Let \(y=f(x)\) be a continuous function. Suppose \(f\) takes the same value at \(x=a\) and \(x=b\); that is, \(f(a)=f(b)\). Also suppose that \(f'(x)\) exists everywhere between \(a\) and \(b\). The graph of \(f\) is then a continuous curve such that we can draw a tangent line at each point of this curve (see Fig. 1). Geometric intuition shows us at once that for at least one value of \(x\) between \(a\) and \(b\), the tangent is parallel to the \(x\)-axis (say at \(P\)); that is, the slope is zero. This illustrates **Rolle’s theorem**:

**Theorem 1. ** Rolle’s Theorem. Let \(f(x)\) be a function with the following properties:

- \(f\) is continuous everywhere on the closed interval \([a,b]\),
- \(f'(x)\) exists at each point of the open interval

\((a,b)\)

If \(f(a)=f(b)\), then there is at least one point \(c\) in the open interval \((a,b)\) such that \(f'(c)=0\).

This theorem is obviously true, because as \(x\) increases from \(a\) to \(b\), \(f(x)\) cannot always increase or always decrease as \(x\) increases, since \(f(a)=f(b)\). Hence, for at least one value of \(x\) between \(a\) and \(b\), \(f(x)\) must cease to increase and begin to decrease, or else cease to decrease and begin to increase, and for that particular value of \(x\), the first derivative must be zero (because that particular value of \(x\) is either a local maximum or local minimum, and Fermat’s

Theorem 31

tells us that the first derivative at that specific point must be zero).

66

Because \(f\) is continuous, we know by the Extreme-Value Theorem 26 that \(f\) has an absolute maximum and an absolute minimum on \([a,b]\). If both the absolute maximum and the absolute minimum occur at the endpoints, then because \(f(a)=f(b)\), the function is constant between \(a\) and \(b\) and \(f'(x)=0\) for every \(x\in(a,b)\). If \(f\) is not constant, then at least one of the absolute maximum or the absolute minimum occurs at an interior point. An interior extremum is also a local extremum and because \(f'(x)\) exists at every point by hypothesis, it follows from

Fermat’s 31

that the derivative must be zero there .

- If the conditions of Rolle’s theorem are satisfied, then there is
*at least one point*\(c\in(a,b)\) such that \(f'(c)=0\), but there may be more than one such point in the interval. For example, there are three points for which the derivative of the function is zero in Figure 2.

In a special case where \(f(a)=f(b)=0\), Rolle’s theorem states that between any pairs of roots of \(f(x)\) lies a root of \(f'(x)\).

- We say \(x=a\) is a root of \(f(x)\) if \(f\) vanishes for \(x=a\); that is, \(f(a)=0\). If we plot the graph of \(f\), then this graph crosses the \(x\)-axis at \(x=a\).

For example, let \[f(x)=(x-3)(x-2)(x+1)(x+2)(x+4).\] This function becomes zero at \(x=3,2,-1,-2\), and \(-4\). \(f\) is a polynomial and thus it is continuous and differentiable everywhere. It follows from Rolle’s Theorem that \(f'(x)\) has (at least) one zero in the interval \((2,3)\), one zero in \((-1,2)\), one zero in \((-2,-1)\), and one in \((-2,-4)\) (See Figure 3).

The zeros of \(g(x)=x^{4}-x^{2}-2\) are \(\pm\sqrt{2}\). So Rolle’s Theorem tells us its derivative \(g'(x)=4x^{3}-2x\) has one zero between \(-\sqrt{2}\) and \(\sqrt{2}\). In fact, \(g'(x)=2x(2x^{2}-1)\) has three roots between \(-\sqrt{2}\) and \(\sqrt{2}\). The roots are \(x=0\) and \(x=\pm1/\sqrt{2}\) (See Figure 4).

We need to emphasize that \(f(x)\) must be continuous for all points in the closed interval \([a,b]\), including the end-points, but its derivative does not need to exist at the the end-points. If \(f(x)\) is discontinuous even at one point of the closed interval or \(f'(x)\) does not exist at an interior point of the interval, then Rolle’s rule does not apply. For example, Fig. 6(a) shows the graph of a function that is discontinuous (goes to \(\pm\infty\)) for \(x=d\), a value lying between \(a\) and \(b\). Or Fig. 6(b) and (c) show the graph of continuous functions whose first derivatives are discontinuous for such an intermediate value \(x=d\). In all of these cases, it is seen that no point on the graph between \(x=a\) and \(x=b\) does the tangent to the curve become parallel to the

\(x\)-axis.

(a) | (b) | (c) |

Figure 6. |

#### A physical interpretation

Consider a train that moves on a straight railroad. If we see the train in a train station at 10 am and see the same train in the same station at 4 pm, we know that at some point, its speed was zero, because if the train never left the station, its speed was zero all the time between 10 am to 4 pm, and if it left the station and came back, it had first to stop, even for a second, to change its direction and come back. This is exactly what Rolle’s theorem states.

An important usage of Rolle’s theorem is in proving the mean-value theorem that will be discussed in the next section.