3.1 Definition of the Derivative

We begin with two questions: What is the precise definition of a tangent line? And how can we compute its slope? To answer these questions, let’s return to the relationship between tangent and secant lines first mentioned in Section 2.1.

The secant line through distinct points \(P = (a, f(a))\) and \(Q = (x, f(x))\) on the graph of a function \(f(x)\) has slope [Figure 3.1(A)]

\[\frac{\Delta f}{\Delta x} = \frac{f(x) - f(a)}{x-a}\]

where

\[\Delta f = f(x) - f(a)\text{ and }\Delta x = x - a\]

REMINDER A secant line is any line through two points on a curve or graph.

The expression \(\dfrac{f(x) - f(a)}{x-a}\) is called the difference quotient.

Figure 3.1: The secant line has slope \(\frac{\Delta f}{\Delta x}\). Our goal is to compute the slope of the tangent line at \((a, f(a))\).

Now observe what happens as \(Q\) approaches \(P\) or, equivalently, as \(x\) approaches \(a\). Figure 3.2 suggests that the secant lines get progressively closer to the tangent line. If we imagine \(Q\) moving toward \(P\), then the secant line appears to rotate into the tangent line as in (D). Therefore, we may expect the slopes of the secant lines to approach the slope of the tangent line.

Based on this intuition, we define the derivative \(f(a)\) (which is read “\(f\) prime of \(a\)”) as the limit

\[f'(a) = \underbrace{\lim\limits_{x\rightarrow a}\frac{f(x) - f(a)}{x-a}}_{\scriptstyle\text{Limit of slopes of secant lines}}\]

121

Figure 3.2: The secant lines approach the tangent line as \(Q\) approaches \(P\).

There is another way of writing the difference quotient using a new variable \(h\):

\[h = x - a\]

We have \(x = a + h\) and, for \(x \neq a\) (Figure 3.3),

\[\frac{f(x) - f(a)}{x-a} = \frac{f(a+h) - f(a)}{h}\]

The variable \(h\) approaches 0 as \(x \rightarrow a\), so we can rewrite the derivative as

\[f'(a) = \lim\limits_{h\rightarrow 0}\frac{f(a+h) - f(a)}{h}\]

Each way of writing the derivative is useful. The version using h is often more convenient in computations.

Figure 3.3: The difference quotient can be written in terms of \(h\).

DEFINITION The Derivative

The derivative of \(f(x)\) at \(x = a\) is the limit of the difference quotients (if it exists):

\[f'(a) = \lim\limits_{h\rightarrow 0}\frac{f(a+h) - f(a)}{h}\tag{1}\]

When the limit exists, we say that \(f\) is differentiable at \(x = a\). An equivalent definition of the derivative is

\[f'(a) = \lim\limits_{x\rightarrow a}\frac{f(x) - f(a)}{x-a}\tag{2}\]

We now have the concepts to define precisely what is meant by a curve given by \(y=f(x)\) having a (non-vertical) tangent line at the point \(P(a,f(a))\), namely as the line with slope \(f'(a)\) through \(P\).

DEFINITION Tangent Line

Assume that \(f(x)\) is differentiable at \(x = a\). The tangent line to the graph of \(y = f(x)\) at \(P = (a, f(a))\) is the line through \(P\) of slope \(f'(a)\). The equation of the tangent line in point-slope form is

REMINDER The equation of the line through \(P = (a, b) \) of slope \(m\) in point-slope form:

\(y - b = m(x - a)\)

\[y-f(a) = f'(a) (x-a)\tag{3}\]

122

EXAMPLE 1 Equation of a Tangent Line

Find an equation of the tangent line to the graph of \(f(x) = x^2\) at \(x = 5\).

Solution First, we must compute \(f'(5)\). We are free to use either Eq. (1) or Eq. (2). Using Eq. (2), we have

\begin{align*} f'(5) & =\lim\limits_{x\rightarrow 5} \frac{f(x) - f(5)}{x-5} = \lim\limits_{x\rightarrow 5} \frac{x^2 - 25}{x-5} = \lim\limits_{x\rightarrow 5}\frac{(x-5)(x+5)}{x-5}\\[5pt] & =\lim\limits_{x\rightarrow 5} (x+5) = 10 \end{align*}

Next, we apply Eq. (3) with \(a = 5\). Because \(f(5) = 25\), an equation of the tangent line is \(y - 25 = 10(x - 5)\), or, in slope-intercept form: \(y = 10x - 25\) (Figure 3.4).

Figure 3.4: The tangent line to the curve \(y = x^{2}\) at \(x = 5\) has the equation \(y=10x-25\). Move the slider to see other tangent lines and their equations.

Isaac Newton referred to calculus as the “method of fluxions” (from the Latin word for “flow”), but the term “differential calculus”, introduced in its Latin form “calculus differentialis” by Gottfried Wilhelm Leibniz, eventually won out and was adopted universally.

The next two examples illustrate differentiation (the process of computing the derivative) using Eq. (1). For clarity, we break up the computations into three steps.

EXAMPLE 2

Compute \(f'(3)\), where \(f(x) = x^{2} - 8x\).

Solution Using Eq. (1), we write the difference quotient at \(a = 3\) as

\[\frac{f(a+h)-f(a)}{h} = \frac{f(3+h)-f(3)}{h} \]

Step 1. Write out the numerator of the difference quotient.

\begin{align*}\frac{f(3+h)-f(3)}{h} & = \left((3+h)^2 - 8(3+h)\right) - \left(3^2 - 8(3)\right)\\ & = \left((9+6h+h^2) - (24+8h)\right) - \left(9-24\right)\\ & = h^2 - 2h\end{align*}

Step 2. Divide by \(h\) and simplify.

\[\frac{f(3+h)-f(3)}{h} = \frac{h^2 - 2h}{h} = \underbrace{\frac{h(h-2)}{h}=h-2}_{\text{Cancel }h}\]

Step 3. Compute the limit.

\[f'(3) = \lim\limits_{h\rightarrow 0}\frac{f(3+h)-f(3)}{h} =\lim\limits_{h\rightarrow 0}h-2 = -2 \]

Question 3.1 Derivative Progress Check 1

xdnzduW6ihz5qzz6PvsbfdQYYWB12WxK4CJNNlOpL0IlX+Wq+h6XLhl2fE640Zn6UOfb9K7AtkKS0KkAPzXs9yIJ/UjJ7NLsiskWLHb4q27oXeg9ii5ttZUetFIbhuk/lJ67fTm5SgXC8+wMta5jNQqQPbZtubR0kMRHsYgFv0V0GzReZwSHaX8AYgaM1ORXWB+kympXYqVyhnu8JoNoLQxP9fD8JQ/i
3
Correct.
Try again. First find the slope by computing \(f'(-2)\). Also, remember to answer this question by typing in an equation in the form \(y=mx+b\). In other words, if you find that the slope is \(5\) and the \(y\)-intercept \(12\), you would have to type in the equation \(y=5x+12\).
Incorrect.

EXAMPLE 3

Sketch the graph of \(f(x)=\dfrac{1}{x}\) and the tangent line at \(x = 2\).

(a) Based on the sketch, do you expect \(f'(2)\) to be positive or negative?

(b) Find an equation of the tangent line at \(x = 2\).

Solution The graph and tangent line at \(x = 2\) are shown in Figure 3.5.

Figure 3.5: Graph of \(f(x) = \frac{1}{x}\). The tangent line at \(x = 2\) has equation \(y=-\frac{1}{4}x+1\).

(a) We see that the tangent line has negative slope, so \(f'(2)\) must be negative.

(b) We compute \(f'(2)\) in three steps as before.

Step 1. Write out the numerator of the difference quotient.

\[f(2+h) - f(2) = \frac{1}{2+h} - \frac{1}{2} = \frac{2}{2(2+h)} - \frac{2+h}{2(2+h)} = -\frac{h}{2(2+h)}\]

Step 2. Divide by \(h\) and simplify.

\[\frac{f(2+h) - f(2)}{h} = \frac{1}{h}\cdot\left(-\frac{h}{2(2+h)}\right) = -\frac{1}{2(2+h)}\]

123

Step 3. Compute the limit.

\[f'(2) = \lim\limits_{h\rightarrow 0}\frac{f(2+h) - f(2)}{h} =\lim\limits_{h\rightarrow 0}-\frac{1}{2(2+h)} = -\frac{1}{4}\]

The function value is \(f(2) = \frac{1}{2}\), so the tangent line passes through \(\left(2,\frac{1}{2}\right)\) and has equation

\[y-\frac{1}{2} = -\frac{1}{4}(x-2)\]

In slope-intercept form, \(y=-\frac{1}{4}x+1\).

Question 3.2 Derivatives Progress Check 2

6Cjm/CyOAGmVZJ2ln1h1J71/1ogX4ppuVxu1SfsOthvYlOCNf/lOYc9gWDNtx6Qqlvk4LFz6f6uyO6kL
3
Correct.
Try again. Be careful with your algebra. This problem is not easy. You will later see (in section 3.7 to be precise) how much easier this particular problem becomes once we can evaluate the derivative more directly.
Incorrect.

The graph of a linear function \(f(x) = mx + b\) (where \(m\) and \(b\) are constants) is a line of slope \(m\). The tangent line at any point coincides with the line itself (Figure 3.6), so we should expect that \(f'(a) = m\) for all \(a\). Let’s check this by computing the derivative:

\begin{align*}f'(a) &= \lim\limits_{h\rightarrow 0} \frac{f(a+h)-f(a)}{h} = \lim\limits_{h\rightarrow 0} \frac{(m(a+h)+b)-(ma+b)}{h} \\[5pt] & =\lim\limits_{h\rightarrow 0}\frac{mh}{h} = \lim\limits_{h\rightarrow 0}m = m \end{align*}

Figure 3.6: The derivative of \(f(x) = mx + b\) is \(f'(a) = m\) for all \(a\).

If \(m = 0\), then \(f(x) = b\) is constant and \(f'(a) = 0\) (Figure 3.7). In summary,

THEOREM 1 Derivative of Linear and Constant Functions

  • If \(f(x) = mx + b\) is a linear function, then \(f'(a) = m\) for all \(a\).
  • If \(f(x) = b\) is a constant function, then \(f'(a) = 0\) for all \(a\).
Figure 3.7: The derivative of a constant function \(f(x) = b\) is \(f'(a) = 0\) for all \(a\).

EXAMPLE 4

Find the derivative of \(f(x) = 9x - 5\) at \(x = 2\) and \(x = 5\).

Solution We have \(f'(a) = 9\) for all \(a\). Hence, \(f'(2) = f'(5) = 9\).

Question 3.3 Derivative Progress Check 3

MuvEjcMwCiSP3nQ24XXeTInbIyz2M+nF9XLRsMzaR37YgcNgSBcJSIt6jSbafNQvOpacEVZM+v8jPmzilylpmuPAEWQ=
3
Correct.
Try again. Find \(f'(x)\) and see what you get.
Incorrect.

3.1.1 Estimating the Derivative

Approximations to the derivative are useful in situations where we cannot evaluate \(f(a)\) exactly. Since the derivative is the limit of difference quotients, the difference quotient should give a good numerical approximation when \(h\) is sufficiently small:

\[f'(a)\approx\frac{f(a+h)-f(a)}{h}\text{ if }h\text{ is small.}\]

Graphically, this says that for small \(h\), the slope of the secant line is nearly equal to the slope of the tangent line (Figure 3.8).

Figure 3.8: When \(h\) is small, the secant line has nearly the same slope as the tangent line.

EXAMPLE 5

Estimate the derivative of \(f(x) = \sin x\) at \(x=\frac{\pi}{6}\).

Solution We calculate the difference quotient for several small values of \(h\):

\[\frac{\sin(\frac{\pi}{6} + h)-\sin\frac{\pi}{6}}{h} = \frac{\sin(\frac{\pi}{6} + h)-0.5}{h}\]

Table 3.1 below suggests that the limit has a decimal expansion beginning \(0.866\). In other words, \(f'(\frac{\pi}{6})\approx0.866\).

124

\(h > 0\) \(\frac{\sin(\frac{\pi}{6} + h)-0.5}{h}\) \(h < 0\) \(\frac{\sin(\frac{\pi}{6} + h)-0.5}{h}\)
0.01 0.863511 −0.01 0.868511
0.001 0.865775 −0.001 0.866275
0.0001 0.8660 00 −0.0001 0.866 050
0.00001 0.8660 229 −0.00001 0.8660 279
Table 3.1: Values of the Difference Quotient for Small \(h\)

In the next example, we use graphical reasoning to determine the accuracy of the estimates obtained in Example 5.

EXAMPLE 6 Determining Accuracy Graphically

Let \(f(x) = \sin x\). Show that the approximation \(f'(\frac{\pi}{6})\approx0.8660\) is accurate to four decimal places.

Solution Observe in Figure 3.9 that the position of the secant line relative to the tangent line depends on whether \(h\) is positive or negative. When \(h > 0\), the slope of the secant line is smaller than the slope of the tangent line, but it is larger when \(h < 0\). This tells us that the difference quotients in the second column of Table 3.1 are smaller than \(f'(\frac{\pi}{6})\) and those in the fourth column are greater than \(f'(\frac{\pi}{6})\). From the last line in Table 3.1 we may conclude that

Figure 3.9: The tangent line is squeezed in between the secant lines with \(h > 0\) and \(h < 0\).

This technique of estimating an unknown quantity by showing that it lies between two known values (“squeezing it”) is used frequently in calculus.

\[0.866022\leq f'(\tfrac{\pi}{6}) \leq 0.866028\]

It follows that the estimate \(f'(\frac{\pi}{6})\approx0.866\) is accurate to four decimal places. In Section 3.6, we will see that the exact value is \(f'(\frac{\pi}{6})=\cos(\frac{\pi}{6})=\frac{\sqrt{3}}{2} \approx0.8660254\), just about midway between 0.866022 and 0.866028.

CONCEPTUAL INSIGHT Are Limits Really Necessary?

It is natural to ask whether limits are really necessary. The tangent line is easy to visualize. Is there perhaps a better or simpler way to find its equation? History gives one answer: The methods of calculus based on limits have stood the test of time and are used more widely today than ever before.

History aside, we can see directly why limits play such a crucial role. The slope of a line can be computed if the coordinates of two points \(P = (x_{1}, y_{1})\) and \(Q = (x_{2}, y_{2})\) on the line are known:

\[\text{Slope of Line} = \frac{y_2-y_1}{x_2 - x_1}\]

This formula cannot be applied to the tangent line because we know only that it passes through the single point \(P = (a, f(a))\). Limits provide an ingenious way around this difficulty. We choose a point \(Q = (a + h, f (a + h))\) on the graph near \(P\) and form the secant line. The slope of this secant line is just an approximation to the slope of the tangent line:

\[\text{Slope of the secant line} = \frac{f(a+h)-f(a)}{h} \approx \text{slope of tangent line}\]

But this approximation improves as \(h \rightarrow 0\), and by taking the limit, we convert our approximations into the exact slope.

125

3.1.2 Section 3.1 Summary

\[\frac{f(a+h)-f(a)}{h}\]

The difference quotient is the slope of the secant line through the points \(P = (a, f(a))\) and \(Q = (a + h, f(a + h))\) on the graph of \(f(x)\).

\[f'(a) = \lim\limits_{h\rightarrow 0}\frac{f(a+h)-f(a)}{h} = \lim\limits_{x\rightarrow a}\frac{f(x) - f(a)}{x-a}\]

If the limit exists, we say that \(f\) is differentiable at \(x = a\).

\[y - f(a) = f'(a)(x- a)\]

Step 1. Write out the numerator of the difference quotient.
Step 2. Divide by \(h\) and simplify.
Step 3. Compute the derivative by taking the limit.