Calculus › Section D · Applications of derivatives › Topic D.2

Rolle's theorem and the mean value theorem

If you average 80 km/h on a road trip, at some instant your speedometer read exactly 80. That is not folk wisdom but a theorem, and it quietly holds up most of what comes after it in calculus.

Two toll booths

Suppose a motorway logs your number plate at kilometre zero and again 240 km later, three hours afterwards. Your average speed was 80 km/h. Had you stayed strictly above 80 the whole time, you would have covered more than 240 km; strictly below, less. So either you sat at exactly 80 or you spent time on both sides of it, and a speedometer needle that moves without jumping cannot get from one side to the other without passing 80. That argument is the Mean Value Theorem in everyday form. This topic states it precisely, starts with the level special case named after Michel Rolle, and then collects its consequences.

Rolle's theorem

Let \(f\) be continuous on the closed interval \([a, b]\) and differentiable on the open interval \((a, b)\), with equal endpoints \(f(a) = f(b)\). Then there is at least one point \(c\) strictly inside where

f'(c) = 0

A smooth, unbroken curve that starts and ends at the same height must flatten out somewhere in between. Note the hypotheses exactly: continuity is demanded everywhere including the endpoints, but differentiability only between them, so the derivative may fail to exist at \(a\) or \(b\). And the conclusion is existence, not uniqueness and not location: one level point is guaranteed, and there may be several.

Why Rolle's theorem is true

The proof assembles two results you already own. If \(f\) is constant, every point works, done. Otherwise, by the Extreme Value Theorem from D.1, \(f\) attains an absolute maximum and minimum on \([a, b]\). Since the endpoints share one value and the function takes another value somewhere, at least one of those extremes occurs strictly inside. At an interior extreme where the function is differentiable, Fermat's theorem, also from D.1, forces the derivative to be zero there. That is the entire argument: Rolle's theorem is D.1's two theorems applied to a new question.

Worked example: spotting the promised point

Verify Rolle's theorem for \(f(x) = x^2 - 4x + 3\) on \([1, 3]\). Hypotheses: \(f\) is a polynomial, so continuous and differentiable everywhere; the endpoints are \(f(1) = 1 - 4 + 3 = 0\) and \(f(3) = 9 - 12 + 3 = 0\), equal. So the theorem promises a \(c\) in \((1, 3)\) with \(f'(c) = 0\). Computing: \(f'(x) = 2x - 4 = 0\) at \(c = 2\), which lies inside the interval. Geometrically, \(c = 2\) is the vertex of the parabola, the one point where it levels out. The theorem's value is that it promises such a point even for functions where you cannot solve \(f'(c) = 0\) by hand.

The Mean Value Theorem

Now tilt the picture. Let \(f\) be continuous on \([a, b]\) and differentiable on \((a, b)\), with no requirement on the endpoint values. Then there is at least one \(c\) in \((a, b)\) where the tangent slope equals the average slope across the whole interval:

f'(c) = \frac{f(b) - f(a)}{b - a}

The right side is the slope of the secant line through the endpoints. Rolle's theorem is the special case where that secant is level, with slope 0; the Mean Value Theorem (MVT) is the same picture tilted. Somewhere between \(a\) and \(b\), the curve runs exactly parallel to its own average. Applied to a position function, that says instantaneous velocity must equal average velocity at some instant.

Worked example: finding c

For \(f(x) = \sqrt{x}\) on \([0, 4]\), the hypotheses hold: \(\sqrt{x}\) is continuous on \([0, 4]\) and differentiable on \((0, 4)\). The average slope is \(\frac{\sqrt{4} - \sqrt{0}}{4 - 0} = \frac{2}{4} = \frac{1}{2}\). The promised point satisfies \(f'(c) = \frac{1}{2\sqrt{c}} = \frac{1}{2}\), so \(\sqrt{c} = 1\) and \(c = 1\). At \(x = 1\) the square-root curve has slope exactly the interval's average, and the tangent line there is parallel to the secant from \((0, 0)\) to \((4, 2)\). Notice that \(\sqrt{x}\) has no derivative at the endpoint 0, and the theorem still applies, because it asks for differentiability only inside the interval.

The camera's argument, made precise

Average-speed cameras rest on this theorem. Model the car's position \(s(t)\), in kilometres after \(t\) hours, as a differentiable function of time, which is reasonable because a car cannot jump from place to place. If \(s\) increases by 240 between two cameras three hours apart, the Mean Value Theorem gives an instant \(c\) where \(s'(c) = \frac{240}{3} = 80\) km/h. Varying your speed cannot dodge it: slowing down for a visible camera or stopping at a service station only lowers the average, and if the average still exceeds the limit, then at some instant your speed did too. Two timestamps give a fact about one instant, with no record of the journey in between.

Two consequences that run the course

The Mean Value Theorem yields two workhorse corollaries. First: if \(f'(x) = 0\) for every \(x\) in an interval, then \(f\) is constant there. Proof in one line: pick any two points \(a < b\) of the interval; the MVT gives an in-between \(c\) with \(\frac{f(b) - f(a)}{b - a} = f'(c) = 0\), so \(f(a) = f(b)\), and the points were arbitrary. Second: if \(f' = g'\) throughout an interval, then \(f - g\) is constant: apply the first corollary to \(f - g\), whose derivative is zero. This is exactly why the antiderivatives of Section E all come with a constant \(C\): functions with the same derivative differ only by a constant.

Why the MVT is really here

The Mean Value Theorem is used far more often to prove other results than to find a particular \(c\). The increasing and decreasing test of the next topic, "positive derivative means rising", is a direct application: if \(f' > 0\) throughout an interval, then for any two points \(a < b\) in it the secant slope \(\frac{f(b) - f(a)}{b - a} = f'(c)\) is positive, so \(f(b) > f(a)\). Later, the proof of the fundamental theorem of calculus uses the constant-difference corollary. When a theorem in this course feels abstract, ask what it makes provable. For the MVT, the answer is the link between the sign of a derivative and the shape of a graph.

Common mistakes to avoid

The usual slips: demanding differentiability at the endpoints, when only continuity is required there, and so rejecting functions like \(\sqrt{x}\) on \([0, 4]\) that qualify perfectly; reading "there exists a \(c\)" as "every \(c\) works" or "there is exactly one"; applying the theorem when a hypothesis fails: \(|x|\) on \([-1, 1]\) has equal endpoint values but no level tangent anywhere, because it is not differentiable at the corner \(x = 0\); forgetting to divide \(f(b) - f(a)\) by \(b - a\) when computing the average slope; and assuming \(c\) must be a maximum or minimum. The MVT fixes a tangent slope, not a peak.

Key takeaways

Rolle's theorem: continuous on \([a, b]\), differentiable inside, equal endpoints, and a level tangent is guaranteed in between. The Mean Value Theorem drops the equal-endpoint condition: some interior tangent slope equals the secant slope \(\frac{f(b) - f(a)}{b - a}\). Both are existence results, promising at least one \(c\) strictly inside, and both need exactly the hypotheses stated. Their corollaries (zero derivative means constant, equal derivatives mean a constant difference) underlie the rest of the course, from the sign tests of the next topics to the \(C\) in every antiderivative.

Flashcards

Flashcard

Rolle's theorem

f(a) = f(b), with the continuity and differentiability conditions ⇒ f′(c) = 0 for some c in (a, b)

The level special case of the MVT.

Test your understanding

Question 1 of 5

Rolle's theorem applies to f(x) = x² − 4x + 3 on [1, 3]. At which c is f′(c) = 0?

Question 2 of 5

For f(x) = x² on [0, 4], the Mean Value Theorem promises a point c where f′(c) equals the average slope. What is c?

Question 3 of 5

Why does Rolle's theorem not apply to f(x) = |x| on [−1, 1]?

Question 4 of 5

Two functions have equal derivatives at every point of an interval. What follows from the Mean Value Theorem?

Question 5 of 5

A car's position log shows 240 km gained in exactly 3 hours. What does the Mean Value Theorem say about its speedometer?

Standards: 5

← Previous topicNext topic →