chord slope and solve. The theorem only guarantees that the equation has a solution in the interval.

21. Example 4.15: the square root from 0 to 9

Worked example

Show the Mean Value Theorem applies and find the point it guarantees.

\[ f(x) = \sqrt{x} \quad \text{on } [0, 9] \]

Check the hypotheses

Why: The square root is continuous on the closed interval and differentiable wherever x is positive, which covers the open interval.

\[ \text{continuous on } [0,9], \quad \text{differentiable on } (0,9) \]

Compute the chord's slope

Why: Rise over run between the end points.

\[ \frac{f(9) - f(0)}{9 - 0} = \frac{\sqrt9 - \sqrt0}{9} = \frac{3}{9} = \frac13 \]

Differentiate

Why: Power rule with exponent one half.

\[ f'(x) = \frac{1}{2\sqrt{x}} \]

Set the derivative equal to the slope

Why: This is the theorem's equation.

\[ \frac{1}{2\sqrt c} = \frac13 \]

Solve for the square root of c

Why: Cross-multiply.

\[ 2\sqrt c = 3 \;\Longrightarrow\; \sqrt c = \frac32 \]

Square

Why: Both sides are positive.

\[ c = \frac94 \]

Figure (svg): The curve y equals square root of x from 0 to 10, the dashed chord from the origin to the point 9, 3, and the tangent at the point 9 over 4, 3 over 2, drawn parallel to the chord.

The tangent at nine quarters has slope one third, the same as the chord. The theorem only needs a derivative inside the interval, so the vertical tangent at the origin does no harm.

Check by substituting back

Why: The root of nine quarters is three halves, and one over twice that is one third: the chord's slope.

\[ f'\left(\tfrac94\right) = \frac{1}{2 \cdot \tfrac32} = \frac13 \;\checkmark, \qquad 0 < \tfrac94 < 9 \]

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 334 — Example 4.15 and Figure 4.27

The hypotheses need a moment of thought here, because the square root is not differentiable at 0: its tangent there is vertical. The theorem only asks for differentiability on the open interval, which excludes 0, so it applies.

The rest is the standard routine. The chord rises from 0 to 3 over a run of 9, so its slope is one third. Setting one over twice root c equal to one third gives root c equal to three halves, and squaring gives nine quarters.

In the figure the tangent at nine quarters runs exactly parallel to the chord. Notice that nine quarters is well to the left of the midpoint, four and a half. The square root is steep near 0 and flat further out, so its average steepness is reached early. The check substitutes back and confirms both the slope and that c lies inside.

22. Exercise 161: the cube from 0 to 2

Worked example

Find every c between 0 and 2 with f(2) minus f(0) equal to f prime of c times 2.

\[ f(x) = x^3 \quad \text{on } [0, 2] \]

Compute the chord's slope

Why: The cube is 8 at the right end and 0 at the left.

\[ \frac{f(2) - f(0)}{2 - 0} = \frac{8 - 0}{2} = 4 \]

Set the derivative equal to it

Why: Power rule gives three x squared.

\[ 3c^2 = 4 \]

Solve

Why: Only the positive root is in the interval.

\[ c^2 = \frac43 \;\Longrightarrow\; c = \frac{2}{\sqrt3} \approx 1.1547 \]

Figure (svg): The curve y equals x cubed from 0 to 2, the dashed chord from the origin to the point 2, 8, with slope 4, and a tangent of the same slope at x equals 2 over root 3, about 1.155.

The guaranteed point is not the midpoint. The cube is steep near 2 and flat near 0, so the average steepness is reached a little past the middle.

Check the answer and discard the other root

Why: Minus 1.1547 also solves the equation but lies outside; the positive root is inside and gives slope 4.

\[ 3\left(\tfrac{2}{\sqrt3}\right)^2 = 3 \cdot \tfrac43 = 4 \;\checkmark, \qquad 0 < 1.1547 < 2 \]

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 338 — Exercise 161

This exercise is phrased in the increment form, f of 2 minus f of 0 equals f prime of c times 2, but it is the same computation. The chord's slope is 8 over 2, which is 4.

Setting three c squared equal to 4 gives two roots, plus and minus two over root 3. Only the positive one, about 1.155, is inside the interval, so it is the only answer. This is the trap from the Rolle part in a new setting.

Compare with the square root on the previous slide. The cube is flat near 0 and steep near 2, the opposite shape, so this time the guaranteed point sits to the right of the midpoint. You can often predict which side of the middle c will land on just by looking at which end of the curve is steeper.

23. Exercise 164: finish the equation for c

Fill the middle

\[ f(x) = 1 + x + x^2 \quad \text{on } [0, 2] \]

Fill in the blanks

\frac31 = \frac______ = ___ = 1 + 2c \;\Longrightarrow\; c = ___

Why: The chord rises from 1 to 7 over a run of 2, so its slope is 3. The derivative is 1 + 2c, and 1 + 2c = 3 gives c = 1: the midpoint, as it always is for a parabola.

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 338 — Exercise 164

Fill in both boxes before checking. The first is the chord's slope: the function goes from 1 to 7, a rise of 6, over a run of 2. The second is the solution of the equation the theorem produces.

The derivative of one plus x plus x squared is one plus two x, so the equation is one plus two c equals 3, and c is 1. That is the midpoint of the interval from 0 to 2.

Every parabola does this. Its derivative is a linear function, and a linear function takes its average value over an interval at the midpoint. So for any quadratic on any interval, the Mean Value point is the middle. That is a handy check on your arithmetic whenever the function is a parabola.

24. Exercise 159: how many points for a wiggly curve?

Estimation

\[ f(x) = x^2 \cos(\pi x) \quad \text{on } [-2, 2] \]

Predict first

Both ends are at height 4, so the chord is level. Without computing, how many c in (−2, 2) have f′(c) equal to the chord's slope?

  • 1
  • 3
  • 5
  • 7

Correct: 5

Why: A level chord means you want the level tangents, one at every turning point. The curve is even and turns at x = 0, at about ±0.343 and at about ±1.160: five points. The theorem only promised one.

Figure (svg): The curve y equals x squared cos pi x on the interval from minus 2 to 2: height 4 at both ends, so the chord is level; five short level tangents mark the turning points at about minus 1.16, minus 0.34, 0, 0.34 and 1.16.

Level chord, so the Mean Value Theorem is Rolle's theorem here, and the curve turns five times inside the interval.

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 338 — Exercise 159

Make your guess before revealing. You do not need to compute anything: the ends are both at height 4, so the chord is level, and you are counting level tangents. That means counting the turning points of the curve inside the interval.

The figure shows the answer: five. The function is even, symmetric about the vertical axis, so the turning points come in mirror pairs, plus one at the centre. They sit at 0, at about plus and minus 0.343, and at about plus and minus 1.160.

The lesson here is about what the theorem guarantees. It said at least one. The actual function has five. The theorem is a floor, not a count, and you should never read at least one as exactly one.

25. Exercise 179: a logarithm from 0 to e minus 1

Worked example

\[ f(x) = \ln(x + 1) \quad \text{on } [0, e - 1] \]

Check the hypotheses

Why: The argument x plus 1 is at least 1 on this interval, so the logarithm is continuous and differentiable throughout.

\[ x + 1 \ge 1 > 0 \text{ on } [0, e - 1] \]

Compute the chord's slope

Why: The log of e is 1 and the log of 1 is 0.

\[ \frac{\ln e - \ln 1}{(e - 1) - 0} = \frac{1}{e - 1} \]

Differentiate

Why: Chain rule, with inner derivative 1.

\[ f'(x) = \frac{1}{x + 1} \]

Set equal and solve

Why: Equal fractions with equal numerators have equal denominators.

\[ \frac{1}{c + 1} = \frac{1}{e - 1} \;\Longrightarrow\; c + 1 = e - 1 \;\Longrightarrow\; c = e - 2 \]

Figure (svg): The curve y equals the natural log of x plus 1 from 0 to e minus 1, rising from 0 to 1, with the dashed chord of slope about 0.582 and a parallel tangent at x equals e minus 2, about 0.718.

The interval was chosen so that the rise is exactly 1; the guaranteed point then comes out as the tidy number e minus 2.

Check numerically

Why: e minus 2 is about 0.718, inside the interval, and both slopes are about 0.582.

\[ f'(0.71828) = \frac{1}{1.71828} \approx 0.58198 = \frac{1}{e-1} \]

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 338 — Exercise 179

The interval looks strange until you evaluate the function at its ends. At 0 the logarithm of 1 is zero, and at e minus 1 the logarithm of e is one. So the rise is exactly 1, and the chord's slope is one over e minus 1, about 0.582.

The derivative of the log of x plus 1 is one over x plus 1. Setting that equal to one over e minus 1 means the denominators must agree, so c plus 1 equals e minus 1, and c is e minus 2, about 0.718.

Once again c sits left of the midpoint, which is about 0.859, because the logarithm is steepest at the left end. The numerical check confirms both slopes to five decimal places.

26. Rates in motion

Section

Part 3

27. Example 4.16: the rock dropped from 100 ft

Worked example

A rock falls from 100 feet; its height after t seconds is given below. Find (a) when it lands, (b) its average velocity, (c) the instant when its velocity equals the average.

\[ s(t) = -16t^2 + 100 \]

(a) Set the height to zero

Why: Landing means height zero.

\[ -16t^2 + 100 = 0 \;\Longrightarrow\; t^2 = \frac{100}{16} \]

Keep the positive root

Why: Time after release is not negative.

\[ t = \frac{10}{4} = \frac52 \text{ s} \]

(b) Average velocity over the fall

Why: Change in position over elapsed time.

\[ v_{\text{avg}} = \frac{s(5/2) - s(0)}{5/2 - 0} = \frac{0 - 100}{5/2} = -40 \text{ ft/s} \]

(c) Differentiate the position

Why: Velocity is the derivative of position.

\[ s'(t) = -32t \]

Apply the theorem and solve

Why: s is a polynomial, so the hypotheses hold on the interval from 0 to five halves.

\[ -32c = -40 \;\Longrightarrow\; c = \frac{40}{32} = \frac54 \text{ s} \]

Figure (svg): The position of a dropped rock, s equals minus 16 t squared plus 100, falling from 100 feet at t equals 0 to the ground at t equals 5 halves. The dashed chord joins the start and the landing; a parallel tangent touches the curve at t equals 5 quarters, height 75.

Halfway through the fall in time, the rock is moving at exactly its average velocity, minus 40 feet per second.

Check the velocity at c

Why: Minus 32 times five quarters is minus 40, the average, and five quarters is inside the fall.

\[ s'\left(\tfrac54\right) = -32 \cdot \tfrac54 = -40, \qquad 0 < \tfrac54 < \tfrac52 \]

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 335 — Example 4.16 and Figure 4.28

This example reads the theorem as a statement about motion. Part (a) finds the landing time by setting the height to zero, which gives five halves of a second after throwing away the negative root. Part (b) is the chord's slope in physical units: the rock drops 100 feet in two and a half seconds, an average velocity of minus 40 feet per second. The minus sign means downward.

Part (c) is where the theorem enters. Position is a polynomial in time, so the hypotheses hold, and some instant has velocity exactly minus 40. The derivative of the position is minus 32 t, and solving gives five quarters of a second.

In the figure, the tangent at five quarters is parallel to the chord from release to landing. Five quarters is exactly half of five halves: halfway through the fall in time, though not halfway down in distance, since the rock is still at 75 feet.

28. Checkpoint 4.15: the ball dropped from 200 ft

Worked example

\[ s(t) = -16t^2 + 200 \]

Find the landing time

Why: Solve s equal to zero and keep the positive root.

\[ t^2 = \frac{200}{16} = 12.5 \;\Longrightarrow\; T = \frac{5}{\sqrt2} = \frac{5\sqrt2}{2} \approx 3.536 \text{ s} \]

Average velocity

Why: The height drops by 200 over the whole fall.

\[ v_{\text{avg}} = \frac{0 - 200}{5\sqrt2/2} = -\frac{400}{5\sqrt2} = -40\sqrt2 \approx -56.57 \text{ ft/s} \]

Differentiate

Why: Same form as the example.

\[ s'(t) = -32t \]

Set equal and solve

Why: Divide both sides by minus 32.

\[ -32c = -40\sqrt2 \;\Longrightarrow\; c = \frac{5\sqrt2}{4} \approx 1.768 \text{ s} \]

Check against the landing time

Why: c is exactly half of T, the same midpoint pattern as the 100-foot rock, and the speed there matches.

\[ \frac{T}{2} = \frac{5\sqrt2}{4}, \qquad s'(1.768) = -32(1.768) \approx -56.57 \]

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 335 — Checkpoint 4.15

Work through this one yourself before revealing the steps. The structure is the same as the example, but the landing time is no longer a tidy fraction: t squared equals 12.5, so the ball lands at five over root 2 seconds, about 3.536.

The average velocity is minus 200 divided by that time, which simplifies to minus 40 root 2, about minus 56.57 feet per second. Setting minus 32 c equal to it gives c equal to five root 2 over 4, about 1.768 seconds.

The check spots a pattern: c is exactly half the landing time, just as it was for the 100-foot drop. That is the parabola fact from Exercise 164 again. Position under constant gravity is quadratic in time, so the moment when the velocity equals the average is always the midpoint of the time interval.

29. Two cameras, 39 miles, 36 minutes

Worked example

A car passes one camera and, 36 minutes later, a second camera 39 miles further along. The limit is 60 mph. Can the driver be ticketed without anyone measuring an instantaneous speed?

Put time in hours

Why: Thirty-six minutes is six tenths of an hour.

\[ t \in [0, 0.6], \qquad s(0) = 0, \quad s(0.6) = 39 \]

Average speed

Why: Distance over time.

\[ \frac{s(0.6) - s(0)}{0.6 - 0} = \frac{39}{0.6} = 65 \text{ mph} \]

Apply the theorem

Why: A car's position is a differentiable function of time.

\[ s'(c) = 65 \text{ for some } c \in (0, 0.6) \]

Figure (svg): A car's distance in miles against time in hours, from 0 miles at time 0 to 39 miles at time 0.6. The curve wiggles around the dashed chord of slope 65; tangent segments at t equals 0.15 and 0.45 are parallel to the chord.

The driver sped up and slowed down, but at the two marked instants the speedometer read exactly the average, 65 miles per hour.

Check the conclusion against the limit

Why: At that instant the speed was 65, which is over 60. Nothing was measured but two positions and two times.

\[ s'(c) = 65 > 60 \]

This is the theorem at work in the world. Nobody measured the car's speed at any instant. All anyone knows is two positions and two times. Yet the conclusion is a certain fact about an instantaneous speed.

Convert the time to hours first, because the speed limit is in miles per hour. Then the average speed is 39 divided by 0.6, which is 65. A car's position is a differentiable function of time, so the theorem guarantees an instant at which the speedometer read exactly 65, over the limit.

The figure makes it concrete. The driver sped up and slowed down, so the curve wiggles around the chord, but at the two marked instants the tangent is parallel to the chord. Notice what the theorem does not give: it does not say when, and it does not say how long the car was over 60. It proves only that at least one instant exists.

30. Find the error: the theorem asked to say when

Error analysis

Annotate

On: \( \frac{39}{0.6} = 65 \;\Longrightarrow\; s'(0.3) = 65 \text{ (halfway through)} \)

  • Correct: 39 miles in 0.6 hours is an average of 65 mph.
  • The error. The theorem guarantees SOME c in (0, 0.6); it does not say c is the midpoint. In the figure on the previous slide the speed at t = 0.3 is well below 65.
  • For the falling rock c WAS the midpoint, because a quadratic has a linear derivative. That is a property of quadratics, not of the theorem.

Read the claimed argument before revealing the notes. The average is computed correctly, so the arithmetic is not the problem.

The problem is the jump from some instant to the midpoint. The theorem is an existence statement: it guarantees a c in the interval and says nothing about where. Look back at the camera figure. At the midpoint, three tenths of an hour, the car was going well under 65; the instants where it read exactly 65 were at 0.15 and 0.45 hours.

The confusion is understandable, because the falling rock did hit its average velocity exactly at the midpoint of its time interval. But that came from the position being quadratic, which makes the velocity linear. A car's position is not a quadratic, and in general c can be anywhere in the interval.

31. Average-speed cameras on a motorway

Real world

Discussion prompt

A vehicle enters a 12-mile camera zone at 14:00:00 and leaves at 14:10:00. The limit is 65 mph. The driver says: 'You never saw me speeding.' What can the authority prove, and what can it not claim?

Write your answer before revealing. Think about what the authority actually recorded, and what the theorem lets it deduce from that.

Twelve miles in ten minutes is 72 miles per hour. Since position is differentiable, there was an instant at which the vehicle's speed was exactly 72, and 72 is over 65. That is a proof, not an estimate, which is why average-speed enforcement is mathematically sound.

The limits of the conclusion matter just as much. The theorem does not locate the instant, so a notice cannot say where the offence happened. And it gives no upper bound on the speed: the driver might have reached 90 somewhere. It guarantees one value, the average, and nothing about the extremes.

32. How large can f(2) be?

Estimation

\[ f(0) = -3, \qquad f'(x) \le 5 \text{ for every } x \]

Predict first

f is differentiable everywhere. What is the largest value f(2) could possibly have?

  • 5
  • 7
  • 10
  • There is no limit

Correct: 7

Why: By the theorem, f(2) − f(0) = f′(c)(2 − 0) for some c, and f′(c) is at most 5, so f(2) − f(0) is at most 10. Starting from −3, f(2) is at most 7, reached only by the straight line of slope 5.

Figure (svg): A cone of possibilities starting at the point 0, minus 3: the straight line y equals minus 3 plus 5x is a ceiling, reaching 7 at x equals 2, and a sample curve whose slope never exceeds 5 stays below it, ending near 3.

A slope that never exceeds 5 cannot climb faster than the dashed line, so by x equals 2 the function is at most 7.

\[ f(2) = f(0) + 2f'(c) \le -3 + 2 \cdot 5 = 7 \]

Make your prediction first. The information is local: a starting value, and a cap on the slope at every point. The question is global: how far can the function get by x equal to 2?

The increment form answers it in one line. The change from 0 to 2 equals the derivative at some c times 2, and that derivative is at most 5, so the change is at most 10. Starting at minus 3, the function can reach at most 7.

The figure shows the geometry. The dashed line of slope 5 is the fastest possible climb, and any function whose slope never exceeds 5 stays on or below it. This use of the theorem, turning a bound on the derivative into a bound on the function, is how error estimates are proved throughout calculus.

33. The proof, by tilting

Section

Part 4

34. The chord is a line you can write down

Concept

The proof needs the chord as a formula. It passes through the left end point and has slope equal to the average rate, so point-slope form gives it at once.

\[ m = \frac{f(b) - f(a)}{b - a} \]

\[ y = f(a) + m(x - a) \]

Check both ends: at x equal to a the second term vanishes, leaving f(a); at x equal to b the second term is m times b minus a, which is exactly the rise.

\[ x = b: \quad f(a) + \frac{f(b) - f(a)}{b - a}(b - a) = f(b) \]

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 333 — proof of Theorem 4.5

The proof of the Mean Value Theorem subtracts the chord from the function, so you need the chord as an equation. It is a line, so all it takes is a slope and a point. The slope is the average rate m, and the left end point is on it.

Point-slope form then gives the chord at once: f of a plus m times x minus a. It is worth checking that it really passes through both ends. At a the second term is zero. At b, the factor b minus a in the slope cancels the b minus a from the run, leaving f of b minus f of a, and adding f of a gives f of b.

That cancellation is the whole reason the proof works. Keep it in mind for the fill-in slide later in this part.

35. Subtract the chord and the ends level out

Picture it

Figure (svg): Two panels. Left: a wavy curve f from a to b with its dashed chord, a vertical gap marked g(x) between them, and tangents parallel to the chord at two points. Right: the difference g equals f minus the chord, which starts and ends at zero and has level tangents at the same two points.

Subtracting the chord straightens it into the x-axis. Rolle's level tangents for g sit exactly where f's tangents are parallel to the chord.

Let g be the vertical gap between the curve and the chord, the pink segment on the left. Both ends of the curve sit on the chord, so the gap is zero at a and at b. On the right, g has equal end values: Rolle's picture, recovered by tilting.

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 333 — Figure 4.26

In the left panel, the pink segment is the vertical gap between the curve and the chord at one value of x. Call that gap g of x. At a and at b the curve meets the chord, so the gap is zero there.

The right panel plots g itself. It starts at zero and ends at zero, so it has equal end values, exactly Rolle's situation. Rolle's level tangents for g appear at c one and c two.

Now compare the panels. At c one and c two, the tangents on the left are parallel to the chord. That is not a coincidence. Subtracting a line of slope m lowers every slope by m, so where g has slope zero, f has slope m. The tilt changes the picture but not where the special points are.

36. Proving the Mean Value Theorem from Rolle's

Worked example

Assume f is continuous on the closed interval and differentiable on the open one, and let m be the chord's slope.

Define the gap function

Why: The curve minus the chord.

\[ g(x) = f(x) - \left[f(a) + m(x - a)\right] \]

Evaluate at a

Why: The bracket is f(a).

\[ g(a) = f(a) - f(a) = 0 \]

Evaluate at b

Why: The bracket is f(b), by the previous slide.

\[ g(b) = f(b) - f(b) = 0 \]

Check g inherits the hypotheses

Why: A line is continuous and differentiable everywhere, and so is a difference of such functions.

\[ g \text{ continuous on } [a,b], \quad \text{differentiable on } (a,b) \]

Apply Rolle to g

Why: Equal end values, so some interior c has a level tangent.

\[ g'(c) = 0 \text{ for some } c \in (a,b) \]

Differentiate g

Why: The derivative of the chord is its slope m.

\[ g'(x) = f'(x) - m \]

Conclude

Why: Set x equal to c.

\[ f'(c) - m = 0 \;\Longrightarrow\; f'(c) = \frac{f(b) - f(a)}{b - a} \]

Check on Exercise 161

Why: For the cube on the interval from 0 to 2, m is 4 and g is the cube minus 4x; Rolle's point for g is the same c found before.

\[ g(x) = x^3 - 4x, \quad g'(c) = 3c^2 - 4 = 0 \;\Longrightarrow\; c = \tfrac{2}{\sqrt3} \]

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 333 — proof of Theorem 4.5

This is the proof in the book, one move per line. Define g as f minus the chord. The next two lines show g is zero at both ends, using the chord's formula from two slides ago.

Then g inherits the smoothness of f, since subtracting a line cannot create a break or a corner. So g satisfies all three of Rolle's hypotheses, and some interior c has g prime of c equal to zero. Differentiating g gives f prime minus m, because the derivative of a line is its slope, and setting that to zero at c gives the theorem.

The check runs the proof on a real example, the cube on the interval from 0 to 2. There m is 4 and g is x cubed minus 4x. Rolle's point for g is two over root 3, the same c you found in Exercise 161 by the direct method. The two routes always agree, because they are the same equation.

37. Slide b and watch the tilted cube

Tweak it

Parameter explorer

The curve is g(x) = x³ − b²x: the cube minus its chord on [0, b]. Slide b. Where are g's end values, and where is its lowest point?

\[ g(x) = x^3 - {b}^2 x \]

  • b — from 1 to 3: right end b

\[ g'(x) = 3x^2 - b^2 = 0 \;\Longrightarrow\; c = \frac{b}{\sqrt3} \approx 0.577\,b \]

The curve on this slide is the gap function g for the cube on the interval from 0 to b: the cube minus its chord. Start with b at 2, which is the example just checked, then move the slider.

Two things stay true for every b. The curve starts at zero and comes back to zero exactly at x equal to b, because subtracting the chord always levels the ends. And the curve has a single lowest point between them, where its tangent is level. That lowest point is Rolle's point for g, which is the Mean Value point for the cube.

Watch where the minimum sits as b changes. It is always at b over root 3, about 0.577 of the way along. The shape scales, but the proportion does not, which is a nice way to see that the cube's Mean Value point is never the midpoint.

38. Why g vanishes at b

Fill the middle

Fill in the blanks

g(b) = f(b) - \left[f(a) + \fracf(b) - f(a)___(b - a)\right] = f(b) - \left[f(a) + ___\right] = 0

Why: The run b − a cancels, leaving the rise f(b) − f(a). Added to f(a) that gives f(b), and f(b) − f(b) = 0. This one cancellation is why the chord was the right thing to subtract.

Fill in the blank before checking. It is the single piece of algebra the proof depends on.

Inside the bracket, the slope's denominator b minus a meets the run b minus a, and they cancel, leaving the rise f of b minus f of a. Add f of a and you get f of b, so g of b is f of b minus f of b, which is zero.

If you had subtracted any other line, this would fail: a line of a different slope would miss the right-hand end point, and g would not have equal end values. The chord is the unique line through both end points, and that is why it is the thing to subtract.

39. Rolle versus the Mean Value Theorem

Comparison

Comparison matrix

Rolle's theoremMean Value Theorem
continuityon [a, b]on [a, b]
differentiabilityon (a, b)on (a, b)
end valuesf(a) = f(b)any
conclusionf′(c) = 0f′(c) = chord slope
proved fromEVT and FermatRolle

Exercise 150 asks when the two theorems are the same. Exactly when f(a) equals f(b): then the chord's slope is zero and the two conclusions coincide.

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 337 — Exercise 150

Fill in the blanks by comparing the two statements line by line. The first two rows are identical: both theorems need continuity on the closed interval and differentiability on the open one.

The differences are in the last three rows. Rolle demands equal end values; the Mean Value Theorem allows any end values. Rolle concludes a level tangent; the Mean Value Theorem concludes a tangent parallel to the chord. And the logical order is the reverse of the generality: Rolle is proved first, from the Extreme Value Theorem and Fermat, and then the general theorem is proved from Rolle.

Exercise 150 asks when the two are equivalent. Exactly when the end values are equal, because then the chord is level and its slope is zero.

40. Why both hypotheses matter

Section

Part 5

41. Continuous on the closed interval, differentiable on the open one

Concept

The two hypotheses are asked of different sets, and the difference is deliberate. Continuity must include the endpoints, so the ends of the chord are really on the curve. Differentiability is only needed inside, because c is always inside.

\[ \text{continuous on } [a, b] \qquad \text{differentiable on } (a, b) \]

So a vertical tangent at an endpoint is harmless. The square root on the interval from 0 to 4 has no derivative at 0, yet the theorem applies:

\[ \frac{\sqrt4 - \sqrt0}{4 - 0} = \frac12 = \frac{1}{2\sqrt c} \;\Longrightarrow\; c = 1 \]

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 337 — Exercises 148 and 149

The two hypotheses are asked of different sets, and it is worth understanding why. Continuity at the endpoints matters because the chord is drawn to the values at a and at b. If the function jumped at an endpoint, the chord would be aimed at a point the curve never really reaches.

Differentiability at the endpoints does not matter, because c is always strictly inside. The theorem never looks at the derivative at a or at b.

The square root shows this in action. Its tangent at 0 is vertical, so it has no derivative there, yet on the interval from 0 to 4 it satisfies both hypotheses, and the equation gives c equal to 1. Exercises 148 and 149 ask you to show both hypotheses are needed; the next slide does it.

42. One hypothesis dropped, one conclusion lost

Picture it

Figure (svg): Two panels on the interval from minus 1 to 1. Left: the floor function, a staircase with a jump at 0, with the dashed chord of slope 1 from the point minus 1, minus 1 to the point 1, 1, while every step is flat. Right: y equals absolute value of x minus one half, with its corner at one half and the dashed chord of slope minus one half; the two pieces have slopes minus 1 and plus 1.

Each failure kills the conclusion in its own way: the jump supplies all the rise while every tangent stays flat; the corner skips from slope minus 1 to plus 1 and never takes the chord's slope.

On the left the floor function breaks continuity: all of its rise happens in the jump at 0, while every tangent is flat. On the right the corner at one half breaks differentiability: the slope is minus 1 or plus 1, never the chord's minus one half.

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 338 — Exercises 167 and 170

Each panel keeps one hypothesis and breaks the other. On the left, the floor function is differentiable everywhere except at its jumps, and wherever it has a derivative, the derivative is zero. But it rises from minus 1 to 1 across the interval, all of it in a jump. The chord has slope 1, and no tangent ever does.

On the right, the absolute value of x minus one half is continuous everywhere, but it has a corner at one half. The chord has slope minus one half. The two straight pieces have slopes minus 1 and plus 1, and the corner lets the slope skip everything in between.

Together they show that neither hypothesis can be dropped. Continuity stops the function from gaining height without slope; differentiability stops it from changing slope without passing through the values in between.

43. Exercise 167: a corner at one half

Worked example

Show no c satisfies the equation below, and explain why the theorem does not apply.

\[ f(x) = \left|x - \tfrac12\right|, \qquad f(1) - f(-1) = f'(c)(2) \]

Evaluate the ends

Why: Distance from one half.

\[ f(1) = \tfrac12, \qquad f(-1) = \tfrac32 \]

Solve for the required slope

Why: Divide the change by 2.

\[ \tfrac12 - \tfrac32 = -1 = 2f'(c) \;\Longrightarrow\; f'(c) = -\tfrac12 \]

List the slopes the function actually has

Why: Each piece is a line.

\[ f'(x) = \begin{cases} -1 & x < \tfrac12 \\ +1 & x > \tfrac12 \end{cases} \]

Compare

Why: Neither minus 1 nor plus 1 equals minus one half, and at one half there is no derivative.

\[ f'(c) \in \{-1, 1\} \;\Longrightarrow\; f'(c) \ne -\tfrac12 \]

Check which hypothesis failed

Why: f is continuous on the whole interval, but the corner at one half is inside it, so differentiability on the open interval fails.

\[ \tfrac12 \in (-1, 1) \text{ and } f'\left(\tfrac12\right) \text{ does not exist} \]

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 338 — Exercise 167

The exercise asks you to show directly that no c works, and then to diagnose why. Start by finding what the slope would have to be. The function falls from three halves to one half, a change of minus 1 over a run of 2, so the required slope is minus one half.

Then list the slopes the function actually has. Each piece is a straight line, one with slope minus 1 and one with slope plus 1, and at the corner itself there is no slope. Minus one half is not on the list, so no c exists.

The final step identifies the broken hypothesis. The function is continuous, so the failure is differentiability, and it fails at one half, which is inside the interval. If the corner had been outside, or exactly at an endpoint, the theorem would have applied.

44. Does the Mean Value Theorem apply?

Sorting

Sort into buckets

Exercises 171 to 181. Continuous on the closed interval and differentiable on the open one?

Applies
eˣ on [0, 1]; √(9 − x²) on [−3, 3]; x sin(πx) on [0, 2]; 1/|x + 1| on [0, 3]
Does not apply
ln(2x + 3) on [−3/2, 0]; tan(2πx) on [0, 2]; (x² + 3x + 2)/x on [−1, 1]; 5 + |x| on [−1, 1]
yes
eˣ and x sin(πx) are smooth everywhere. √(9 − x²) is continuous on [−3, 3] and differentiable inside; its vertical tangents are at the endpoints only. 1/|x + 1| has its trouble at x = −1, outside [0, 3].
no
ln(2x + 3) is undefined at x = −3/2, so it is not continuous on the closed interval. tan(2πx) has an asymptote at x = 1/4. (x² + 3x + 2)/x is undefined at 0. 5 + |x| has a corner at 0, inside the interval.

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 338 — Exercises 171 to 181

Sort each function by checking the two hypotheses, and look for trouble in two places: points where the function is undefined or jumps, and points where it has a corner or a vertical tangent. Then ask whether the trouble is inside the interval, at an endpoint, or outside it.

The four that apply include two worth a second look. The square root of nine minus x squared is a semicircle with vertical tangents at plus and minus 3, but those are the endpoints, where only continuity is needed. And one over the absolute value of x plus 1 blows up at minus 1, which is outside the interval from 0 to 3.

The four that fail each fail inside the interval, or at an endpoint where continuity is required. The logarithm is undefined at minus three halves, the left endpoint. The tangent function has an asymptote at one quarter. The rational function is undefined at 0. And five plus the absolute value of x has a corner at 0.

45. Exercise 168: before solving anything

Step zero

\[ f(x) = \frac{1}{x^2} \quad \text{on } [-1, 1], \qquad f(1) - f(-1) = f'(c)(2) \]

Discussion prompt

Before you look for c, which hypothesis should you test first, and what does it tell you? Then say whether a c exists.

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 338 — Exercise 168

Write your answer before revealing. The instinct is to jump straight into the equation, and this slide is here to slow that instinct down.

The first thing to check is continuity, and it fails immediately: one over x squared is undefined at 0, and 0 is inside the interval. That settles whether the theorem applies. It does not.

It is still instructive to finish the calculation. The end values are both 1, so you would need a level tangent. But the derivative, minus two over x cubed, is never zero, so no such point exists. The graph is two separate pieces, each shooting up towards the gap at 0, and neither piece ever levels off.

46. Exercise 151: a jump, and the equation holds anyway

Counterexample

Discussion prompt

Can a function with a discontinuity still have a c with f′(c)(b − a) = f(b) − f(a)? Build an example or explain why not.

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 337 — Exercise 151

Think about this before revealing. The question is whether a discontinuous function is forbidden from satisfying the conclusion, or only unprotected by the theorem.

The example on the slide answers it. The function follows x squared up to 1, then drops by 1 and continues as x squared minus 1. It is discontinuous at 1. Its chord from 0 to 2 has slope three halves, and on the first piece the derivative two x equals three halves at three quarters. So the equation holds, jump and all.

This is a lesson about logic that applies to every theorem you will meet. The hypotheses are sufficient conditions. When they hold, the conclusion is guaranteed. When they fail, the theorem simply says nothing: the conclusion may fail, as it did for the floor function, or may hold by luck, as it does here.

47. Trap: demanding a derivative at the endpoints

Trap

The trap

A rejected application:

\[ f(x) = \sqrt x \text{ on } [0, 4] \]

\[ f'(0) \text{ does not exist} \]

\[ \Longrightarrow \text{ theorem fails?} \]

Wrong. Differentiability is not asked for at 0.

The fix

The hypothesis is differentiability on the open interval. The endpoints only need continuity, and the root is continuous at 0. The theorem applies, and c is 1.

\[ f'(1) = \tfrac12 = \frac{2 - 0}{4 - 0} \]

The rejected argument on the left is being too careful in the wrong place. It notices, correctly, that the square root has no derivative at 0. It then concludes, wrongly, that the theorem cannot be used on the interval from 0 to 4.

Read the hypothesis again: differentiable on the open interval. The endpoints are excluded on purpose, because the point c is always strictly inside. At the endpoints the theorem only needs continuity, and the square root is continuous at 0.

The theorem applies and gives c equal to 1, where the tangent slope is one half, the same as the chord from the origin to the point 4, 2. Whenever you check differentiability, check it on the open interval and nowhere else.

48. The corollaries

Section

Part 6

49. The increment form: local slope, global change

Concept

Multiply the theorem through by the length of the interval. It now says the change in f over the whole interval is some derivative value times the run.

\[ f(b) - f(a) = f'(c)(b - a) \]

Anything you know about the derivative everywhere, such as its sign, a bound, or that it is zero, now becomes a statement about the change in f between any two points. That is the engine behind all three corollaries.

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 335 — introduction to the corollaries

This slide is the hinge of the lesson. Multiply both sides of the theorem by b minus a and you get the increment form: the change in f between two points equals the derivative somewhere between them, times the distance between them.

Read it as a machine. You feed in a fact about the derivative that holds everywhere in the interval, because then it holds at the unknown c. You get out a fact about the change in f between any two points. If the derivative is zero, the change is zero. If the derivative is positive and the run is positive, the change is positive. If the derivative is at most 5, the change is at most 5 times the run.

Each of those is one of the corollaries, or the bound you proved on the estimation slide. The unknown location of c never matters, because the fact you feed in is true everywhere.

50. Corollary 1: a zero derivative means a constant

Concept

Corollary 1 (Theorem 4.6) — If f is differentiable over an interval I and its derivative is zero at every point of I, then f is constant on I.

Take any two points a less than b in the interval and apply the increment form.

\[ f(b) - f(a) = f'(c)(b - a) \]

\[ f(b) - f(a) = 0 \cdot (b - a) = 0 \]

\[ f(b) = f(a) \text{ for every pair } a, b \in I \]

The book argues by contradiction, supposing two different values; this direct version is the same idea.

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, pp. 335-336 — Theorem 4.6 and its proof

You have known since Chapter 3 that a constant function has zero derivative. This corollary is the converse, and it needs the Mean Value Theorem, because a zero derivative is a statement about each point separately while being constant is a statement about all pairs of points.

The proof is the increment form with the derivative set to zero. Pick any two points in the interval. The change between them is the derivative at some c times the run, and that derivative is zero, so the change is zero. Every pair of points has the same value, so the function is constant.

The book writes this as a proof by contradiction, supposing two points with different values and deriving a nonzero derivative. It is the same argument read backwards. Notice the word interval in the statement. It is essential, as the trap later in this part shows.

51. Corollary 2: equal derivatives differ by a constant

Concept

Constant Difference Theorem (Theorem 4.7) — If f and g are differentiable over an interval I and have the same derivative at every point of I, then f equals g plus a constant.

\[ h = f - g \;\Longrightarrow\; h' = f' - g' = 0 \]

\[ h = C \;\Longrightarrow\; f = g + C \]

Figure (svg): Four parabolas y equals x squared plus C for C equal to minus 1, 0, 1 and 2, stacked vertically, each with a short tangent at x equals 1; all four tangents have slope 2 and are parallel.

Same derivative everywhere means the same slope at every x, so the curves are vertical translates of each other. Corollary 2 says there are no other possibilities on an interval.

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 336 — Theorem 4.7 and its proof

This follows from Corollary 1 in two lines. Subtract the two functions. The difference has derivative f prime minus g prime, which is zero, so by Corollary 1 the difference is a constant.

The figure shows what that means geometrically. The four parabolas all have derivative two x, so at every x they have the same slope, as the parallel tangents at x equal to 1 show. They are vertical translates of one another.

The content of the corollary is that there is no other way for two functions to share a derivative on an interval. You cannot bend one of them somewhere without changing its slope there. That fact is about to become very important.

52. Why this is the constant of integration

Intuition

Later in this chapter you will undo derivatives: given a derivative, find the function. If one answer is F, Corollary 2 says every other answer is F plus a constant, and nothing else. That is why every antiderivative carries a plus C, and why one known value pins down the constant.

\[ F' = f \;\Longrightarrow\; \text{every antiderivative of } f \text{ is } F + C \]

Without the Mean Value Theorem you could not rule out some strange function with the right derivative that is not a shifted copy of F.

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 336 — the remark after Theorem 4.7

Later in this chapter you will work backwards from derivatives: given f, find a function whose derivative is f. Such a function is called an antiderivative. If you find one, F, you can add any constant and still have an antiderivative, because constants differentiate to zero.

Corollary 2 is the converse: those are the only ones. Any function with derivative f on an interval is F plus some constant. So when you write plus C after an antiderivative, you are describing all the answers, not just some of them.

This is also why one extra piece of information, such as a starting value, is enough to pin down the answer completely. It fixes the constant. You will use exactly this reasoning with initial-value problems in Section 4.10.

53. Two functions, one derivative: sine squared and minus cosine squared

Worked example

Show these two functions have the same derivative, and find the constant they differ by.

\[ f(x) = \sin^2 x, \qquad g(x) = -\cos^2 x \]

Differentiate f

Why: Chain rule: the outer square, then sine.

\[ f'(x) = 2\sin x \cos x \]

Differentiate g

Why: Chain rule: the derivative of cosine is minus sine, and the two minus signs cancel.

\[ g'(x) = -2\cos x \cdot (-\sin x) = 2\sin x \cos x \]

Apply Corollary 2 on the whole real line

Why: The derivatives agree everywhere, and the real line is an interval.

\[ \sin^2 x = -\cos^2 x + C \]

Find C from one value

Why: Put x equal to zero.

\[ 0 = -1 + C \;\Longrightarrow\; C = 1 \]

Check at another point

Why: At x equal to 0.7 the difference is still 1, which is the Pythagorean identity in disguise.

\[ \sin^2(0.7) - \left(-\cos^2(0.7)\right) = 0.41502 + 0.58498 = 1 \]

These two functions look quite different, yet they have the same derivative. Each derivative is a chain-rule calculation, and in the second one two minus signs cancel. Both come out to two sine x cosine x.

Corollary 2 applies because the derivatives agree on the whole real line, which is an interval. So sine squared minus the other function is a constant, and evaluating at 0 finds it: sine squared of 0 is 0, and minus cosine squared of 0 is minus 1, so the constant is 1.

Rearranged, that is sine squared plus cosine squared equals 1, the Pythagorean identity. You have just proved a trigonometric identity using calculus, without any triangles. The numerical check at 0.7 confirms it to five decimal places.

54. Trap: a zero derivative on a broken domain

Trap

The trap

Both terms are differentiable for every x except 0:

\[ h(x) = \arctan x + \arctan\tfrac1x \]

\[ h'(x) = \frac{1}{1+x^2} - \frac{1}{1+x^2} = 0 \]

\[ \Longrightarrow h \text{ is one constant?} \]

Wrong. Corollary 1 needs an interval.

The fix

The domain is two intervals with a gap at 0, and h is constant on each separately, with different constants.

\[ h(1) = \tfrac{\pi}{2}, \quad h(-1) = -\tfrac{\pi}{2} \]

Figure (svg): The function arctan x plus arctan of one over x: a horizontal line at height pi over 2 for positive x and at minus pi over 2 for negative x, with a gap at x equals 0.

Zero derivative everywhere it is defined, but two different constants: the domain is two intervals, not one.

The computation on the left is correct. The derivative of arctan of one over x, by the chain rule, is minus one over one plus x squared, which exactly cancels the derivative of arctan x. So the derivative of h is zero wherever it is defined.

The error is concluding that h is one constant. Corollary 1 needs an interval, and the domain of h is not one: it is all nonzero x, two separate pieces with a gap at 0. The Mean Value Theorem cannot be applied across the gap, because h is not continuous there, not even defined.

The figure shows the result. On the positive side h is pi over 2, and on the negative side it is minus pi over 2. The correct statement is that h is constant on each interval of its domain, possibly with different constants.

55. Corollary 3: the sign of the derivative decides the direction

Concept

Corollary 3 (Theorem 4.8) — Let f be continuous over the closed interval from a to b and differentiable over the open interval. If the derivative is positive throughout the open interval, f is increasing on the closed interval; if it is negative throughout, f is decreasing.

Figure (svg): Top: the curve y equals x cubed minus 3x from minus 2.2 to 2.2, rising to a peak at x equals minus 1, falling to a valley at x equals 1, then rising. Bottom: a number line for the sign of the derivative, plus to the left of minus 1, minus between minus 1 and 1, plus to the right of 1.

Read the sign of the derivative off the number line and the direction of the curve follows: Corollary 3 is the licence for that reading.

\[ f(x) = x^3 - 3x, \quad f'(x) = 3(x - 1)(x + 1) \]

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, pp. 336-337 — Theorem 4.8 and Figure 4.29

This is the corollary you will use most, starting in the very next section. If the derivative is positive at every point inside an interval, the function increases across the whole interval; if negative, it decreases.

The figure shows how you will apply it. Factor the derivative of x cubed minus 3x as 3 times x minus 1 times x plus 1. The factors change sign at minus 1 and at 1, so the number line underneath splits into three pieces, and on each the derivative has one sign. The curve above rises, falls, and rises accordingly.

Notice the hypotheses: continuity on the closed interval, differentiability on the open one, the same as the Mean Value Theorem. That is because the proof is one application of it.

56. Reading Corollary 3 piece by piece

Notation

Annotate

On: \( f' > 0 \text{ on } (a,b) \;\Longrightarrow\; x_1 < x_2 \Rightarrow f(x_1) < f(x_2) \)

  • A statement about every single point inside the interval, one slope at a time.
  • ANY two points of the interval, however far apart. This is the global statement.
  • The definition of increasing: further right means higher.
  • The bridge from local to global is the Mean Value Theorem, applied on the interval from x₁ to x₂.

The hypothesis and the conclusion of this corollary are about different kinds of thing. The hypothesis is a statement at each point: the slope there is positive. The conclusion is a statement about pairs of points, however far apart: the one on the right is higher.

Step through the notes and notice where the difficulty lies. A positive slope at a point only says the function is rising near that point. Why should all those local rises add up to a global rise? Without a theorem to connect them, you could not be sure.

The connection is the Mean Value Theorem on the interval from the first point to the second. It produces one point c where the slope equals the average rate between them. That single slope is positive, so the average rate is positive, so the function rose.

57. Proving Corollary 3(i) directly

Worked example

Assume the derivative is positive on the open interval, and take any two points in the closed interval.

\[ a \le x_1 < x_2 \le b \]

Apply the Mean Value Theorem on the smaller interval

Why: f is continuous on it and differentiable inside it.

\[ f(x_2) - f(x_1) = f'(c)(x_2 - x_1), \quad c \in (x_1, x_2) \]

Sign of each factor

Why: c is inside the big interval, so its derivative is positive; the run is positive because the points are in order.

\[ f'(c) > 0, \qquad x_2 - x_1 > 0 \]

Multiply

Why: A positive times a positive.

\[ f(x_2) - f(x_1) > 0 \]

Conclude

Why: Rearrange.

\[ f(x_1) < f(x_2) \]

Check on the sign-chart function

Why: x cubed minus 3x has positive derivative beyond 1; try 1.2 and 1.8.

\[ f(1.2) = -1.872 < 0.432 = f(1.8) \]

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 337 — proof of Theorem 4.8

Pick any two points in the interval, with the first to the left of the second. Apply the Mean Value Theorem on the smaller interval between them; the hypotheses carry over, since a function continuous and differentiable on a big interval is so on any piece of it.

The increment form expresses the change as a product of two numbers. The derivative at c is positive, because c is inside the big interval where every derivative is positive. The run is positive because the points are in order. So the change is positive, and the right-hand value is bigger.

The book proves this by contradiction, supposing a pair of points where the function did not rise and deriving a negative derivative. This direct version is the same idea without the detour. The check tests the result on x cubed minus 3x, which has positive derivative beyond 1: the value at 1.8 is indeed bigger than the value at 1.2.

58. Exercise 182: exactly one real root

Worked example

\[ f(x) = x^3 + 4x + 16 \]

At least one root: the Intermediate Value Theorem

Why: The function is continuous and changes sign between minus 3 and 0.

\[ f(-3) = -27 - 12 + 16 = -23 < 0, \qquad f(0) = 16 > 0 \]

At most one root: the derivative

Why: It is never less than 4.

\[ f'(x) = 3x^2 + 4 \ge 4 > 0 \]

Apply Corollary 3

Why: A strictly increasing function cannot hit the same height twice.

\[ f \text{ increasing on } (-\infty, \infty) \]

Find the root

Why: Try small integers between minus 3 and 0.

\[ f(-2) = -8 - 8 + 16 = 0 \]

Figure (svg): The curve y equals x cubed plus 4x plus 16 from minus 3 to 1.5, always rising, crossing the x-axis exactly once, at x equals minus 2.

The IVT gives at least one crossing between minus 3 and 0; a derivative that never drops below 4 forbids a second one.

Check by factoring

Why: Divide out x plus 2; the quadratic left over has negative discriminant, so no other real roots.

\[ x^3 + 4x + 16 = (x + 2)(x^2 - 2x + 8), \quad (-2)^2 - 4 \cdot 8 = -28 < 0 \]

OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem §4.4, p. 338 — Exercise 182

Exactly one is two claims, and each needs its own tool. At least one comes from the Intermediate Value Theorem: the function is negative at minus 3 and positive at 0, so it crosses zero in between.

At most one comes from the derivative. Three x squared plus 4 is never less than 4, so by Corollary 3 the function is increasing on the whole line, and an increasing function cannot hit the same height twice. You could also argue with Rolle's theorem: two roots would mean equal values, so the derivative would be zero somewhere between them, and it never is.

Finding the root is then a matter of trying integers in the bracket, and minus 2 works. The check factors the cubic and confirms that the quadratic left over has a negative discriminant, so it contributes no other real roots.

59. Match each derivative fact to what it forces

Matching

Match the pairs

  • a. f′(x) = 0 for all x in an interval
  • b. f′(x) = g′(x) for all x in an interval
  • c. f′(x) > 0 for all x in (a, b)
  • d. f′(x) ≤ 5 for all x, and f(0) = −3
  • w. f is constant there
  • x. f = g + C for one constant C
  • y. f is increasing on [a, b]
  • z. f(2) ≤ 7

Why: Every one of these comes from the same equation, f(b) − f(a) = f′(c)(b − a), with a different fact about f′(c) plugged in: zero, a difference of zero, positive, or at most 5.

Match each fact on the left before checking. Each one is the increment form with a different piece of information about the derivative plugged in.

The first three are the three corollaries in order: zero derivative gives a constant; equal derivatives give a constant difference; a positive derivative gives an increasing function. The fourth is the bound from the estimation slide, where a cap of 5 on the slope and a starting value of minus 3 limit the value at 2 to at most 7.

If any match felt uncertain, go back to the increment form on the corollaries' first slide and plug the fact in yourself. It should take one line each.

60. Putting it together

Section

Part 7

61. Pattern: using the Mean Value Theorem

Pattern

Figure (svg): A flow diagram. Top row, left to right: is f continuous on the closed interval; is f differentiable on the open interval; compute the chord slope m. Arrows labelled no from the first two boxes drop to a red note: the theorem says nothing. From the third box an arrow goes down to: solve f prime of c equals m; then keep only solutions strictly inside; then check by substituting back.

Hypotheses first, then one slope, then one equation. The last two boxes are where marks are usually lost.
  1. Check continuity on the closed interval. A break anywhere, endpoints included, and the theorem is silent.
  2. Check differentiability on the open interval. Corners, cusps and vertical tangents inside are fatal; at the ends they are harmless.
  3. Compute the chord slope m, the average rate of change.
  4. Solve f′(c) = m, and keep only solutions strictly inside the interval.
  5. Check by substituting c back. For the corollaries, use the increment form with whatever you know about f′.

This is the order to work in whenever a problem invokes the theorem. The flow diagram puts the hypotheses first, because if either fails, nothing that follows is guaranteed, and the theorem gives you no right to expect a c.

For the hypotheses, remember where to look: continuity on the closed interval, endpoints included, and differentiability on the open interval only. Corners, cusps and vertical tangents are fatal inside, harmless at the ends.

Then compute one slope and solve one equation. The last two boxes are where marks are lost: discarding solutions outside the interval, and checking by substituting back. For corollary-style problems, skip the solving and plug what you know about the derivative straight into the increment form.

62. Check: finding the guaranteed point

Check

Check your understanding

For f(x) = x² on [1, 5], which c does the Mean Value Theorem give?

  • A. c = 3 (correct)
  • B. c = 6
  • C. c = √13 ≈ 3.61
  • D. No such c: the endpoint values differ

Answer: A

Why: The chord's slope is (25 − 1)/(5 − 1) = 6, and f′(c) = 2c = 6 gives c = 3, the midpoint, as always for a parabola.

Why B tempts people
6 is the chord's slope, the value f′(c) must take. The question asks for c itself: solve 2c = 6.
Why C tempts people
√13 is where f takes the average of its end values, 13. The theorem is about the derivative, not the function's value.
Why D tempts people
Unequal end values rule out Rolle's theorem, not the Mean Value Theorem, which has no such requirement.

The chord's slope is 24 over 4, which is 6, and the derivative two c equal to 6 gives c equal to 3. As a parabola, the answer had to be the midpoint of the interval from 1 to 5, so the answer checks itself.

The distractors are the three most common confusions. Reporting 6 answers the wrong question: 6 is the value the derivative must take, not the point where it takes it. Root 13 answers a different question entirely, where the function reaches its average value. And refusing because the end values differ confuses this theorem with Rolle's.

63. Check: the hypotheses

Check

Check your understanding

On the interval [−1, 1], to which function does the Mean Value Theorem apply?

  • A. f(x) = |x|
  • B. f(x) = x^(2/3)
  • C. f(x) = √(x + 1) (correct)
  • D. f(x) = 1/x

Answer: C

Why: √(x + 1) is continuous on [−1, 1] and differentiable wherever x + 1 is positive, which is all of (−1, 1). Its vertical tangent is at the endpoint −1, where differentiability is not required.

Why A tempts people
|x| has a corner at 0, inside the interval, so it is not differentiable on (−1, 1).
Why B tempts people
x^(2/3) has a cusp at 0: its derivative (2/3)x^(−1/3) blows up there, inside the interval.
Why D tempts people
1/x is not even defined at 0, so continuity on [−1, 1] fails.

For each option, look for trouble inside the open interval and for a break anywhere on the closed one. The square root of x plus 1 has its only trouble at minus 1, a vertical tangent at the left endpoint, where differentiability is not required, so the theorem applies.

The others all fail inside. The absolute value has a corner at 0. The two-thirds power has a cusp at 0, where its derivative blows up. And one over x is not defined at 0, so it fails continuity before differentiability even comes up.

64. Check: the corollaries

Check

Check your understanding

F and G are differentiable on the whole real line with F′(x) = G′(x) = 2x, F(0) = 3 and G(0) = −1. What is F(x) − G(x)?

  • A. 4, for every x (correct)
  • B. 0, since the derivatives are equal
  • C. 2x
  • D. It depends on x and cannot be found

Answer: A

Why: By Corollary 2 on the whole real line, F − G is a constant. At x = 0 it is 3 − (−1) = 4, so F(x) − G(x) = 4 everywhere.

Why B tempts people
Equal derivatives force a constant difference, not a zero difference. The values at 0 differ by 4.
Why C tempts people
2x is the common derivative. The derivative of F − G is 2x − 2x = 0.
Why D tempts people
The real line is an interval, so Corollary 2 applies and one value fixes the constant.

This is Corollary 2 on the whole real line, which is an interval, so F minus G is a constant. Evaluating at 0 gives 3 minus minus 1, which is 4, and the difference is 4 for every x.

The tempting wrong answer is zero, from the feeling that functions with equal derivatives must be equal. They need only differ by a constant, and here the given values show the constant is 4. The answer 2x confuses the functions with their derivative.

65. Explain the tilt

Explain it to yourself

Discussion prompt

In two or three sentences: why does subtracting the chord turn the Mean Value Theorem into Rolle's theorem, and why does the derivative of the new function vanish exactly where the old tangent is parallel to the chord?

Write your explanation before revealing. Being able to reconstruct this proof in your own words is the best evidence that you understand the theorem rather than just its formula.

A strong answer makes two points. First, the chord passes through both end points of the curve, so the difference between them is zero at both ends, which gives Rolle's equal end values. Second, subtracting a line of slope m lowers every slope by exactly m, so a zero slope for the difference is a slope of m for the original function.

If your explanation mentioned only the first point, you have explained why Rolle applies but not why its conclusion is the one you wanted. The second point is what translates Rolle's level tangent back into the Mean Value Theorem's parallel tangent.

66. Exit ticket

Exit ticket

\[ f(x) = x^3 - x \quad \text{on } [0, 2] \]

Discussion prompt

Find every c the Mean Value Theorem guarantees, then use Corollary 3 to decide whether f is increasing on [1, 2].

This ticket uses both halves of the lesson on one function. First the standard routine: the chord from 0 to 2 has slope 3, and three c squared minus 1 equal to 3 gives c equal to two over root 3, about 1.155. The negative root is outside the interval and is discarded.

Then Corollary 3. On the interval from 1 to 2 the derivative three x squared minus 1 is at least 2, which is positive, so the function increases there. You did not need to compare any values of the function to know that.

If either part felt shaky, revisit Exercise 161 for the first and the corollary proof for the second.

67. Recap

Recap

resulthypothesesconclusion
Rolle's theoremcontinuous on [a, b], differentiable on (a, b), f(a) = f(b)f′(c) = 0 for some c in (a, b)
Mean Value Theoremcontinuous on [a, b], differentiable on (a, b)f′(c) = (f(b) − f(a))/(b − a) for some c in (a, b)
Corollary 1f′ = 0 on an intervalf is constant
Corollary 2f′ = g′ on an intervalf = g + C
Corollary 3f′ > 0 (or < 0) on (a, b)f increasing (or decreasing) on [a, b]

\[ f(b) - f(a) = f'(c)(b - a) \]

Next, Section 4.5 uses Corollary 3 to read maxima, minima and the shape of a graph from the sign of the derivative.

Stewart, Calculus: Early Transcendentals 8e, §4.2 The Mean Value Theorem §4.2, pp. 287-292 — the same material in Stewart

Two theorems and three corollaries, all resting on one picture: a tangent parallel to the chord. Rolle's theorem is the level-chord case, proved from the Extreme Value Theorem and Fermat's theorem. The Mean Value Theorem follows from Rolle by subtracting the chord.

The hypotheses are the same for both, continuity on the closed interval and differentiability on the open one, and each is genuinely needed, as the jump and the corner showed. When they fail, the theorem is silent rather than false.

The increment form in the middle of the slide is the version to remember. Feed it any fact about the derivative and it returns a fact about the function. Section 4.5 feeds it the sign of the derivative, and uses Corollary 3 to find where functions rise and fall and where their maxima and minima are.

Sources

  1. OpenStax Calculus Volume 1, §4.4 The Mean Value Theorem — Strang & Herman, OpenStax / Rice University, CC BY-NC-SA 4.0, pp. 329-338
  2. Stewart, Calculus: Early Transcendentals 8e, §4.2 The Mean Value Theorem — James Stewart, Cengage Learning, 2016, pp. 287-292

Want this taught 1-on-1? Alexander tutors Calculus I — $55/session, free consultation.

Book on Wyzant · Text (657) 465-8108