3.1 Defining the Derivative

The tangent line as the limit of secant lines, the two equivalent forms of the difference quotient, computing a derivative at a point directly from the definition, writing the equation of a tangent line, velocity as an instantaneous rate of change, and estimating a derivative from a table of values.

Subject: Calculus I · 67 slides · symbolic lesson

Open the interactive version of this deck

What this lesson covers

The lesson, slide by slide

1. Defining the Derivative

Title

Calculus I · Section 3.1

One limit that measures slope, speed and every other rate of change

2. What this lesson gives you

Objectives

Chapter 2 built limits. This section spends them: one particular limit, the limit of a difference quotient, turns out to answer the tangent problem and the velocity problem at once, and it gets a name of its own, the derivative.

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 188 — learning objectives 3.1.1 to 3.1.7

Chapter 2 was an investment. You learned what a limit is, how to compute one, and how to cope with the zero over zero that keeps appearing. This section is where the investment pays out: a single limit, built from a difference quotient, solves the tangent problem and the velocity problem from Section 2.1 in one stroke.

That limit is the derivative, and it is the most important object in the rest of this course. Everything in Chapter 3 either computes derivatives faster or uses them to say something about a function.

By the end of today you should be able to compute a derivative at a point straight from its definition, turn it into a tangent line, read it as a rate with units, estimate it from data, and spot the places where it does not exist.

3. Before anything new: a secant slope

Warm-up

Discussion prompt

On the parabola y = x², find the slope of the secant line through (3, 9) and (4, 16). Then say what you would do to get the slope of the tangent line at (3, 9), using only secants.

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §2.1, pp. 108-111 — the tangent problem

Commit to an answer before you reveal. The secant slope is ordinary rise over run: the heights 9 and 16, the inputs 3 and 4, so the slope is 7.

The second half is the real question. A secant through two points is not the tangent, but you can make it better by moving the second point closer. At 3.1 the slope is 6.1, at 3.01 it is 6.01, and the pattern is plain: the slopes are homing in on 6.

Section 2.1 stopped there, reading the answer off a table by eye. Today you replace the eye with a limit, which turns a good guess into an exact value.

4. Secant lines and tangent lines

Section

Part 1

5. The slope of a secant line

Concept

Figure (svg): Two copies of the same upward-curving graph with a secant line through a point at a and a point further right. Left: the far point is labelled x and the horizontal run is x minus a. Right: the far point is labelled a plus h and the run is h. The vertical rise is marked in both.

One line, two names for its second point. The slope is rise over run either way; only the bookkeeping differs.

A secant joins two points on a graph. Its slope is rise over run, and when the first point sits at a and the second at x, that ratio has a name.

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

This is the difference quotient of Definition 3.1: a difference of outputs over the matching difference of inputs.

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, pp. 188-189 — Definition, equation 3.1, and Figure 3.3

Look at the left panel. The two marked points are the fixed point at a and a second point at x. The dashed yellow legs are the run, along the horizontal, and the rise, straight up. The orange secant is the line through both points, and its slope is rise over run.

Written with function notation, the rise is f of x minus f of a and the run is x minus a. That ratio is called a difference quotient, because it is a difference of outputs divided by a difference of inputs. It is also the average rate of change of f between the two inputs, which is the reading you will use in Part 3.

The right panel is the same picture with the second point renamed. Hold that thought for two slides.

6. Reading the difference quotient piece by piece

Notation

Annotate

On: \( Q = \frac{f(x) - f(a)}{x - a} \)

  • The rise: how much the output changes between the two points. It can be negative when the graph falls.
  • The run: how far apart the two inputs are. It must not be zero, so x ≠ a.
  • The slope of the secant, which is also the AVERAGE rate of change of f between a and x.
  • Think of a as pinned and x as free to slide. Everything in this lesson happens as x slides toward a.

Step through the annotations one at a time. The numerator is a rise and can be negative if the graph falls between the two points; the denominator is a run and must never be zero, because two equal inputs do not determine a line.

The most useful way to picture this formula is with a pinned and a free point. The input a is pinned. The input x is free to slide along the axis, dragging its point along the curve. Each position of x gives a different secant and a different quotient.

The whole of this lesson is about what happens to that quotient as x slides toward a. It never reaches a, since the run would be zero, but it can get as close as you like.

7. Name the second point by its distance h

Concept

Instead of saying where the second point is, say how far it is from a. Call that distance h, so the second input is a plus h.

\[ x = a + h \quad\iff\quad h = x - a \]

Substitute into the difference quotient: the run becomes h, and the second output becomes f of a plus h.

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

This is the difference quotient with increment h, equation 3.2. Same number, different bookkeeping; h may be negative, which puts the second point to the left of a.

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 189 — equation 3.2

This is a change of bookkeeping, not a change of mathematics. Instead of naming the second input by where it sits, you name it by how far it is from a. That distance is h, so the second input is a plus h.

Follow the substitution line by line. The run, x minus a, becomes a plus h minus a, which is just h. The second output becomes f of a plus h. The result is the difference quotient with increment h, equation 3.2 in the book.

Two small points save a lot of confusion later. First, h can be negative: that simply puts the second point to the left of a. Second, the quotient has the same value in both forms for the same pair of points, so you are free to use whichever form makes your algebra easier.

8. Secants swinging into the tangent

Picture it

Figure (svg): The parabola y equals x squared with the point 3, 9 marked. Three orange dashed secants run to points right of 3 at 5, 4.2 and 3.5; three purple dashed secants run to points left of 3 at 1, 2 and 2.5. A solid yellow tangent line of slope 6 sits between the two families.

Right-hand secants are too steep, left-hand ones too shallow, and both families close in on the same line: the tangent, slope 6.

Every secant through (3, 9) on the parabola has slope x plus 3, where x is the second point. Slide x toward 3 from either side and the slopes close in on one number, 6. The line with that slope through (3, 9) is the tangent.

\[ \frac{x^2 - 9}{x - 3} = x + 3 \]

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, pp. 189-190 — Figure 3.4

In the figure the fixed point is 3, 9 on the parabola. The orange dashed lines are secants to points on the right, at 5, 4.2 and 3.5; the purple ones go to points on the left, at 1, 2 and 2.5. Read their slopes from the labels: the right-hand ones come down, 8, 7.2, 6.5, and the left-hand ones go up, 4, 5, 5.5.

Both families are squeezing toward the solid yellow line, which has slope exactly 6. That line is the tangent. The formula under the figure explains the pattern: for this parabola, every secant through 3, 9 has slope x plus 3, and x plus 3 tends to 6 as x tends to 3.

Notice the geometry too. Because the parabola bends upward, right-hand secants are steeper than the tangent and left-hand ones are shallower. The tangent slope is the one number trapped between them.

9. Secants from the left: too steep or too shallow?

Prediction

\[ f(x) = x^2, \quad a = 3, \quad m_{\tan} = 6 \]

Predict first

Take the secant from (3, 9) to a point on y = x² just to the LEFT of 3, say at x = 2.9. Compared with the tangent slope 6, its slope is…

  • less than 6
  • more than 6
  • exactly 6
  • undefined, because the run is negative

Correct: less than 6

Why: The secant slope is x + 3, which is 5.9 at x = 2.9. A negative run pairs with a negative rise, so the slope is positive and perfectly defined. On a graph that bends upward, left-hand secants are too shallow and right-hand ones too steep, and the tangent is squeezed between them.

Pick an answer before revealing. The trap here is the negative run. When the second point is on the left, x minus 3 is negative, which can feel like it should break something.

It does not. The rise is negative too, because the parabola is lower at 2.9 than at 3, and a negative over a negative is positive. The slope is 5.9, a little less than 6.

Keep the picture from the previous slide in mind: on a curve that bends upward, secants on the left undershoot the tangent slope and secants on the right overshoot it. That is why a table needs both sides: each side gives a bound, and the derivative is where they meet.

10. Slide h and watch the secant pivot

Tweak it

Parameter explorer

The line drawn is the secant to y = x² through (3, 9) and (3 + h, (3 + h)²). Slide h toward 0 from both sides. What slope does the line settle on, and does the line ever break at h = 0?

\[ h = {h}, \qquad m_{\text{sec}} = 6 + {h} \]

  • h — from -2 to 2: increment h

\[ \frac{(3+h)^2 - 9}{h} = \frac{6h + h^2}{h} = 6 + h \quad (h \ne 0) \]

Start with h at 1.5 and drag it slowly toward zero. The line pivots about the point 3, 9, and the readout shows its slope, 6 plus h. Now push h past zero into negative values and watch the line keep turning smoothly through the tangent.

The formula under the slider is the whole derivative computation in one line. The difference quotient simplifies to 6 plus h for every nonzero h. At h equal to zero the original quotient is undefined, but the simplified expression is perfectly happy there, and its value, 6, is the limit.

That is the pattern of every calculation in this lesson: simplify the quotient until the h in the denominator is gone, and then letting h go to zero is just substitution.

11. Zoom in until the curve is a line

Picture it

Figure (svg): Three panels of y equals the square root of x (blue) and its tangent line y equals x plus 1 over 2 (orange) at the point 1, 1. Left, x from 0 to 3: the curve bends visibly away from the line. Middle, x from 0.7 to 1.3: the gap is small. Right, x from 0.97 to 1.03: the two are indistinguishable.

Each panel is a closer look at the same point. Zoom in far enough and the curve is its tangent line: the graph is locally linear.

The book zooms in on the root curve at the point 1, 1 together with its tangent. In a window 0.06 wide the two can no longer be told apart. That is local linearity: near a point where a tangent exists, the curve and its tangent line are nearly the same.

\[ \sqrt{1.02} \approx 1.00995, \qquad \frac{1.02 + 1}{2} = 1.01 \]

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 190 — Figure 3.5

This is Figure 3.5 in the book, drawn with the root curve and its tangent at the point 1, 1. In the left panel you can clearly see the curve bending away from the orange line. In the middle panel, a window a fifth as wide, the gap is barely visible. In the right panel the two are one line.

The numbers under the figure make the same point. At 1.02, the curve is about 1.00995 and the tangent line gives 1.01: they differ by about five hundred-thousandths.

This property is called local linearity, and it is the practical reason derivatives matter. Close to a point, a complicated function behaves like its tangent line, and a line is easy to work with. Section 4.2 builds linear approximation on exactly this picture.

12. A tangent is not a line that touches once

Intuition

Figure (svg): Left: the cubic y equals x cubed with its tangent at 0, the x-axis, which passes straight through the curve. Right: the parabola y equals x squared with the vertical line x equals 1, which meets the curve exactly once but is not a tangent.

Touching once is the wrong test in both directions. A tangent is defined by a limit of slopes, not by how many times it meets the curve.

The school-geometry picture, a line that touches a circle at one point, breaks down for general graphs. The horizontal axis is the tangent to the cubic at the origin and passes straight through it; a vertical line meets a parabola once and is not a tangent at all.

So the definition cannot count intersections. It uses the only thing that works for every graph: the limit of secant slopes.

You probably met tangent lines first with circles, where a tangent is a line touching the circle at exactly one point. That definition does not survive contact with general graphs, and the two panels show why.

On the left, the horizontal axis is the tangent to the cubic at the origin: the cubic is momentarily flat there. Yet the axis passes straight through the curve. On the right, a vertical line meets the parabola at exactly one point, and it is obviously not the tangent; it is not even close to the direction of the curve.

So counting intersections is the wrong test, in both directions. What does work, for every graph, is the limit of secant slopes. That is the definition on the next slide.

13. The tangent line, defined

Concept

Tangent line (equations 3.3 and 3.4) — The line through the point with first coordinate a on the graph of f whose slope is the limit of the secant slopes, provided that limit exists.

\[ m_{\tan} = \lim_{x\to a}\frac{f(x) - f(a)}{x - a} \]

\[ m_{\tan} = \lim_{h\to 0}\frac{f(a+h) - f(a)}{h} \]

Two forms, one number. Choose whichever makes the algebra easier; the rest of the lesson shows how to tell.

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 191 — Definition, equations 3.3 and 3.4

Here is the formal definition, in both forms. The tangent line at a is the line through the point on the graph above a whose slope is the limit of the secant slopes. The first formula uses the x form; the second uses the h form.

Pay attention to the phrase provided this limit exists. The definition is honest: it does not promise that every graph has a tangent at every point. Part 4 is about the places where the limit fails.

In practice, a tangent line needs two ingredients, a point and a slope. The limit supplies the slope. The point is simply the point on the graph, a and f of a, which you get by substituting. Keep those two jobs separate and tangent lines become routine.

14. Example 3.1: the tangent to x squared at 3

Worked example

Find an equation of the line tangent to the graph at x equal to 3, using the x form.

\[ f(x) = x^2, \quad a = 3 \]

Apply the definition

Why: Equation 3.3 with a equal to 3.

\[ m_{\tan} = \lim_{x\to 3}\frac{f(x) - f(3)}{x - 3} \]

Substitute the function

Why: The value at 3 is 9.

\[ = \lim_{x\to 3}\frac{x^2 - 9}{x - 3} \]

Factor the numerator

Why: A difference of squares.

\[ = \lim_{x\to 3}\frac{(x-3)(x+3)}{x - 3} \]

Cancel and evaluate

Why: Inside the limit x is never 3, so the common factor may cancel.

\[ = \lim_{x\to 3}(x + 3) = 6 \]

Find the point

Why: The tangent passes through the point on the graph.

\[ f(3) = 9 \;\Longrightarrow\; (3, 9) \]

Use point-slope form

Why: Slope 6 through the point 3, 9.

\[ y - 9 = 6(x - 3) \]

Simplify

Why: Distribute and add 9.

\[ y = 6x - 9 \]

Figure (svg): The parabola y equals x squared from x equals minus 1 to 4.5 and the line y equals 6x minus 9, which touches it at the point 3, 9.

The line shares the point (3, 9) with the parabola and has the parabola's slope there, 6.

Check the point and the slope

Why: The line gives 9 at x equal to 3, and a secant over a tiny interval has slope next to 6.

\[ 6(3) - 9 = 9, \qquad \frac{3.001^2 - 9}{0.001} = 6.001 \]

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 191 — Example 3.1 and Figure 3.6

Follow the lines with the words covered: they run from the definition to the answer without a gap. Substituting gives x squared minus 9 over x minus 3, which is zero over zero at 3. Factoring the difference of squares exposes the factor x minus 3, and cancelling it is legitimate because inside the limit x is never equal to 3.

The slope is 6. The point comes from a different calculation: the curve's height at 3, which is 9. Point-slope form then gives y minus 9 equals 6 times x minus 3, and tidying up gives y equals 6x minus 9. The figure shows the line resting against the parabola at 3, 9.

The check does two things at once. The line passes through the right point, and a secant over a tiny interval has a slope of 6.001, right next to 6. Both halves of the tangent line are confirmed.

15. Example 3.2: the same slope in the h form

Worked example

Use equation 3.4 to find the slope of the tangent to the same parabola at x equal to 3.

Apply the definition

Why: Equation 3.4 with a equal to 3.

\[ m_{\tan} = \lim_{h\to 0}\frac{f(3+h) - f(3)}{h} \]

Substitute the function

Why: Square the input 3 plus h; the value at 3 is 9.

\[ = \lim_{h\to 0}\frac{(3+h)^2 - 9}{h} \]

Expand the square

Why: Three terms, not two.

\[ = \lim_{h\to 0}\frac{9 + 6h + h^2 - 9}{h} \]

Simplify the numerator

Why: The nines cancel, which is exactly what makes the limit work.

\[ = \lim_{h\to 0}\frac{6h + h^2}{h} \]

Factor out h

Why: Now the zero in the denominator has a matching zero on top.

\[ = \lim_{h\to 0}\frac{h(6 + h)}{h} \]

Cancel and evaluate

Why: h is not zero inside the limit.

\[ = \lim_{h\to 0}(6 + h) = 6 \]

Check against Example 3.1 and a number

Why: Same slope as the x form, and h equal to 0.01 gives 6.01.

\[ \frac{3.01^2 - 9}{0.01} = \frac{0.0601}{0.01} = 6.01 \]

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 192 — Example 3.2

The same slope, reached through the other door. Now you expand three plus h, squared, which gives 9 plus 6h plus h squared. Be careful with that expansion: the middle term, 6h, is the one that survives, and forgetting it is the most common slip in this whole topic.

Once the nines cancel, every remaining term on top has a factor of h. Factor it, cancel it, and what is left, 6 plus h, can be evaluated at h equal to zero. The answer is 6 again.

The two forms had to agree, since they are the same limit written with a change of variable. Seeing it happen once is reassuring; after that, just choose the form whose algebra you like better for the function in front of you.

16. Trap: putting h equal to zero too early

Trap

The trap

A line that appears on a lot of first attempts:

\[ \lim_{h\to 0}\frac{(3+h)^2 - 9}{h} = \frac{9 - 9}{0} \]

\[ = \frac{0}{0} \;\Longrightarrow\; \text{no slope?} \]

Wrong. Zero over zero is not an answer.

The fix

Every difference quotient gives zero over zero at h equal to 0; that is built in. It means simplify first: expand, cancel the h, and only then let h go to 0.

\[ \frac{6h + h^2}{h} = 6 + h \to 6 \]

This mistake comes from treating a limit as if it were substitution. It is true that for continuous functions you can often find a limit by plugging in. But the difference quotient is never continuous at h equal to zero, because it is not even defined there.

Zero over zero is not a number and not a verdict. It is a signal that the numerator and denominator share a factor, and that you have more algebra to do. Expand, simplify, cancel, and only then substitute. After cancelling, 6 plus h is continuous, so direct substitution is now allowed and gives 6.

17. Why h is always a factor

Concept

The numerator of the h form is a function of h that equals zero when h is zero, because both terms are then the same number.

\[ N(h) = f(a+h) - f(a), \qquad N(0) = f(a) - f(a) = 0 \]

For a polynomial, a zero at h equal to 0 means h is a factor. That is why every polynomial example factors so neatly.

\[ N(h) = (2+h)^3 - 8 = 12h + 6h^2 + h^3 = h(12 + 6h + h^2) \]

For reciprocals and roots the factor is still there, but hidden: combining fractions or multiplying by a conjugate is how you dig it out.

Here is the reason every calculation works out, and it is worth understanding rather than trusting. The numerator of the h form subtracts f of a from f of a plus h. At h equal to zero those two terms are the same number, so the numerator is zero.

For a polynomial in h, being zero at h equal to zero means there is no constant term, and that means h is a factor. The example with two plus h, cubed, minus 8 shows it happening: every term that survives carries an h.

For reciprocals and roots the factor is still there, just disguised. Combining small fractions or multiplying by a conjugate are ways of rewriting the numerator so the factor becomes visible. If you ever finish your simplification and h has not appeared as a factor, you have made an algebra slip somewhere.

18. Example 3.3: the tangent to one over x at 2

Worked example

Find an equation of the line tangent to the graph at x equal to 2.

\[ f(x) = \frac1x, \quad a = 2 \]

Apply the definition

Why: The x form; the value at 2 is one half.

\[ m_{\tan} = \lim_{x\to 2}\frac{\frac1x - \frac12}{x - 2} \]

Clear the small fractions

Why: Multiply top and bottom by 2x.

\[ = \lim_{x\to 2}\frac{2 - x}{2x(x - 2)} \]

Expose the common factor

Why: Two minus x is minus one times x minus 2.

\[ = \lim_{x\to 2}\frac{-(x - 2)}{2x(x - 2)} \]

Cancel

Why: x is not 2 inside the limit.

\[ = \lim_{x\to 2}\frac{-1}{2x} \]

Evaluate the limit

Why: Substitute 2.

\[ m_{\tan} = -\frac14 \]

Write the line

Why: Through the point 2, one half with slope minus one quarter.

\[ y - \frac12 = -\frac14(x - 2) \]

Simplify

Why: Distribute and add one half.

\[ y = -\frac14 x + 1 \]

Figure (svg): Both branches of y equals 1 over x for x from minus 6 to 6.5, and the line y equals minus x over 4 plus 1 touching the right-hand branch at the point 2, one half.

The tangent slopes downward, slope minus one quarter, because 1/x is falling at x = 2.

Check with a nearby secant and the point

Why: A secant from 2 to 2.001 has slope about minus 0.25, and the line passes through the point.

\[ \frac{\frac{1}{2.001} - \frac12}{0.001} \approx -0.24988, \qquad -\frac14(2) + 1 = \frac12 \]

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, pp. 192-193 — Example 3.3 and Figure 3.7

Reciprocals bring fractions inside a fraction, and the cleanest cure is to multiply the top and bottom by the product of the small denominators, here 2x. That turns one over x minus one half into 2 minus x, with 2x joining the denominator.

The next move is the one people miss: 2 minus x is the negative of x minus 2. Write it that way and the factor cancels, leaving minus one over 2x. At x equal to 2 that is minus one quarter.

Look at the figure. The curve is falling at x equal to 2, so a negative slope is exactly what you should expect, and the orange line leans downward through the point 2, one half. The check confirms both the slope, with a secant of about minus 0.25, and the point.

19. Checkpoint 3.1: the slope of the root curve at 4

Worked example

Find the slope of the line tangent to the graph at x equal to 4.

\[ f(x) = \sqrt{x}, \quad a = 4, \quad f(4) = 2 \]

Apply the h form

Why: The root of 4 plus h, minus 2, over h.

\[ m_{\tan} = \lim_{h\to 0}\frac{\sqrt{4+h} - 2}{h} \]

Multiply by the conjugate

Why: Top and bottom by the root of 4 plus h, plus 2.

\[ = \lim_{h\to 0}\frac{(\sqrt{4+h} - 2)(\sqrt{4+h} + 2)}{h(\sqrt{4+h} + 2)} \]

Multiply out the top

Why: A difference of squares removes the root.

\[ = \lim_{h\to 0}\frac{(4 + h) - 4}{h(\sqrt{4+h} + 2)} \]

Simplify the top

Why: Only h is left.

\[ = \lim_{h\to 0}\frac{h}{h(\sqrt{4+h} + 2)} \]

Cancel h

Why: h is not zero inside the limit.

\[ = \lim_{h\to 0}\frac{1}{\sqrt{4+h} + 2} \]

Evaluate the limit

Why: The root of 4 is 2.

\[ = \frac{1}{2 + 2} = \frac14 \]

Figure (svg): The curve y equals the square root of x from 0 to 9, and the line y equals x over 4 plus 1 touching it at the point 4, 2.

At x = 4 the root curve rises a quarter of a unit per unit of x, and the tangent shows that slope.

Check with a small h

Why: h equal to 0.001 gives a secant slope a hair under one quarter.

\[ \frac{\sqrt{4.001} - 2}{0.001} \approx 0.24998 \]

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 193 — Checkpoint 3.1

A root minus a number is the signal for the conjugate. Multiply the top and bottom by the root of 4 plus h, plus 2. On top you get a difference of squares, and the square of a square root is just what was inside it, so the root disappears.

What is left on top is 4 plus h minus 4, which is h, and it cancels the h in the denominator. Do not multiply out the denominator; you want to keep the factor h visible so you can cancel it.

The slope is one quarter. The figure shows the tangent at 4, 2, with equation y equals x over 4 plus 1, and the check with h equal to 0.001 gives a secant slope a hair below a quarter, just as the concave shape of the root curve predicts.

20. Exercise 26: rationalise the root

Fill the middle

\[ f(x) = \sqrt{x - 2}, \quad a = 6: \quad f(6 + h) = \sqrt{4 + h}, \quad f(6) = 2 \]

Fill in the blanks

\frach - 2}1/4\cdot\frac___ + 2}___ + 2} = \frac___} }___ + 2)} \;\to\; ___

Why: The conjugate turns the top into (4 + h) minus 4, which is just h. After cancelling h, what is left is one over the root of 4 plus h, plus 2, and at h equal to 0 that is one quarter. The shift by 2 inside the root moves the whole calculation of Checkpoint 3.1 over by two units.

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 199 — Exercise 26

Fill in the two blanks before checking. This is Checkpoint 3.1 shifted two units to the right: the function is the root of x minus 2 at 6, and substituting 6 plus h gives the root of 4 plus h, exactly as before.

After multiplying by the conjugate, the numerator is 4 plus h minus 4, so the first blank is simply h. Cancel it, let h go to zero, and the value is one quarter.

The fact that the answer matches Checkpoint 3.1 is not a coincidence. Shifting a graph sideways moves its tangent lines sideways too, without changing their slopes. You will see that formalised as the chain rule later in the chapter.

21. Exercise 16: a tangent to a shifted root

Worked example

Use equation 3.4 to find the slope, then an equation of the tangent line, at x equal to 1.

\[ f(x) = \sqrt{x + 8}, \quad a = 1, \quad f(1) = \sqrt{9} = 3 \]

Apply the h form

Why: The input 1 plus h gives the root of 9 plus h.

\[ m_{\tan} = \lim_{h\to 0}\frac{\sqrt{9+h} - 3}{h} \]

Multiply by the conjugate

Why: Top and bottom by the root of 9 plus h, plus 3.

\[ = \lim_{h\to 0}\frac{(9 + h) - 9}{h(\sqrt{9+h} + 3)} \]

Cancel h

Why: The top is h.

\[ = \lim_{h\to 0}\frac{1}{\sqrt{9+h} + 3} \]

Evaluate the limit

Why: The root of 9 is 3.

\[ = \frac{1}{3 + 3} = \frac16 \]

Write the line

Why: Slope one sixth through the point 1, 3.

\[ y - 3 = \frac16(x - 1) \]

Simplify

Why: Three is eighteen sixths.

\[ y = \frac16 x + \frac{17}{6} \]

Figure (svg): The curve y equals the square root of x plus 8, starting at x equals minus 8, and its tangent line at the point 1, 3, which has slope one sixth.

Near x = 1 the curve and the line are almost the same; far from it, the curve bends away below the line.

Check close to the point

Why: At x equal to 1.1 the line and the curve agree to four decimal places.

\[ \frac{1.1}{6} + \frac{17}{6} \approx 3.01667, \qquad \sqrt{9.1} \approx 3.01662 \]

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 199 — Exercise 16

This is a book exercise in the same family, and it asks for the whole tangent line, not just the slope. The function value at 1 is the root of 9, which is 3, so the point is 1, 3.

The conjugate removes the root in one step, leaving h over h times the root of 9 plus h, plus 3. Cancel and let h go to zero: one over 6. Then point-slope form gives y minus 3 equals one sixth of x minus 1, which tidies to y equals x over 6 plus 17 over 6.

The check shows local linearity in numbers. At 1.1, a tenth away from the point, the line and the curve agree to four decimal places. The figure shows how the curve bends below the line farther away.

22. Which move clears the zero over zero?

Sorting

Sort into buckets

Each limit is zero over zero as written. Sort it by the algebra move that exposes the cancelling factor.

factor the numerator
lim (x² − 16)/(x − 4), x → 4; lim (x³ − 1)/(x − 1), x → 1
expand, then factor out h
lim ((2 + h)³ − 8)/h, h → 0; lim ((1 + h)² − 1)/h, h → 0
combine the fractions
lim (1/(3 + h) − 1/3)/h, h → 0; lim (2/x − 1)/(x − 2), x → 2
multiply by the conjugate
lim (√(9 + h) − 3)/h, h → 0; lim (√x − 5)/(x − 25), x → 25
fac
x² − 16 = (x − 4)(x + 4) and x³ − 1 = (x − 1)(x² + x + 1): the x form of a polynomial factors directly.
exp
(2 + h)³ − 8 = 12h + 6h² + h³ and (1 + h)² − 1 = 2h + h²: expand the power and h comes out of every term.
den
A fraction inside a fraction: put the small fractions over a common denominator and the factor appears on top.
con
A root minus a number: the conjugate turns the top into a difference with no root in it.

Sort each limit before checking. You are not asked to compute anything, only to recognise which algebra move the expression needs. That recognition is most of the skill in computing derivatives from the definition.

Polynomials in the x form usually factor directly, with x minus a as one factor. Powers of a plus h get expanded, and then h comes out of every term. A fraction inside a fraction wants a common denominator. A root minus a number wants the conjugate.

One item is worth a second look: the root of x minus 5, over x minus 25. You could also factor x minus 25 as a difference of squares of roots. Both routes are the same idea, rewriting the expression so the root is cleared.

23. Find the error: squaring a sum term by term

Error analysis

Annotate

On: \( \lim_{h\to 0}\frac{(2+h)^2 - 4}{h} = \lim_{h\to 0}\frac{4 + h^2 - 4}{h} = \lim_{h\to 0} h = 0 \)

  • The square of a sum is not the sum of the squares: (2 + h)² = 4 + 4h + h².
  • Dropping the middle term threw away the only part that survives the limit.
  • (4h + h²)/h = 4 + h, which tends to 4. The tangent to x² at 2 has slope 4, not 0.
  • x² is clearly rising at x = 2, so a slope of 0 should have looked wrong before any algebra.

Try to find the slip before revealing the annotations. The answer, zero, should already bother you: x squared is clearly rising at 2, so its tangent cannot be flat.

The error is in the first move. Two plus h, squared, is not 4 plus h squared. The square of a sum has a cross term, twice the product, so it is 4 plus 4h plus h squared. Dropping that cross term threw away the only piece that survives the limit.

With the cross term back, the quotient is 4 plus h, and the slope is 4. Build the habit of asking whether an answer is believable from the graph before you trust your algebra.

24. The derivative at a point

Section

Part 2

25. One limit, a new name: the derivative

Concept

The tangent-slope limit turns up in physics, economics and biology so often that it gets its own name and symbol.

Derivative of f at a (equations 3.5 and 3.6) — The limit of the difference quotient as the second point approaches a, provided the limit exists. It is written f prime of a, and computing it is called differentiation.

\[ f'(a) = \lim_{x\to a}\frac{f(x) - f(a)}{x - a} = \lim_{h\to 0}\frac{f(a+h) - f(a)}{h} \]

Nothing new to compute: the derivative at a IS the slope of the tangent at a. What is new is that the number now has a life of its own, as a rate.

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 193 — Definition, equations 3.5 and 3.6

Nothing on this slide asks you to compute anything new. The derivative of f at a is exactly the limit you have been computing as a tangent slope. What changes is its status: it now has a symbol, f prime of a, and a name, and computing it is called differentiation.

Why the new name? Because this limit shows up everywhere, not only in geometry. The book lists velocity and acceleration in physics, marginal profit in business and growth rates in biology. In every case it is the same limit of a difference quotient, read in different units.

The definition again comes in two forms, equations 3.5 and 3.6, and again ends with provided this limit exists. You now have three names for one number: the slope of the tangent, the derivative, and, in Part 3, the instantaneous rate of change.

26. Reading the definition of f′(a)

Notation

Annotate

On: \( f'(a) = \lim_{h\to 0}\frac{f(a+h) - f(a)}{h} \)

  • Read 'f prime of a'. A single number, the slope of the graph at the one input a.
  • h approaches 0 from both sides and is never equal to 0; that is what lets you cancel h.
  • Always 0 at h = 0, so the quotient is always 0/0 at first sight: the simplification is the work.
  • If the two one-sided limits differ, or the quotient blows up, f has no derivative at a (Part 4).

Step through the annotations one by one. The symbol is read f prime of a, and it is a single number, the slope of the graph at one input.

The limit sign carries a subtle promise. The variable h approaches zero from both sides but never equals zero, and that is precisely what makes cancelling h legal. The numerator is always zero at h equal to zero, which is why every computation starts as zero over zero and why the simplification is where the work lives.

The last note points ahead to Part 4. If the quotient approaches different numbers from the two sides, or grows without bound, the derivative does not exist at that point.

27. Converting between the two forms

Concept

The two forms are one limit with a change of variable. Put x equal to a plus h.

\[ x = a + h: \quad x - a = h, \qquad x \to a \iff h \to 0 \]

\[ \lim_{x\to a}\frac{f(x) - f(a)}{x - a} = \lim_{h\to 0}\frac{f(a+h) - f(a)}{h} \]

Rule of thumb: the x form suits polynomials and reciprocals whose numerator factors with x minus a; the h form suits roots and powers you can expand. Either always works in principle.

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 191 — discussion after equation 3.4

The two forms are the same limit with a change of variable, and it helps to see that once. Put x equal to a plus h. Then the run x minus a is h, the value f of x is f of a plus h, and x approaching a is the same as h approaching zero.

So which should you use? For a polynomial where you can see that f of x minus f of a factors, the x form is usually shorter. For roots, and for powers you would rather expand, the h form is usually cleaner. With reciprocals either works well.

None of this is a rule you must obey. Both forms always give the same answer; the choice is only about how much algebra stands between you and it.

28. Match each limit to the derivative it computes

Matching

Match the pairs

  • a. lim ((2 + h)³ − 8)/h, h → 0
  • b. lim (√x − √5)/(x − 5), x → 5
  • c. lim (1/(1 + h) − 1)/h, h → 0
  • d. lim (x² − 49)/(x − 7), x → 7
  • w. f(x) = x³ at a = 2
  • x. f(x) = √x at a = 5
  • y. f(x) = 1/x at a = 1
  • z. f(x) = x² at a = 7

Why: In the h form, a is whatever h is added to and f is what is done to it; in the x form, a is where x is heading and f(a) is the number subtracted on top. Reading a limit backwards like this is a skill you will need when a limit is secretly a derivative.

Make each match before checking. You are reading the definition backwards: given a limit, identify the function and the point.

In the h form, the point a is whatever h is added to, and the function is what is being done to a plus h. So two plus h, cubed, minus 8 is the cube function at 2. In the x form, a is where x is heading, and the number subtracted on top is f of a.

This backwards reading matters more than it looks. Later, some limits that seem impossible turn out to be derivatives in disguise, and recognising one lets you evaluate it instantly with a derivative rule.

29. Example 3.4: estimating f′(3) from a table

Worked example

Use a table of values of the difference quotient, equation 3.5, to estimate the derivative at 3.

\[ f(x) = x^2, \qquad f'(3) = \lim_{x\to 3}\frac{x^2 - 9}{x - 3} \]

Tabulate just below 3

Why: Each value is the slope of one secant from the left.

\[ x = 2.9, \; 2.99, \; 2.999: \quad 5.9, \; 5.99, \; 5.999 \]

Tabulate just above 3

Why: Secants from the right.

\[ x = 3.1, \; 3.01, \; 3.001: \quad 6.1, \; 6.01, \; 6.001 \]

Figure (svg): The graph of the difference quotient x squared minus 9 over x minus 3 plotted against x from 2.85 to 3.15: a straight line through the six table values with a hollow circle at 3, 6.

Every table value is the slope of one secant. Left values climb to 6, right values fall to 6; the quotient itself is undefined at x = 3, the hollow circle.

Read the common target

Why: Both columns close in on the same number.

\[ f'(3) \approx 6 \]

Check against the exact limit

Why: Factoring gives x plus 3, which tends to 6, as in Example 3.1.

\[ \frac{x^2 - 9}{x - 3} = x + 3 \to 6 \]

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, pp. 193-194 — Example 3.4

Sometimes you cannot, or would rather not, do the algebra. A table of difference quotients gives an estimate: pick inputs closer and closer to 3 from each side and watch the quotients.

The figure plots each table value as a dot. The purple dots on the left climb toward 6 and the orange ones on the right fall toward 6, both along the line x plus 3, with a hollow circle at 3 itself where the quotient is undefined. Each dot is the slope of one secant.

Always use both sides. If the left and right columns headed for different numbers, there would be no derivative, which is exactly what happens at a corner in Part 4. The check confirms the estimate with the exact factoring from Example 3.1.

30. Checkpoint 3.2: the table in the h form

Estimation

h−0.1−0.01−0.0010.0010.010.1
((3 + h)² − 9)/h5.95.995.9996.0016.016.1

Predict first

This table uses equation 3.6 for f(x) = x² at a = 3. What does it estimate f′(3) to be?

  • 5.9
  • 6
  • 6.1
  • No estimate: the two sides disagree

Correct: 6

Why: Negative h is the left side and positive h the right side. Each side misses 6 by exactly h, since the quotient simplifies to 6 + h, and both sides squeeze toward 6. The h table is the x table with every x replaced by 3 + h.

\[ \frac{(3+h)^2 - 9}{h} = 6 + h \]

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 194 — Checkpoint 3.2

Choose an estimate before revealing. The table is the h-form version of the previous slide: negative h are the left-hand points and positive h are the right-hand points.

Each entry misses 6 by exactly h, because the quotient simplifies to 6 plus h. The two sides straddle 6 and squeeze toward it, so the estimate is 6. Neither 5.9 nor 6.1 is special; they are just the values furthest from the point.

Notice how the table and the formula under it support each other. The table gives you confidence about what the limit is; the formula tells you why, and makes the estimate exact.

31. Example 3.5: a quadratic, in the x form

Worked example

Find the derivative at 2 using equation 3.5.

\[ f(x) = 3x^2 - 4x + 1, \quad f(2) = 12 - 8 + 1 = 5 \]

Apply the definition

Why: The x form at a equal to 2.

\[ f'(2) = \lim_{x\to 2}\frac{f(x) - f(2)}{x - 2} \]

Substitute

Why: Put in the formula and the value 5.

\[ = \lim_{x\to 2}\frac{(3x^2 - 4x + 1) - 5}{x - 2} \]

Simplify the numerator

Why: One minus five is minus four.

\[ = \lim_{x\to 2}\frac{3x^2 - 4x - 4}{x - 2} \]

Factor

Why: x minus 2 must be a factor, because the top is zero at 2.

\[ = \lim_{x\to 2}\frac{(x - 2)(3x + 2)}{x - 2} \]

Cancel the common factor

Why: x is not 2 inside the limit.

\[ = \lim_{x\to 2}(3x + 2) \]

Evaluate the limit

Why: Substitute 2.

\[ f'(2) = 8 \]

Figure (svg): The parabola y equals 3x squared minus 4x plus 1 and the line y equals 8x minus 11 touching it at the point 2, 5.

The derivative 8 is the slope of the orange line, the one line through (2, 5) that the parabola hugs.

Check the factoring and a nearby quotient

Why: Multiplying back gives the numerator, and x equal to 2.001 gives a quotient of 8.003.

\[ (x - 2)(3x + 2) = 3x^2 - 4x - 4, \qquad \frac{f(2.001) - 5}{0.001} = 8.003 \]

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 194 — Example 3.5

Start by computing f of 2, which is 5, because you will need to subtract it. Substituting gives three x squared minus 4x plus 1, minus 5, over x minus 2, and the constants combine to leave three x squared minus 4x minus 4.

Now a very useful fact: because the numerator is zero at x equal to 2, x minus 2 must be one of its factors. So you already know half of the factoring, and the other factor, 3x plus 2, follows from matching the leading term and the constant.

Cancel, substitute 2, and the derivative is 8. The check multiplies the factors back to confirm the numerator and shows a nearby quotient of 8.003. The figure shows the tangent line through 2, 5 with slope 8, which is y equals 8x minus 11.

32. Example 3.6: the same derivative, in the h form

Worked example

Find the derivative at 2 again, now using equation 3.6.

Apply the definition

Why: The h form at a equal to 2.

\[ f'(2) = \lim_{h\to 0}\frac{f(2+h) - f(2)}{h} \]

Substitute

Why: Replace every x by 2 plus h; subtract 5.

\[ = \lim_{h\to 0}\frac{3(2+h)^2 - 4(2+h) + 1 - 5}{h} \]

Expand

Why: Three times the square, then distribute the minus four.

\[ = \lim_{h\to 0}\frac{12 + 12h + 3h^2 - 8 - 4h + 1 - 5}{h} \]

Collect terms

Why: The constants add to zero, as they must.

\[ = \lim_{h\to 0}\frac{3h^2 + 8h}{h} \]

Factor out h

Why: Every surviving term carries an h.

\[ = \lim_{h\to 0}\frac{h(3h + 8)}{h} \]

Cancel and evaluate

Why: Let h go to 0.

\[ = \lim_{h\to 0}(3h + 8) = 8 \]

Check against Example 3.5

Why: Both forms give 8; and at h equal to 1 the unsimplified and simplified quotients agree.

\[ \frac{3(9) - 4(3) + 1 - 5}{1} = 11 = 3(1) + 8 \]

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, pp. 194-195 — Example 3.6

The same derivative through the h form, and it is a good exercise in careful expansion. Every x in the formula becomes 2 plus h. Expand three times two plus h, squared, and distribute the minus four, and write every term out before collecting.

When you collect, the constant terms add up to zero: 12 minus 8 plus 1 minus 5. They must, because of the fact from Part 1 that the numerator vanishes at h equal to zero. If your constants do not cancel, look for an arithmetic slip before going further.

What remains, three h squared plus 8h, has h in every term. Factor, cancel, and let h go to zero: 8, exactly as in the x form. The check at h equal to 1 confirms the simplification by evaluating both versions of the quotient.

33. Checkpoint 3.3: fill in the factoring

Fill the middle

\[ f(x) = x^2 + 3x + 2, \quad f(1) = 6, \quad f'(1) = \lim_{x\to 1}\frac{f(x) - 6}{x - 1} \]

Fill in the blanks

\frac45 = \frac___})}___ \;\to\; ___

Why: The numerator is x² + 3x − 4, which factors as (x − 1)(x + 4); the factor x − 1 is guaranteed because the numerator is zero at x = 1. After cancelling, x + 4 tends to 5, so f′(1) = 5.

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 195 — Checkpoint 3.3

Fill in both blanks before checking. The function value at 1 is 6, so the numerator is x squared plus 3x minus 4.

You know one factor must be x minus 1, because the numerator is zero at 1. The other factor has to multiply with it to give minus 4 as the constant, so it is x plus 4. After cancelling, x plus 4 tends to 5.

This trick of knowing one factor in advance works for every polynomial in the x form. It turns the factoring from a search into a one-step division.

34. Order the steps of a derivative from the definition

Ranking

Put in order

Put the steps of an h-form computation of f′(a) in order.

  1. Write the quotient (f(a + h) − f(a))/h
  2. Substitute the formula for f(a + h) and the number f(a)
  3. Expand, combine fractions or rationalise until h is a factor of the top
  4. Cancel the common factor h
  5. Let h → 0 by direct substitution

Why: Substitution and simplification come before the limit is taken. If you let h go to 0 before the cancellation you get 0/0 every time; after it, direct substitution finishes the job.

Put the steps in order before checking. They are the same five steps in every example in this lesson, whatever the function.

The two that people swap are the last two. You must cancel the h before you let h go to zero. Letting h go to zero first always gives zero over zero; cancelling first leaves an expression that you can evaluate by substitution.

35. A number at each point, and soon a function

Intuition

Nothing stops you from leaving a as a letter. Do Example 3.2 for a general point.

\[ \frac{(a+h)^2 - a^2}{h} = \frac{2ah + h^2}{h} = 2a + h \]

\[ \lim_{h\to 0}(2a + h) = 2a \]

So the parabola has slope 2a at every point a: 6 at 3, 4 at 2, 0 at the bottom, negative on the left. Section 3.2 makes this rule a function in its own right.

\[ a = 3: \; 2(3) = 6 \quad\checkmark \]

Here is a glimpse of the next section. Redo Example 3.2 but keep the point as a letter a instead of the number 3. The algebra is identical, and the limit comes out as 2a.

That single formula gives the slope of the parabola at every point at once. At 3 it gives 6, matching Example 3.1. At 2 it gives 4, matching the error-analysis slide. At zero, the bottom of the parabola, it gives zero, a flat tangent. On the left side, where the parabola falls, it is negative.

So the derivative is a number at each point, and those numbers together form a new function. Section 3.2 takes this seriously and studies the derivative as a function in its own right.

36. Trap: using f′(a) as the height of the point

Trap

The trap

The tangent to the parabola at 3, with the slope already found:

\[ f(x) = x^2, \quad a = 3 \]

\[ f'(3) = 6 \;\Longrightarrow\; (3, 6) \]

\[ y - 6 = 6(x - 3) \]

Wrong point: the curve is not at height 6.

The fix

The derivative is a slope, not a height. The point on the graph comes from the function itself, f of 3.

\[ (3, f(3)) = (3, 9) \]

\[ y = 6x - 9 \]

This slip happens when two numbers are in play and one gets used for both jobs. The derivative, 6, is a slope. It says how steep the line is. It says nothing about where the line is.

The point on the tangent line is the point on the graph, and its height is f of a, found by substituting into the original function. For the parabola at 3 that is 9, not 6. Using 6 produces a line with the right slope that never touches the curve.

A quick check catches this every time: substitute a into your tangent line and see whether it returns f of a.

37. Velocity and rates of change

Section

Part 3

38. Average velocity is a secant slope

Concept

If s of t is the position of an object moving along a line, the average velocity over a time interval is the change in position over the time elapsed, a difference quotient.

\[ v_{\text{ave}} = \frac{s(t) - s(a)}{t - a} \]

Figure (svg): A rising, flattening position curve s of t. A green secant line joins the points at times a and t; an orange tangent line touches the curve at a and is steeper than the secant.

Average velocity over the interval is the green slope; the speedometer reading at time a is the orange slope.

On the graph of position against time that ratio is the slope of a secant. Shrink the interval and it becomes the slope of the tangent: the instantaneous velocity.

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 195 — equation 3.7 and Figure 3.8

Now the same mathematics gets a physical reading. If s of t is where an object is at time t, then the change in position divided by the time elapsed is its average velocity over that interval. That is a difference quotient, equation 3.7.

In the figure the green line is a secant from time a to time t, and its slope is that average velocity. The orange line is the tangent at a. As the interval shrinks, the green line swings onto the orange one.

So the tangent slope must be the velocity at the single instant a, the number a speedometer shows. That is the instantaneous velocity, and the next slide gives it a formula.

39. Reading instantaneous velocity

Notation

Annotate

On: \( v(a) = s'(a) = \lim_{t\to a}\frac{s(t) - s(a)}{t - a} \)

  • The instantaneous velocity at time a: what a speedometer (with a sign for direction) reads at that instant.
  • Velocity is nothing but the derivative of position. No new mathematics, a new reading.
  • An average velocity over [a, t] or [t, a]: displacement over elapsed time.
  • Units of s over units of t: metres per second, feet per second. A negative value means moving in the negative direction.

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 195 — equation 3.8

Read the annotations in order. The instantaneous velocity at a is written v of a, and the definition says it is simply the derivative of the position function at a. There is no new mathematics here, only a new interpretation.

The quotient inside the limit is an average velocity. The limit shrinks the interval to nothing. The book writes the interval as a to t when t is later than a and t to a when t is earlier; the formula handles both, just as h could be positive or negative.

Units come along automatically: position units over time units. Velocity also has a sign. A negative velocity means the object is moving in the negative direction, downward for a height, backward along a road.

40. Example 3.7: the weight on a spring

Worked example

The weight's position relative to its rest line is sine t. Estimate its velocity at t equal to 0 from a table, then confirm with equation 3.5.

\[ s(t) = \sin t, \qquad v(0) = \lim_{t\to 0}\frac{\sin t - \sin 0}{t - 0} = \lim_{t\to 0}\frac{\sin t}{t} \]

Tabulate average velocities

Why: The quotient is the same for t and minus t, so both sides agree at each size.

\[ t = \pm 0.1: \; 0.998334, \quad t = \pm 0.01: \; 0.999983, \quad t = \pm 0.001: \; 0.9999998 \]

Estimate

Why: The averages close in on one from both sides.

\[ v(0) \approx 1 \]

Figure (svg): The curve s equals sine t for t from minus 3.2 to 3.2, the tangent line s equals t through the origin, and two dashed secants from the origin to t equals 1 and t equals minus 1, each with slope 0.84.

The weight passes its rest line at speed 1. Secants over a whole unit of time undershoot, at 0.84; shorter intervals close the gap.

Check with the limit you already know

Why: Section 2.3 proved this limit equals one with the squeeze theorem.

\[ v(0) = s'(0) = \lim_{t\to 0}\frac{\sin t}{t} = 1 \]

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, pp. 195-196 — Example 3.7 and Table 3.1

The book follows the recommended routine: estimate from a table, then confirm with the definition. The average velocities over shorter and shorter intervals around time zero are 0.998, then 0.99998, then 0.9999998, the same on both sides. They are heading to 1.

To confirm, substitute into equation 3.5. Sine of zero is zero, so the quotient is sine t over t, and you proved in Section 2.3 with the squeeze theorem that this limit is exactly 1. That is not algebra you could do by cancelling; it needed the earlier theorem.

The figure shows the weight's position curve with its tangent at the origin, the line of slope 1. The dashed secants over a whole unit of time have slope about 0.84, a clear undershoot, and shorter secants close the gap. The weight is passing through its rest position at 1 unit per second, moving upward.

41. Checkpoint 3.4: the dropped rock

Worked example

A rock dropped from 64 feet has height s of t feet after t seconds. Find its instantaneous velocity one second after it is dropped.

\[ s(t) = -16t^2 + 64, \quad 0 \le t \le 2, \qquad s(1) = 48 \]

Apply equation 3.5

Why: The x form, with time as the variable.

\[ v(1) = \lim_{t\to 1}\frac{s(t) - s(1)}{t - 1} \]

Substitute

Why: Subtract the height 48.

\[ = \lim_{t\to 1}\frac{-16t^2 + 64 - 48}{t - 1} \]

Simplify the numerator

Why: 64 minus 48 is 16.

\[ = \lim_{t\to 1}\frac{-16t^2 + 16}{t - 1} \]

Factor

Why: Take out minus 16, then a difference of squares.

\[ = \lim_{t\to 1}\frac{-16(t - 1)(t + 1)}{t - 1} \]

Cancel and evaluate

Why: Let t go to 1.

\[ = \lim_{t\to 1}-16(t + 1) = -32 \]

Figure (svg): Height of the rock, s equals minus 16 t squared plus 64, falling from 64 feet at t equals 0 to the ground at t equals 2, with the tangent line at t equals 1 through the point 1, 48 with slope minus 32.

One second in, the rock is 48 feet up and losing height at 32 feet per second: the negative slope is the downward motion.

Check with averages on both sides

Why: Averages over a hundredth of a second either side bracket minus 32 feet per second.

\[ \frac{s(1.01) - s(1)}{0.01} = -32.16, \qquad \frac{s(0.99) - s(1)}{-0.01} = -31.84 \]

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 196 — Checkpoint 3.4

Set up the derivative at 1 with the x form. The height at 1 second is 48 feet, so the numerator is minus 16 t squared plus 64 minus 48, which is minus 16 t squared plus 16.

Take out the minus 16 first; then t squared minus 1 is a difference of squares. Cancel t minus 1 and substitute: minus 16 times 2 is minus 32. The velocity is minus 32 feet per second, and the minus sign means the height is decreasing: the rock is falling.

The check brackets the answer with average velocities over a hundredth of a second on each side, minus 32.16 and minus 31.84. The figure shows the tangent at 1 second, a line dropping 32 feet for each second.

42. Average or instantaneous?

Sorting

Sort into buckets

Sort each quantity: is it an average rate over an interval, or an instantaneous rate at one moment?

average: a secant slope
a 120-mile trip took 2 hours: 60 mph; (s(5) − s(2))/3; slope of the secant from t = 1 to t = 1.1
instantaneous: a tangent slope
the speedometer reading at 3:15 pm; lim (s(t) − s(2))/(t − 2) as t → 2; slope of the tangent at t = 1
avg
Each uses two separate times and divides a change by the gap between them. However short the gap (even 0.1 s), it is still an average.
inst
Each is a limit as the gap shrinks to nothing: one moment, one number. A speedometer approximates exactly this limit.

Sort each item before checking. The dividing question is simple: does the quantity involve two separate times, or one moment?

Anything that divides a change by a time gap is an average, however short the gap. The secant over a tenth of a second is still an average; it is a good approximation to the instantaneous value, but not the same thing.

Instantaneous means a limit: the gap shrinks to nothing. The speedometer reading is the physical version of that limit, and the slope of the tangent is its geometric version.

43. Instantaneous rate of change, and its units

Concept

Instantaneous rate of change — The instantaneous rate of change of a function f at a value a is its derivative f prime of a.

Velocity is one example. The same limit measures how fast any quantity changes with respect to any other, and its units are always output units per input unit.

functioninputf′(a) meansunits of f′(a)
position s(t), feettime, secondsvelocityft/s
velocity v(t), ft/stime, secondsaccelerationft/s per s
temperature T(t), °Ftime, hoursheating or cooling rate°F per hour
profit P(x), dollarsunits mademarginal profitdollars per unit

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 196 — Definition of instantaneous rate of change

This definition generalises velocity. For any function, the derivative at a measures how fast the output is changing relative to the input at that instant. That is its meaning in every application you will meet.

The units follow from the difference quotient: a difference of outputs over a difference of inputs, so output units per input unit. Look at the table. A position in feet against time in seconds gives feet per second. A velocity against time gives acceleration, feet per second per second. A temperature against hours gives degrees per hour. A profit in dollars against units made gives dollars per unit, called marginal profit.

Getting the units right is the fastest way to interpret a derivative correctly, and a good check on whether you have set up a problem sensibly.

44. Example 3.8: the Venom GT's acceleration

Worked example

The car reached 60 mph in 3.05 s, 100 in 5.88 s, 200 in 14.51 s and 229.9 in 19.96 s. Is its acceleration increasing, decreasing or constant as it nears 229.9 mph?

Convert miles per hour to feet per second

Why: 5280 feet per mile, 3600 seconds per hour.

\[ 1 \text{ mph} = \frac{5280 \text{ ft}}{3600 \text{ s}} = \frac{22}{15} \text{ ft/s} \]

Convert each speed

Why: Multiply by 22 over 15.

\[ 60 \to 88, \quad 100 \to 146.67, \quad 200 \to 293.33, \quad 229.9 \to 337.19 \]

Average acceleration up to the last reading

Why: A difference quotient of velocity on an interval ending at 19.96 seconds.

\[ a_{\text{ave}} = \frac{v(19.96) - v(t)}{19.96 - t} = \frac{337.19 - v(t)}{19.96 - t} \]

Intervals from 0 and 3.05 s

Why: Divide each gain in speed by the time it took.

\[ \frac{337.19 - 0}{19.96} = 16.89, \qquad \frac{337.19 - 88}{16.91} = 14.74 \]

Intervals from 5.88 and 14.51 s

Why: The same, closer to the end.

\[ \frac{337.19 - 146.67}{14.08} = 13.53, \qquad \frac{337.19 - 293.33}{5.45} = 8.05 \]

Figure (svg): The five data points of velocity against time for the Hennessey Venom GT, from 0 at t equals 0 to 337.19 feet per second at t equals 19.96, with four dashed secant lines from each earlier point to the last one; the secants get flatter as the starting time increases.

Each dashed line's slope is an average acceleration. They flatten as the start time moves toward 19.96 s: the car is gaining speed ever more slowly.

Check the trend

Why: The averages fall steadily as the interval closes in on 19.96 seconds, so the acceleration there is decreasing.

\[ 16.89 > 14.74 > 13.53 > 8.05 \;\text{ ft/s}^2 \]

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 197 — Example 3.8, Tables 3.2 and 3.3

This example uses real test data rather than a formula. The first job is units: miles per hour must become feet per second, and one mile per hour is 22 fifteenths of a foot per second. Multiplying gives the four speeds, 88, 146.67, 293.33 and 337.19.

Acceleration is the rate of change of velocity, so the average accelerations are difference quotients of the velocity data. The book fixes the right end of each interval at 19.96 seconds and moves the left end closer: 16.89, 14.74, 13.53, and 8.05 feet per second per second.

Look at the figure. Each dashed line's slope is one of those averages, and the lines get flatter as their starting point approaches the last data point. The car is still speeding up, but ever more slowly. With only data and no formula, a trend like this is the honest conclusion you can draw.

45. Example 3.9: how fast the house cools at midnight

Worked example

The temperature t hours after 9 p.m. is T of t degrees Fahrenheit. Find the instantaneous rate of change at midnight.

\[ T(t) = 0.4t^2 - 4t + 70, \quad 0 \le t \le 10; \qquad T(3) = 3.6 - 12 + 70 = 61.6 \]

Set up the derivative at 3

Why: Midnight is 3 hours after 9 p.m.

\[ T'(3) = \lim_{t\to 3}\frac{T(t) - T(3)}{t - 3} \]

Substitute

Why: Subtract 61.6.

\[ = \lim_{t\to 3}\frac{0.4t^2 - 4t + 70 - 61.6}{t - 3} \]

Simplify

Why: 70 minus 61.6 is 8.4.

\[ = \lim_{t\to 3}\frac{0.4t^2 - 4t + 8.4}{t - 3} \]

Factor

Why: Take out 0.4; then t squared minus 10t plus 21 factors.

\[ = \lim_{t\to 3}\frac{0.4(t - 3)(t - 7)}{t - 3} \]

Cancel and evaluate

Why: Let t go to 3.

\[ = \lim_{t\to 3}0.4(t - 7) = 0.4(-4) = -1.6 \]

Figure (svg): The temperature curve T of t equals 0.4 t squared minus 4t plus 70 from 9 p.m. to 7 a.m., falling from 70 degrees to a low of 60 at 2 a.m. and back to 70, with the tangent line at midnight, slope minus 1.6.

At midnight the house is still cooling, 1.6 degrees per hour, and the cooling slows until it stops at the 2 a.m. low.

Check with a short interval and the units

Why: The average over the next 0.001 hours is minus 1.5996, and the units are degrees per hour.

\[ \frac{T(3.001) - T(3)}{0.001} = -1.5996 \approx -1.6 {}^\circ\text{F per hour} \]

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 198 — Example 3.9

First translate the question. Time is measured from 9 p.m., so midnight is t equal to 3, and you want the derivative at 3. The temperature then is 61.6 degrees.

After subtracting 61.6 the numerator is 0.4 t squared minus 4t plus 8.4. Take out the 0.4 to make the factoring comfortable: t squared minus 10t plus 21 is t minus 3 times t minus 7. Cancel and substitute: 0.4 times minus 4, which is minus 1.6.

The figure puts the answer in context. At midnight the curve is still falling, at 1.6 degrees per hour. The fall slows until the low of 60 degrees at 2 a.m., where the tangent is flat, and after that the house warms back to 70 by 7 a.m.

46. Example 3.10: should the toy company make more?

Worked example

x systems sell at p dollars each and cost C of x dollars to make. Find the rate of change of profit at 10,000 systems.

\[ p = -0.01x + 400, \qquad C(x) = 100x + 10{,}000 \]

Build revenue

Why: Number sold times price.

\[ R(x) = xp = -0.01x^2 + 400x \]

Build profit

Why: Revenue minus cost.

\[ P(x) = -0.01x^2 + 300x - 10{,}000 \]

Evaluate profit at 10,000

Why: Minus one million, plus three million, minus ten thousand.

\[ P(10{,}000) = 1{,}990{,}000 \]

Set up the derivative

Why: The x form; subtract the profit value.

\[ P'(10{,}000) = \lim_{x\to 10{,}000}\frac{-0.01x^2 + 300x - 2{,}000{,}000}{x - 10{,}000} \]

Factor

Why: Minus 0.01 times x minus 10,000 times x minus 20,000.

\[ = \lim_{x\to 10{,}000}\frac{-0.01(x - 10{,}000)(x - 20{,}000)}{x - 10{,}000} \]

Cancel and evaluate

Why: Let x go to 10,000.

\[ = -0.01(10{,}000 - 20{,}000) = 100 \]

Figure (svg): The profit P of x for x from 0 to 30 thousand systems, a downward parabola peaking near 2.24 million dollars at 15 thousand, with the tangent at 10 thousand systems rising with slope 100 dollars per system.

At 10,000 systems profit is still climbing, so making more pays. The peak, where the slope is zero, is at 15,000.

Check with the next system

Why: Making system 10,001 adds almost exactly 100 dollars, and profit is positive, so increase production.

\[ P(10{,}001) - P(10{,}000) = 99.99 \approx 100 \text{ dollars} \]

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 198 — Example 3.10

Business problems need a little setup before any calculus. Revenue is the number sold times the price, and the price depends on how many are made, so revenue is a quadratic. Profit is revenue minus cost, another quadratic.

The derivative at 10,000 follows the usual routine. Subtract the profit value, 1,990,000; the numerator must then have x minus 10,000 as a factor, and the other factor is x minus 20,000, with minus 0.01 in front. Cancel and substitute: 100 dollars per system.

Interpretation is the point. A positive rate means each extra system adds about 100 dollars of profit, and the check confirms that making the 10,001st system adds 99.99 dollars. So the company should increase production. The figure shows why the answer would change eventually: profit peaks at 15,000 systems.

47. Checkpoint 3.5: the price of a scone

Prediction

\[ P(s) = -20s^2 + 150s - 10, \qquad s = 3.25 \text{ dollars} \]

Predict first

Daily profit on scones is P(s) at price s. The shop charges $3.25. Work out P′(3.25) and decide: should the shop raise or lower its price?

  • Raise the price
  • Lower the price
  • Keep it: P′(3.25) = 0
  • Cannot tell without P(3.25)

Correct: Raise the price

Why: P(3.25) = 266.25, and the numerator −20s² + 150s − 276.25 factors as −20(s − 3.25)(s − 4.25). After cancelling, the limit is −20(3.25 − 4.25) = 20 dollars per dollar of price. Profit rises as the price rises, so a small increase pays (up to the peak at $3.75).

\[ P'(3.25) = \lim_{s\to 3.25}-20(s - 4.25) = 20 \]

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 198 — Checkpoint 3.5

Work out the derivative before revealing. The profit at 3.25 dollars is 266.25, and after subtracting it the numerator factors with s minus 3.25 as one factor and s minus 4.25 as the other, times minus 20.

The derivative is minus 20 times 3.25 minus 4.25, which is 20. Its units are dollars of daily profit per dollar of price. A positive value says that raising the price a little raises profit, so the shop should consider a small increase.

The answer that tempts people is that you need the profit value itself. You do need it to set up the quotient, but it does not decide the question; the sign of the derivative does.

48. Exercise 45: closing in on the race car

Estimation

\[ s(t) = 8t^2 - \frac{1}{16}t^3 \text{ ft}, \qquad s(4) = 124 \]

interval[4, 4.1][4, 4.01][4, 4.001][4, 4.0001]
average velocity (ft/s)61.724461.072561.007261.0007

Predict first

From the table alone, what is your best estimate of the car's instantaneous velocity at t = 4 s?

  • 61.7 ft/s
  • 61 ft/s
  • 62 ft/s
  • 124 ft/s

Correct: 61 ft/s

Why: Each tenfold shrink of the interval cuts the excess over 61 by about ten: 0.72, 0.07, 0.007, 0.0007. The averages are converging to 61. The exact limit agrees: the quotient simplifies to 61 + 7.25h − h²/16, which tends to 61. And 124 ft is the position, not the velocity.

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 201 — Exercise 45

Make your estimate from the table before revealing. The average velocities over intervals starting at 4 seconds are 61.72, then 61.07, then 61.007, then 61.0007.

Look at how the excess over 61 behaves: each time the interval shrinks by a factor of ten, the excess shrinks by roughly ten too. That steady pattern is strong evidence that the limit is 61 feet per second.

The exact computation agrees, and it also explains the pattern: the quotient simplifies to 61 plus 7.25h minus a tiny h squared term, so the excess is about 7.25 times h. The distractor 124 is the car's position at 4 seconds, a reminder not to confuse where something is with how fast it is going.

49. Exercise 51: gallons per mile

Real world

Discussion prompt

A car gets 30 miles per gallon, and N(x) is the number of gallons used to drive x miles. Find a formula for N(x), then N(100) and N′(100), and say what each means physically.

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 202 — Exercise 51

Write your answer before revealing. At 30 miles per gallon, each mile uses one thirtieth of a gallon, so the gallons used for x miles is x over 30.

N of 100 is a quantity: about 3.33 gallons for a 100-mile drive. N prime of 100 is a rate: the extra fuel used per extra mile at the 100-mile mark, one thirtieth of a gallon per mile. The limit is quick because the numerator is exactly h over 30.

Notice that the rate is the reciprocal of fuel economy, and that it is the same at every mileage because N is a straight line. For a line, the derivative is just the slope everywhere. Real fuel use is not quite linear, which is where derivatives start to earn their keep.

50. Where a derivative fails

Section

Part 4

51. Differentiable at a point

Concept

The definition always ends with the words provided this limit exists. When it does, f is differentiable at a. When it does not, the graph has no tangent line with a slope there.

\[ f'(a) \text{ exists} \iff \lim_{h\to 0^-}\frac{f(a+h) - f(a)}{h} = \lim_{h\to 0^+}\frac{f(a+h) - f(a)}{h} \text{ (finite)} \]

Three things can go wrong: the one-sided limits disagree (a corner), the quotient grows without bound (a vertical tangent), or f is not even defined or continuous at a (a gap or a jump).

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 201 — Exercises 41 to 44

Every definition in this lesson ended with provided this limit exists. This part is about what happens when it does not. A function whose derivative exists at a is called differentiable at a.

Because the derivative is a two-sided limit, it exists exactly when the two one-sided limits of the difference quotient exist, are finite, and agree. That gives a checklist of failures: the one-sided limits disagree, the quotient grows without bound, or the function is not defined or not continuous at a, so the quotient has nothing sensible to compare.

The book's exercises 41 to 44 contain one example of each kind, and you will work two of them in detail.

52. Three ways to have no derivative

Picture it

Figure (svg): Three small graphs. Left: y equals the absolute value of x, a V with a sharp corner at the origin. Middle: y equals the cube root of x, which stands vertical at the origin. Right: y equals the absolute value of x over x, a step from minus 1 to 1 with hollow circles at x equals 0.

A corner has two different one-sided slopes; a vertical tangent has an infinite slope; a jump has no point to draw a tangent at.

The absolute value turns a corner at 0. The cube root stands vertical at 0. The step function jumps at 0 and is not even defined there. In each case, try to draw one sensible tangent line through the origin and you will see why the limit fails.

Look at each panel and try to draw one tangent line through the origin. On the left, the absolute value has a sharp corner. A line leaning one way fits the right arm, a line leaning the other way fits the left arm, and no single line fits both.

In the middle, the cube root becomes vertical at the origin. There is a tangent line there, the vertical axis, but a vertical line has no slope, so there is no derivative.

On the right, the absolute value of x over x jumps from minus 1 to 1 and is not defined at zero at all. With no point on the graph above zero, there is nothing to draw a tangent through. That one is Exercise 44.

53. The corner of the absolute value

Concept

Apply the definition to the absolute value at 0. The value there is 0, so the quotient is the absolute value of h over h.

\[ \frac{f(0+h) - f(0)}{h} = \frac{|h|}{h} \]

\[ h > 0: \; \frac{|h|}{h} = \frac{h}{h} = 1, \qquad h < 0: \; \frac{|h|}{h} = \frac{-h}{h} = -1 \]

\[ \lim_{h\to 0^+} = 1 \ne -1 = \lim_{h\to 0^-} \;\Longrightarrow\; f'(0) \text{ does not exist} \]

The right-hand secants all have slope 1 and the left-hand ones slope minus 1. They never agree, however small h gets.

Apply the definition exactly as before. At zero the absolute value is zero, so the quotient is the absolute value of h divided by h.

Now split into cases, because the absolute value behaves differently on each side. For positive h, the absolute value of h is h, and the quotient is 1. For negative h, the absolute value is minus h, and the quotient is minus 1. These are constant: they do not creep toward each other as h shrinks.

So the right-hand limit is 1, the left-hand limit is minus 1, and the two-sided limit does not exist. That matches the picture: every secant on the right has slope 1, every secant on the left has slope minus 1.

54. Exercise 43: a corner where two pieces meet

Worked example

Use the limit definition to show that the derivative does not exist at x equal to 1.

\[ f(x) = \begin{cases} 1, & x < 1 \\ x, & x \ge 1 \end{cases} \qquad f(1) = 1 \]

Right-hand quotient

Why: For h positive, 1 plus h is at least 1, so use the formula x.

\[ \lim_{h\to 0^+}\frac{(1 + h) - 1}{h} = \lim_{h\to 0^+}\frac{h}{h} = 1 \]

Left-hand quotient

Why: For h negative, 1 plus h is below 1, so the value is 1.

\[ \lim_{h\to 0^-}\frac{1 - 1}{h} = \lim_{h\to 0^-} 0 = 0 \]

Compare the two sides

Why: The one-sided limits differ, so the two-sided limit does not exist.

\[ 1 \ne 0 \;\Longrightarrow\; f'(1) \text{ does not exist} \]

Figure (svg): The graph of f equal to 1 for x less than 1 and equal to x for x at least 1: a horizontal ray meeting a slanted ray at the point 1, 1. A dashed purple extension of the flat piece and a dashed orange extension of the slanted piece show the two different one-sided slopes.

The two pieces meet, so f is continuous at 1, but they arrive with different slopes. No single line fits both.

Check that f is continuous there anyway

Why: Both pieces approach 1 at x equal to 1, so the failure is a corner, not a break; numerically h equal to plus and minus 0.01 give slopes 1 and 0.

\[ \lim_{x\to 1^-} 1 = 1 = \lim_{x\to 1^+} x = f(1) \]

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 201 — Exercise 43

A piecewise function tells you which formula to use by the size of the input, so treat the two sides separately. For positive h, 1 plus h is at least 1, so you use the formula x. For negative h, 1 plus h is less than 1, so the value is 1.

The right-hand quotient is h over h, which is 1. The left-hand quotient is zero over h, which is 0. Different one-sided limits mean no derivative.

The check is important because it tells you what kind of failure this is. Both pieces reach height 1 at x equal to 1, so the function is continuous there. The graph has no break, just a change of direction, as the figure shows with the two dashed extensions arriving at different slopes.

55. Exercise 41: the cube root at 0

Worked example

Use the limit definition to show that the derivative does not exist at x equal to 0.

\[ f(x) = x^{1/3}, \qquad f(0) = 0 \]

Write the quotient

Why: The h form at a equal to 0.

\[ \frac{f(0 + h) - f(0)}{h} = \frac{h^{1/3}}{h} \]

Simplify the power

Why: Subtract exponents: one third minus one.

\[ \frac{h^{1/3}}{h} = h^{-2/3} = \frac{1}{h^{2/3}} \]

Take the limit

Why: The denominator is a square of a cube root, positive on both sides, and shrinks to zero.

\[ \lim_{h\to 0}\frac{1}{h^{2/3}} = +\infty \]

Conclude

Why: An infinite limit is not a number, so there is no derivative.

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

Figure (svg): The curve y equals the cube root of x near the origin with three dashed secants from the origin to x equals 1, 0.3 and 0.05, of slopes 1, 2.23 and 7.37, getting steeper toward the vertical axis.

As the second point slides to the origin the secants stand up without limit. The tangent is the vertical axis, and a vertical line has no slope.

Check with numbers on both sides

Why: h equal to 0.001 and minus 0.001 both give a secant slope of 100, and the slopes keep growing.

\[ \frac{(0.001)^{1/3}}{0.001} = \frac{0.1}{0.001} = 100, \qquad \frac{(-0.001)^{1/3}}{-0.001} = \frac{-0.1}{-0.001} = 100 \]

OpenStax Calculus Volume 1, §3.1 Defining the Derivative §3.1, p. 201 — Exercise 41

This failure is different. The quotient is h to the one third over h, and subtracting exponents gives h to the minus two thirds, which is one over h to the two thirds.

Because h to the two thirds is the square of a cube root, it is positive for both positive and negative h, and it shrinks to zero. One over a shrinking positive number grows without bound. The limit is infinite, and an infinite limit is not a derivative.

The figure shows why geometrically: the secants from the origin stand up more and more steeply, slopes 1, 2.23, 7.37, and the check shows a slope of 100 at a thousandth on each side. The curve has a vertical tangent at the origin.

56. Continuous, yet no derivative

Counterexample

Discussion prompt

Someone claims: if a function is continuous at a, it has a derivative at a. Give a counterexample from this lesson and name exactly which part of the definition fails.

Write your counterexample before revealing. This is the most important logical fact of the section, so it is worth being able to produce the example on demand.

The absolute value is continuous at zero: its limit there is zero, which is its value. But its difference quotient approaches different numbers from the two sides, so the limit does not exist. The cube root is a second counterexample, continuous at zero with an infinite quotient.

So continuity is not enough for a derivative. The reverse direction is true, as Section 3.2 proves: if a function has a derivative at a point, it must be continuous there.

57. Trap: continuous, so differentiable

Trap

The trap

A tempting shortcut:

\[ \lim_{x\to 0}|x| = 0 = |0| \]

\[ \Longrightarrow\; |x| \text{ has a derivative at } 0? \]

Wrong. Continuity says the graph has no break, nothing about its slope.

The fix

To have a derivative, the slopes must settle to one finite number from both sides. At the corner of the V they do not.

\[ \frac{|h|}{h} = 1 \text{ or } -1 \]

The true direction (Section 3.2): differentiable implies continuous, not the other way round.

The shortcut on the left feels natural because smooth-looking functions are both continuous and differentiable. But the two words ask different questions. Continuity asks whether the graph has a break. Differentiability asks whether the graph has a single well-defined slope.

The corner of the V has no break and two slopes. That is enough to kill the derivative. Whenever you are asked whether a derivative exists at a point, compute the one-sided quotients; do not argue from continuity.

58. Differentiable at zero?

Discrimination

Sort into buckets

Decide for each function whether f′(0) exists.

f′(0) exists
f(x) = x²; f(x) = x|x|; f(x) = sin x
f′(0) does not exist
f(x) = |x|; f(x) = x^(1/3); f(x) = x^(2/3); f(x) = |x|/x
yes
x²: quotient h → 0. x|x|: quotient |h| → 0 from both sides, so f′(0) = 0. sin x: quotient (sin h)/h → 1, as in Example 3.7.
no
|x|: slopes ±1 (corner). x^(1/3): quotient → +∞ (vertical tangent). x^(2/3): quotient h^(−1/3) → +∞ from the right and −∞ from the left (a cusp). |x|/x: not defined at 0, so f(0) does not exist (Exercise 44).

Sort each function before checking, and for each one, think about the difference quotient at zero rather than guessing from the formula.

Two items deserve a second look. The function x times the absolute value of x has an absolute value in it, but its quotient is the absolute value of h, which goes to zero from both sides, so the derivative exists and is zero. Absolute value signs do not automatically mean a corner.

The two-thirds power is a cusp: the quotient heads to plus infinity on the right and minus infinity on the left. And the absolute value of x over x is not even defined at zero, so its derivative cannot exist there.

59. Putting it together

Section

Part 5

60. Pattern: from a function to its tangent line

Pattern

Figure (svg): A flow diagram: write the difference quotient, substitute the function, simplify until h is a visible factor, cancel h, let h go to 0 to get f prime of a. A branch from the start computes f of a, and both feed the tangent line y equals f of a plus f prime of a times x minus a.

Two separate numbers, computed separately: the limit gives the slope, plain substitution gives the point.
  1. Pick a form. The x form when f(x) − f(a) factors; the h form for roots and expandable powers.
  2. Substitute, then simplify until h (or x − a) is a visible factor: expand, combine fractions, or multiply by a conjugate.
  3. Cancel, then take the limit by direct substitution. Never substitute before cancelling.
  4. Check the limit exists from both sides; a corner, a vertical tangent or a gap means no derivative.
  5. Tangent line: point (a, f(a)) from the function, slope f′(a) from the limit: y = f(a) + f′(a)(x − a).
  6. Interpret with units: output units per input unit.

This is the whole lesson as a procedure. The flow diagram shows the two separate strands: the top row computes the slope through a limit, and the separate branch on the left computes the height by plain substitution. They meet only in the final tangent line.

Choose your form by the algebra it needs, simplify until the factor h or x minus a is visible, cancel, and only then take the limit. Before trusting a derivative at an unusual point, check that the one-sided limits agree and are finite.

Finally, attach units. Output units per input unit turns a bare number into a statement about the world: a velocity, a cooling rate, a marginal profit.

61. One number, four readings

Comparison

Comparison matrix

settingthe derivative is…units
graph of y = f(x)slope of the tangent liney-units per x-unit
position s(t), feet, t in secondsvelocityft/s
temperature T(t), °F, t in hoursrate of heating or cooling°F per hour
profit P(x), dollars, x units mademarginal profitdollars per unit
fuel N(x), gallons, x in milesrate of fuel usegallons per mile

Fill in the blanks before checking. Every row is the same limit of a difference quotient; only the story and the units change.

The units column is the one to practise. Divide the output units by the input units: feet by seconds, degrees by hours, gallons by miles. If you can state the units, you can almost always say in words what the derivative means.

62. Check: the definition

Check

Check your understanding

For f(x) = x², which expression equals f′(5)?

  • A. lim ((5 + h)² − 25)/h as h → 0 (correct)
  • B. lim (25 + h² − 25)/h as h → 0
  • C. ((5 + h)² − 25)/h evaluated at h = 0
  • D. lim (x² − 25)/(x + 5) as x → 5

Answer: A

Why: Equation 3.6 with a = 5: the quotient (f(5 + h) − f(5))/h, and a LIMIT as h → 0. It simplifies to 10 + h, so f′(5) = 10.

Why B tempts people
(5 + h)² was squared term by term; the missing 10h is exactly what carries the answer.
Why C tempts people
At h = 0 the quotient is 0/0, undefined. The derivative is the limit, not the value at h = 0.
Why D tempts people
The x form divides by x − 5, the run; x + 5 gives a limit of 0, which is not a slope of anything here.

Look for the one expression that is both the right quotient and a limit. The h form at 5 subtracts f of 5, which is 25, and divides by h, and takes the limit as h goes to zero.

Each wrong option reflects a real slip from this lesson: squaring a sum term by term, evaluating at h equal to zero instead of taking a limit, and dividing by the wrong expression in the x form.

63. Check: writing the tangent line

Check

Check your understanding

A function has f(2) = 7 and f′(2) = −3. What is the tangent line at x = 2?

  • A. y = 7x − 3
  • B. y = −3x + 13 (correct)
  • C. y = −3x + 7
  • D. y = 7(x − 2) − 3

Answer: B

Why: The point is (2, f(2)) = (2, 7) and the slope is f′(2) = −3, so y − 7 = −3(x − 2), which gives y = −3x + 13. Check: at x = 2 it gives 7.

Why A tempts people
The slope and the height have been swapped: 7 is the height, −3 the slope.
Why C tempts people
This line has the right slope but passes through (0, 7), not (2, 7).
Why D tempts people
Again 7 is used as the slope and −3 as the height; at x = 2 this line gives −3.

Separate the two numbers first. The value 7 tells you the point, 2, 7. The derivative minus 3 tells you the slope. Point-slope form then gives y minus 7 equals minus 3 times x minus 2.

The fastest check is to substitute x equal to 2 into each option and see which gives 7 with slope minus 3. Two of the wrong options swap the roles of the height and the slope, the trap from Part 2.

64. Check: reading a derivative

Check

Check your understanding

A cart's position is s(t) feet after t seconds, and s′(4) = 12. What does this mean?

  • A. At t = 4 s the cart is moving at 12 ft/s in the positive direction (correct)
  • B. At t = 4 s the cart is 12 ft from its start
  • C. Over the first 4 s the cart averaged 12 ft/s
  • D. The cart travels 12 ft in the 4th second

Answer: A

Why: The derivative of position is the instantaneous velocity, with units feet per second. A positive value means s is increasing at that instant.

Why B tempts people
That would be s(4) = 12: a position, not a rate.
Why C tempts people
(s(4) − s(0))/4 is an average velocity, a secant slope, not the derivative.
Why D tempts people
s(4) − s(3) is a change over a whole second, again an average, which need not equal the instantaneous rate.

The derivative of position is velocity, so s prime of 4 equal to 12 is a speed with a direction, at one instant: 12 feet per second in the positive direction.

The distractors mix up a position with a rate, and an instantaneous rate with an average over an interval. The units are the giveaway: a position is in feet, and only a rate is in feet per second.

65. Explain the zero over zero

Explain it to yourself

Discussion prompt

Every derivative from the definition starts as zero over zero. In two or three sentences, explain why that always happens, and why it does not mean the derivative fails to exist.

Write your explanation before revealing. The aim is to make sure the zero over zero never frightens you again.

At h equal to zero both halves of the numerator are f of a, so the numerator is zero, and the denominator is zero too. That happens for every function at every point, so it tells you nothing about whether a derivative exists. The limit ignores h equal to zero entirely and looks at small nonzero h, where the factor h can be cancelled.

The derivative fails only when that limit fails: when the two sides disagree, as at a corner, or when the quotient blows up, as at a vertical tangent.

66. Exit ticket

Exit ticket

\[ f(x) = 2x^2 - x, \qquad a = 1 \]

Discussion prompt

Find f′(1) from the definition (either form) and write the equation of the tangent line at x = 1.

This combines the two main skills in one problem: a derivative from the definition and a tangent line. Use either form.

In the h form, expand two times one plus h, squared, and subtract one plus h and the value f of 1, which is 1. The constants cancel, leaving 3h plus 2h squared over h, which simplifies to 3 plus 2h and tends to 3. The point is 1, 1, so the tangent line is y equals 3x minus 2.

Check it the same two ways as the examples: the line returns 1 at x equal to 1, and a nearby secant has slope 3.002.

67. Recap

Recap

ideaformulameaning
difference quotient(f(x) − f(a))/(x − a) = (f(a + h) − f(a))/hslope of a secant; an average rate
derivative at af′(a) = lim of the quotient as x → a (or h → 0)slope of the tangent; an instantaneous rate
tangent liney = f(a) + f′(a)(x − a)the point from f, the slope from f′
velocityv(a) = s′(a)units of position per unit of time
no derivativeone-sided limits differ, or infinite, or f undefinedcorner, vertical tangent, gap or jump

\[ f'(a) = \lim_{h\to 0}\frac{f(a+h) - f(a)}{h} \]

Next, Section 3.2 lets a vary: the derivative becomes a function, and its graph tells the story of the original function's slopes.

Stewart, Calculus: Early Transcendentals 8e, §2.7 Derivatives and Rates of Change §2.7, pp. 140-151 — the same material in Stewart

One idea, several readings. The difference quotient is the slope of a secant and an average rate of change. Its limit is the derivative, which is the slope of the tangent and the instantaneous rate of change: velocity when the function is position.

To compute a derivative from the definition, simplify until the factor h is visible, cancel it, and only then let h go to zero. To write a tangent line, take the point from the function and the slope from the derivative. And when the one-sided limits disagree or blow up, as at corners and vertical tangents, the derivative does not exist, even if the function is continuous.

Next, Section 3.2 lets the point vary, so the derivative becomes a function, and its graph tells the story of the original function's slopes.

Sources

  1. OpenStax Calculus Volume 1, §3.1 Defining the Derivative — Strang & Herman, OpenStax / Rice University, CC BY-NC-SA 4.0, pp. 188-202
  2. Stewart, Calculus: Early Transcendentals 8e, §2.7 Derivatives and Rates of Change — James Stewart, Cengage Learning, 2016, pp. 140-151

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

Book on Wyzant · Text (657) 465-8108