The tangent problem and the recognition of a tangent as the limit of secant lines, instantaneous velocity as the limit of average velocities, the area problem solved by inscribed polygons and rectangles, and the observation that both ancient problems dissolve under the same limiting manoeuvre — the idea Section 2.2 will define.
Subject: Calculus I · 66 slides · symbolic lesson
Open the interactive version of this deck
Title
Calculus I · Section 2.1
Two old problems, and the one idea that solves both
Objectives
Calculus grew out of two questions: how steep is a curve at a single point, and how much area lies under it. Neither can be answered with the algebra and geometry you already have. This lesson shows you how to approximate each answer as closely as you like, and names the step that makes the approximation exact.
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 106 — learning objectives 2.1.1 to 2.1.7
This section is a preview, and it is meant to feel like one. You will not meet a single new rule to memorise. Instead you will meet two old problems that the mathematics you already know cannot finish, and watch the same idea finish both of them.
The first problem is the slope of a curve at one point. The second is the area under a curve. For each, you will do real computations: secant slopes, average velocities, totals of rectangles. Each computation gives an approximation, and a better one when you push the numbers closer together.
The last step, the one that turns good approximations into an exact answer, is called taking a limit. You will use it informally here. The rest of Chapter 2 is spent making it precise, because every idea in calculus is built on it.
Warm-up
Discussion prompt
Find the slope of the line through (1, 1) and (3, 9). Then try the same formula with only the single point (1, 1), used twice. What goes wrong?
Commit to an answer before you reveal. The first part is ordinary algebra: rise over run between two points gives 4. The second part is the interesting one.
If you try to measure the slope at a single point with the same formula, both the rise and the run come out as zero. Zero over zero is not zero, and it is not undefined in the harmless sense of a vertical line either. It is simply no information at all: the formula has nothing to work with.
Keep this in mind for the whole lesson. The slope of a curve at a point is a perfectly sensible thing to want, and the obvious formula cannot give it. Everything on the following slides is a way around that zero over zero.
Intuition
| problem | the question | what answers it |
|---|---|---|
| tangent problem | how steep is a curve at one point? | the derivative (differential calculus) |
| velocity problem | how fast is an object moving at one instant? | the derivative again |
| area problem | how much area lies under a curve? | the integral (integral calculus) |
All three ask for something that ordinary formulas cannot deliver in one step. All three are answered the same way: compute approximations you can handle, make them better and better, and ask what number they approach.
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 106 — the tangent and area problems
Read the table as a map of the whole course. The tangent problem and the velocity problem are really one problem: velocity is the rate at which position changes, just as slope is the rate at which height changes. Both lead to the derivative, which is Chapters 3 and 4.
The area problem looks different, since it is about accumulation rather than steepness. It leads to the integral, which is Chapter 5. One of the great surprises of the subject, the Fundamental Theorem of Calculus, is that these two problems turn out to be inverses of each other.
For now, notice only the shared strategy in the last sentence. When you cannot compute something directly, compute approximations you can handle, make them better, and ask what they are closing in on. That is what a limit is.
Section
Part 1
Concept
Figure (svg): Three small graphs side by side: the line f(x) = −2x − 3 falling steeply, the line g(x) = x/2 + 1 rising gently, and the horizontal line h(x) = 2. Dashed rise-and-run steps are drawn on the first two.
\[ f(x) = -2x - 3, \qquad g(x) = \tfrac12 x + 1, \qquad h(x) = 2 \]
For a line, the slope is the rate of change: for every one unit you move right, f drops by 2, g rises by one half, and h does not change at all. One number describes the whole graph.
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 106 — Figure 2.2
Look at the dashed steps on the first two graphs. On the falling line, one step right brings you two steps down; on the rising one, two steps right bring you one step up. Take those steps anywhere on the line and the ratio is the same.
That is what makes lines easy. The slope is the rate of change, and it is one number for the whole graph. The horizontal line has slope zero because its output never changes at all.
Hold on to this fact, because it is the tool the whole lesson relies on. Even when a graph is curved, you will always compute slopes of lines, specifically lines through two points of the curve, and then use them to learn about the curve.
Concept
Figure (svg): The parabola k(x) = x squared, with short tangent segments at x = −2, −1, 0, 1 and 2. Their slopes are labelled −4, −2, 0, 2 and 4.
\[ k(x) = x^2 \]
The parabola falls fast, levels off, then rises faster and faster. There is no single slope to quote. So the question becomes: what is the rate of change at one particular point?
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, pp. 106-107 — Figure 2.3
Each short yellow segment shows how steep the parabola is at one point. On the far left it falls four units per unit step; at the bottom it is momentarily flat; on the right it climbs at two and then four units per step. Every point has its own steepness.
So asking for the slope of the parabola makes no sense until you say where. The honest question is: what is the rate of change at this particular point? That is the tangent problem.
You may notice from the labels that the slope at each point is twice the x-coordinate. That pattern is not a coincidence, and by the end of this lesson you will be able to see where it comes from for the point at x equal to 1.
Concept
secant line — The line through two points (a, f(a)) and (x, f(x)) on the graph of f.
\[ m_{\text{sec}} = \frac{f(x) - f(a)}{x - a} \qquad (2.1) \]
Figure (svg): A smooth rising curve with two marked points, (a, f(a)) and (x, f(x)). A straight secant line passes through both. A dashed horizontal leg of length x − a and a dashed vertical leg of length f(x) − f(a) form a right triangle under the secant.
A secant slope is an ordinary rise over run, so you can always compute it. It is the average rate of change of f between a and x, and it approximates the rate of change at a.
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 107 — Definition, equation 2.1 and Figure 2.4
A secant line is the most useful line you can draw on a curve, because you can always compute its slope. Pick the point you care about, pick a second point on the graph, and take rise over run. The figure shows the run along the bottom and the rise up the side.
The name comes from the Latin for cutting: a secant cuts the curve at two points. Its slope is the average rate of change of the function between them. If the curve rose steeply at first and then flattened, the secant averages those behaviours together.
Because it is an average over a stretch, a secant slope is only an approximation to the rate of change at the point a. The shorter the stretch, the less averaging, and the better the approximation. That single observation drives the rest of Part 1.
Notation
Annotate
On: \( m_{\text{sec}} = \frac{f(x) - f(a)}{x - a}, \qquad x \ne a \)
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 107 — equation 2.1
Step through the annotations one by one. The roles of the two letters are different, and keeping them straight prevents most of the errors in this section. The letter a is fixed: it is the point you are asking about. The letter x is the one you move.
The numerator is the change in output and the denominator the change in input, so the formula is literally rise over run. Nothing about it is new. What is new is the way you are going to use it: many times, with x creeping closer to a.
The last annotation is the important one. The formula is only defined when x differs from a. You will never set x equal to a in this fraction; you will only ask what happens to its value as x gets near a.
Prediction
\[ f(x) = x^2, \qquad a = 1 \]
Predict first
You draw a secant from (1, 1) to the point at x = 2, then another from (1, 1) to the point at x = 1.5. Which secant is steeper?
Correct: The one to x = 2
Why: The parabola bends upward, getting steeper as x grows. A secant to a farther point averages in more of that steep part, so the secant to x equal to 2 is steeper. Moving the second point in makes the secants shallower, and they settle toward the slope at x equal to 1.
Make your prediction before revealing. Picture the parabola bending upward and imagine the two secant lines from the point at x equal to 1.
The parabola gets steeper as you move right. A secant to x equal to 2 spans a stretch that includes some of that steeper part, so its average steepness is larger. Pull the second point in to 1.5 and less of the steep part is included, so the secant is shallower.
Keep going and the pattern continues: the closer the second point, the less extra steepness gets averaged in, and the secant slopes settle down toward the steepness at x equal to 1 itself. The next three examples put numbers on exactly this.
Worked example
Estimate the slope of the tangent line to the parabola at x equal to 1, using the secant through (1, 1) and (2, 4).
\[ f(x) = x^2, \qquad a = 1, \qquad x = 2 \]
Write the secant slope formula
Why: Definition 2.1 with a equal to 1 and x equal to 2.
\[ m_{\text{sec}} = \frac{f(2) - f(1)}{2 - 1} \]
Evaluate f at both points
Why: Square each input.
\[ f(2) = 2^2 = 4, \qquad f(1) = 1^2 = 1 \]
Substitute
Why: Rise on top, run underneath.
\[ m_{\text{sec}} = \frac{4 - 1}{2 - 1} \]
Simplify
Why: Three over one.
\[ m_{\text{sec}} = \frac{3}{1} = 3 \]
Check by counting on the graph
Why: From (1, 1) to (2, 4) you move one unit right and three units up, so the rise over run is three.
\[ \text{run} = 1, \quad \text{rise} = 3 \quad\Longrightarrow\quad \frac{3}{1} = 3 \; \checkmark \]
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, pp. 108-109 — Example 2.1a
This is the simplest possible secant computation, and it is worth doing slowly once. Write the formula with the actual numbers in it before evaluating anything, because that line is where you see which point goes where.
The function values are just squares: 4 at x equal to 2, and 1 at x equal to 1. The rise is 3, the run is 1, and the secant slope is 3.
The check is the graph itself. Walking from one point to the other you go one unit right and three units up, which is the same rise over run. Notice that 3 is only an estimate of the slope at x equal to 1; the next example uses a nearer point to get a better one.
Worked example
Repeat with a second point closer to (1, 1).
\[ f(x) = x^2, \qquad a = 1, \qquad x = \tfrac32 \]
Write the secant slope formula
Why: Same definition, new second point.
\[ m_{\text{sec}} = \frac{f\left(\frac32\right) - f(1)}{\frac32 - 1} \]
Evaluate f at three halves
Why: Square the numerator and the denominator.
\[ f\left(\tfrac32\right) = \tfrac94 \]
Simplify the rise
Why: Write 1 as four quarters.
\[ \frac94 - 1 = \frac94 - \frac44 = \frac54 \]
Simplify the run
Why: Write 1 as two halves.
\[ \frac32 - 1 = \frac12 \]
Divide
Why: Dividing by one half is multiplying by 2.
\[ m_{\text{sec}} = \frac{5/4}{1/2} = \frac54 \cdot 2 = \frac52 = 2.5 \]
Figure (svg): Two panels, each showing y = x squared, a faint dashed tangent line at (1, 1), and a secant. Left: the secant through (1, 1) and (2, 4), slope 3. Right: the secant through (1, 1) and (3/2, 9/4), slope 2.5, lying closer to the dashed tangent.
Check with decimals
Why: The same computation with 1.5 and 2.25 gives the same slope. The nearer point gives the better estimate, and the book puts the tangent slope between 2 and 2.5.
\[ \frac{2.25 - 1}{1.5 - 1} = \frac{1.25}{0.5} = 2.5 \; \checkmark \]
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 109 — Example 2.1b and Figure 2.7
The point three halves, nine quarters is closer to (1, 1), so its secant should be a better estimate. The arithmetic now involves fractions, and the cleanest way through is to simplify the rise and the run separately before dividing.
Nine quarters minus one is five quarters. Three halves minus one is one half. Dividing five quarters by one half is the same as multiplying by two, which gives five halves, or 2.5.
Look at the two panels. The secant in panel (b) is visibly closer to the dashed tangent line than the one in panel (a). The book concludes that the tangent slope is somewhere between 2 and 2.5, which is exactly the kind of statement a pair of secant slopes lets you make.
Worked example
Estimate the tangent slope at x equal to 1 with a still closer point.
\[ f(x) = x^2, \qquad a = 1, \qquad x = \tfrac54 \]
Write the secant slope formula
Why: Rise over run from (1, 1).
\[ m_{\text{sec}} = \frac{\frac{25}{16} - 1}{\frac54 - 1} \]
Simplify the rise
Why: One is sixteen sixteenths.
\[ \frac{25}{16} - \frac{16}{16} = \frac{9}{16} \]
Simplify the run
Why: One is four quarters.
\[ \frac54 - \frac44 = \frac14 \]
Divide
Why: Dividing by one quarter is multiplying by 4.
\[ m_{\text{sec}} = \frac{9}{16} \cdot 4 = \frac94 = 2.25 \]
Check with decimals and against the trend
Why: In decimals the rise is 0.5625 and the run 0.25. The slopes 3, 2.5, 2.25 are shrinking toward 2 as the point moves in.
\[ \frac{1.5625 - 1}{1.25 - 1} = \frac{0.5625}{0.25} = 2.25 \; \checkmark \]
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 109 — Checkpoint 2.1
This checkpoint moves the second point closer still, to x equal to five quarters. Work it on your own first, then compare with the steps.
Again simplify the rise and run separately. Twenty-five sixteenths minus one is nine sixteenths; five quarters minus one is one quarter. Dividing by one quarter multiplies by four, and nine sixteenths times four is nine quarters, which is 2.25.
Now line up the three results: 3, then 2.5, then 2.25, as the run shrank from 1 to one half to one quarter. Each time the run halved, the distance above 2 halved too. That is a strong hint that the secant slopes are heading to exactly 2, and you will prove it two slides from now.
Picture it
Figure (svg): The parabola y = x squared with the point (1, 1) marked. Secant lines from (1, 1) to the points at x = 2, 1.5 and 1.25 fan out, each a little less steep, and the tangent line y = 2x − 1 lies at the end of the fan. A legend beneath gives the slopes 3, 2.5, 2.25 and 2.
Three secants from the same point, each to a nearer second point. Their slopes, 3 then 2.5 then 2.25, fall toward a value that no secant ever reaches: the slope of the tangent line, 2.
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 108 — Figures 2.5 and 2.6
Watch the lines appear one after another. Each new secant uses a second point closer to (1, 1), and each is a little less steep than the one before. The yellow line at the end of the sequence is the one they are converging on.
That yellow line is the tangent line at (1, 1). It is not one of the secants; no secant through two different points ever lands exactly on it. It is the line the secants approach, and its slope, 2, is the number the secant slopes approach.
This picture is the definition of the tangent line in this course. Notice that it says nothing about touching the curve at only one point. Two slides from now you will see why that popular description is wrong.
Concept
tangent line — The line that the secant lines through (a, f(a)) approach as x approaches a. Its slope is the rate of change of f at a.
\[ m_{\text{tan}} = \text{the value } \frac{f(x) - f(a)}{x - a} \text{ approaches as } x \to a \]
That slope is called the derivative of f at a, written with a prime. The study of derivatives and their uses is differential calculus.
\[ m_{\text{tan}} = f'(a) \]
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 108 — the tangent and the derivative
This slide puts the picture into words. The tangent line at a point is the line that the secant lines approach as the second point slides toward the first. Its slope is the rate of change of the function at that point.
That slope gets a name and a notation of its own: the derivative of f at a, written f prime of a. For the parabola at 1 you now know that the derivative is 2. Chapter 3 is devoted to computing derivatives quickly; for now you compute them the slow way, as the value the secant slopes approach.
Differential calculus is simply the study of derivatives and what they are good for. Everything in it comes from this one construction.
Notation
Annotate
On: \( f'(a) = \lim_{x \to a} \frac{f(x) - f(a)}{x - a} \)
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 109 — taking a limit
This is the first appearance of the limit notation, and it is worth taking it apart carefully. Read it from the inside out. The fraction is an ordinary secant slope. The letters underneath say that x is approaching a. The whole expression means the value those secant slopes approach.
Two things in the notation deserve emphasis. First, x approaches a from either side: second points to the left of a are just as good as points to the right, and both should lead to the same value. Second, x never actually equals a, because at a the fraction is zero over zero.
Section 2.2 will define the limit carefully and show you how to find it from tables and graphs. For now, read the symbol as a compact way of writing the sentence on the previous slide.
Fill the middle
For the parabola at (1, 1), the secant slope to a general point x can be simplified before x moves anywhere.
Fill in the blanks
\fracx + 12 = \frac______ = ___ \qquad \text___ x \to 1, \quad m____} \to ___
Why: Factor the difference of squares and cancel the common factor, which is allowed because x is not 1. Every secant slope equals x plus 1, and as x approaches 1 that approaches 2.
Fill in both blanks before checking. The first blank asks what is left after the cancellation; the second asks where that expression goes as x approaches 1.
Factoring x squared minus 1 as a difference of squares exposes a factor of x minus 1 on top that matches the one underneath. Because x is not 1, that factor is not zero, and you may cancel it. What remains is x plus 1.
This is the key algebraic move of the whole chapter. The original fraction cannot be evaluated at x equal to 1, but the simplified expression can, and it gives 2. The cancellation removes the zero over zero without ever setting x equal to 1 in the fraction.
Worked example
Replace the table of slopes with one formula, then read off where it goes.
Write the secant slope for a general x
Why: Any x other than 1.
\[ m_{\text{sec}}(x) = \frac{x^2 - 1}{x - 1}, \qquad x \ne 1 \]
Factor the numerator
Why: A difference of squares.
\[ x^2 - 1 = (x - 1)(x + 1) \]
Cancel the common factor
Why: Legal because x minus 1 is not zero.
\[ m_{\text{sec}}(x) = \frac{(x - 1)(x + 1)}{x - 1} = x + 1 \]
Let x approach 1
Why: The expression x plus 1 heads to 2.
\[ x \to 1 \quad\Longrightarrow\quad m_{\text{sec}}(x) \to 1 + 1 = 2 \]
Figure (svg): The graph of the secant slope m(x) = x + 1 for the parabola at (1, 1): a straight line with an open circle at x = 1, height 2. Dots on the line mark the slopes already computed at x = 0.5, 0.9, 1.1, 1.25, 1.5 and 2.
Check against the three slopes already computed
Why: The formula reproduces Example 2.1 and Checkpoint 2.1 exactly.
\[ x = 2: \; 3, \qquad x = \tfrac32: \; 2.5, \qquad x = \tfrac54: \; 2.25 \; \checkmark \]
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 108 — the secant slopes approaching the tangent slope
The table of secant slopes was a sequence of separate computations. This example replaces all of them with one formula, which is always better when you can get it.
Factor, cancel, and the secant slope for any second point x is simply x plus 1. As x approaches 1, that approaches 2, so the tangent slope at (1, 1) is exactly 2, not just approximately.
The figure shows what this means. The secant slopes all lie on the straight line m equal to x plus 1, with a hole at x equal to 1, because there is no secant at the point itself. The limit is the height of the hole: the value the graph is heading to even though it has no value there. You will see this hole picture again and again in Chapter 2.
Worked example
With the slope known, write the equation of the tangent line.
\[ \text{point } (1, 1), \qquad \text{slope } f'(1) = 2 \]
Use point-slope form
Why: A point and a slope determine a line.
\[ y - 1 = 2(x - 1) \]
Distribute
Why: Two times x minus 1.
\[ y - 1 = 2x - 2 \]
Solve for y
Why: Add 1 to both sides.
\[ y = 2x - 1 \]
Check that the line touches the parabola only at (1, 1)
Why: Subtract the line from the parabola: the difference is a perfect square, zero at x equal to 1 and positive everywhere else, so the parabola sits on or above its tangent.
\[ x^2 - (2x - 1) = x^2 - 2x + 1 = (x - 1)^2 \ge 0 \]
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 108 — the tangent line; compare Exercise 3
Once you know a point and a slope, the tangent line is ordinary point-slope algebra from Chapter 1. With the point (1, 1) and slope 2, the line is y equal to 2x minus 1.
The check is worth understanding, not just reading. Subtracting the line from the parabola gives x minus 1, squared. A square is never negative, and it is zero only at x equal to 1. So the parabola lies on or above this line everywhere and meets it at exactly one point.
That repeated factor, x minus 1 squared, is the algebraic signature of touching rather than crossing. A secant would give two different factors, one for each point where it cuts the curve. Keep this in mind for the counterexample coming up shortly, where a tangent does cross.
Trap
A tempting shortcut for the slope at (1, 1):
\[ m = \frac{1^2 - 1}{1 - 1} = \frac{0}{0} \]
\[ \Longrightarrow\; m = 0 \text{ ?} \]
Wrong. Zero over zero is not zero; it is not a number at all.
The secant formula needs two different points. Keep x away from a, simplify, and only then let x approach a.
\[ \frac{x^2 - 1}{x - 1} = x + 1 \]
\[ x + 1 \to 2 \]
This mistake is natural: you want the slope at 1, so you put 1 in. The formula gives zero over zero, and it is tempting to call that zero, or to decide the slope does not exist. Both conclusions are wrong.
Zero over zero carries no information. The fraction was built to compare two different points, and with only one point there is nothing to compare. The fact that the direct substitution fails says nothing about whether the tangent slope exists.
The fix is the order of operations you just practised: simplify while x is still different from a, then let x approach a. For the parabola that gives 2, a perfectly good slope. Whenever substitution produces zero over zero in this chapter, read it as a signal to simplify, never as an answer.
Intuition
Figure (svg): Three panels zooming in on the point (1, 1) of y = x squared, with the dashed tangent line y = 2x − 1. In the widest window the parabola visibly bends away from the tangent; at a width of 0.5 it bends only slightly; at a width of 0.1 the curve and the tangent are indistinguishable.
Zoom in on (1, 1) and the parabola and its tangent become harder and harder to tell apart. That is what the tangent line is: the straight line the curve looks like when you look closely enough.
The three panels show the same point, the same parabola and the same dashed tangent at three magnifications. In the widest window the curve clearly bends away from the line. At the middle magnification it bends only a little. In the narrowest window you cannot see any difference at all.
This is the most useful way to think about a tangent line. It is the line that the curve looks like when you zoom in far enough on the point. A smooth curve is, locally, almost straight, and the tangent is the straight line it almost is.
It also explains why secant slopes work. Two points very close together on the curve lie almost on the tangent, so the line through them has almost the tangent's slope.
Section
Part 2
Worked example
Fill the table of secant slopes for points Q approaching P, guess the tangent slope, and write the tangent line.
\[ f(x) = x^2 + 1, \qquad P(1, 2) \]
Compute the height of Q at x equal to 1.1
Why: Square, then add 1.
\[ f(1.1) = 1.21 + 1 = 2.21 \]
Compute the secant slope
Why: Rise over run from P.
\[ m_{\text{sec}} = \frac{2.21 - 2}{1.1 - 1} = \frac{0.21}{0.1} = 2.1 \]
Repeat for the nearer points
Why: The same two moves at each x.
| x | y | Q(x, y) | slope |
|---|---|---|---|
| 1.1 | 2.21 | (1.1, 2.21) | 2.1 |
| 1.01 | 2.0201 | (1.01, 2.0201) | 2.01 |
| 1.001 | 2.002001 | (1.001, 2.002001) | 2.001 |
| 1.0001 | 2.0002000 | (1.0001, 2.0002000) | 2.0001 |
\[ m_{\text{sec}} \to 2 \]
Write the tangent line (Exercise 3)
Why: Point-slope form through (1, 2) with slope 2.
\[ y - 2 = 2(x - 1) \quad\Longrightarrow\quad y = 2x \]
Check by simplifying the secant slope
Why: The plus 1 cancels, leaving the same quotient as for x squared, so the limit is 2 exactly.
\[ \frac{(x^2 + 1) - 2}{x - 1} = \frac{x^2 - 1}{x - 1} = x + 1 \to 2 \; \checkmark \]
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 113 — Exercises 1-3
These exercises turn the secant method into a table, which is how you will estimate limits numerically in Section 2.2. For each x, first find the height of Q, then the slope of the line from P to Q.
The slopes 2.1, 2.01, 2.001, 2.0001 follow an unmistakable pattern: each time the run shrinks by a factor of ten, the excess over 2 does too. So the tangent slope is 2, and the tangent line through (1, 2) with that slope is y equal to 2x.
The check shows why the answer matches the plain parabola. Adding 1 to the function lifts the whole graph without changing its shape, so every secant slope is unchanged. After the plus 1 cancels, the fraction is the same one you simplified to x plus 1.
Worked example
\[ f(x) = x^3, \qquad P(1, 1) \]
Height and slope at x equal to 1.1
Why: Cube 1.1, then take rise over run.
\[ f(1.1) = 1.331, \qquad m_{\text{sec}} = \frac{1.331 - 1}{0.1} = 3.31 \]
Repeat for the nearer points
Why: Round to eight significant digits, as the exercise asks.
| x | y | slope |
|---|---|---|
| 1.1 | 1.331 | 3.31 |
| 1.01 | 1.030301 | 3.0301 |
| 1.001 | 1.0030030 | 3.003001 |
| 1.0001 | 1.0003000 | 3.0003000 |
\[ m_{\text{sec}} \to 3 \]
Write the tangent line (Exercise 6)
Why: Point-slope form through (1, 1) with slope 3.
\[ y - 1 = 3(x - 1) \quad\Longrightarrow\quad y = 3x - 2 \]
Check by factoring a difference of cubes
Why: The quotient simplifies to a quadratic whose value at 1 is 3.
\[ \frac{x^3 - 1}{x - 1} = x^2 + x + 1 \to 1 + 1 + 1 = 3 \; \checkmark \]
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 113 — Exercises 4-6
The cube grows faster than the square, and the table shows it: the secant slopes are 3.31, 3.0301, 3.003001, 3.0003. Read the trend rather than the last entry, and the tangent slope is 3.
With slope 3 through (1, 1), point-slope form gives the tangent line y equal to 3x minus 2.
The check uses the difference of cubes factorisation: x cubed minus 1 is x minus 1 times x squared plus x plus 1. Cancel the common factor and the secant slope becomes x squared plus x plus 1, which is 3 when x is 1. Notice the pattern forming: for the square the tangent slope at 1 was 2, and for the cube it is 3. Chapter 3 explains why.
Counterexample
Discussion prompt
A classmate defines a tangent as 'a line that meets the curve at exactly one point'. Use the tangent line you just found for x cubed at (1, 1) to break this definition.
Figure (svg): The cubic y = x cubed and its tangent line y = 3x − 2 at (1, 1). The line touches the curve at (1, 1) and crosses it again at (−2, −8), both marked.
Many people carry the definition 'a tangent touches the curve at one point' from geometry, where it is true for circles. Try to break it with the cubic before you reveal the answer.
Set the cubic equal to its tangent line and rearrange. The resulting cubic factors as x minus 1 squared times x plus 2. The squared factor is the touching point at x equal to 1, just as with the parabola. But there is another root, x equal to minus 2, where the tangent line cuts straight through the curve.
So a tangent can meet a curve more than once. And the converse fails too: a vertical line meets a parabola exactly once and is not tangent to it. Meeting once is the wrong idea. The definition that always works is the one from this lesson, the limit of secant lines.
Worked example
\[ f(x) = \sqrt{x}, \qquad P(4, 2) \]
Height and slope at x equal to 4.1
Why: Take the root, then rise over run.
\[ \sqrt{4.1} \approx 2.0248457, \qquad m_{\text{sec}} \approx \frac{0.0248457}{0.1} = 0.24845673 \]
Repeat for the nearer points
Why: Eight significant digits.
| x | y | slope |
|---|---|---|
| 4.1 | 2.0248457 | 0.24845673 |
| 4.01 | 2.0024984 | 0.24984395 |
| 4.001 | 2.0002500 | 0.24998438 |
| 4.0001 | 2.0000250 | 0.24999844 |
\[ m_{\text{sec}} \to 0.25 = \tfrac14 \]
Write the tangent line (Exercise 9)
Why: Slope one quarter through (4, 2).
\[ y - 2 = \tfrac14(x - 4) \quad\Longrightarrow\quad y = \tfrac14 x + 1 \]
Figure (svg): The curve y = square root of x from 0 to 9.5 and the dashed tangent line y = x/4 + 1 touching it at the marked point (4, 2). The tangent lies above the curve everywhere else.
Check by rationalising
Why: Factor x minus 4 as a difference of squares of roots, then cancel.
\[ \frac{\sqrt{x} - 2}{x - 4} = \frac{\sqrt{x} - 2}{(\sqrt{x} - 2)(\sqrt{x} + 2)} = \frac{1}{\sqrt{x} + 2} \to \frac14 \; \checkmark \]
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 114 — Exercises 7-9
The square root rises more and more slowly, and its secant slopes near 4 are a little under one quarter: 0.2484, 0.2498, 0.24998, 0.249998. The tangent slope is one quarter.
Point-slope form with slope one quarter through (4, 2) gives y equal to one quarter x plus 1. In the figure the dashed tangent lies above the curve everywhere except at the point of contact, because the square root bends downward.
The check introduces a technique you will need in Section 2.3. The difference x minus 4 can be written as root x minus 2 times root x plus 2, a difference of squares in disguise. Cancelling root x minus 2 leaves one over root x plus 2, which is one quarter at x equal to 4.
Comparison
\[ f(x) = \frac1x, \qquad P(-1, -1) \]
Comparison matrix
| x | y = 1/x | secant slope |
|---|---|---|
| −1.05 | −0.95238095 | −0.95238095 |
| −1.01 | −0.99009901 | −0.99009901 |
| −1.005 | −0.99502488 | −0.99502488 |
| −1.001 | −0.99900100 | −0.99900100 |
| limit | −1 | −1 |
\[ \frac{\frac1x - (-1)}{x - (-1)} = \frac{\frac{1 + x}{x}}{x + 1} = \frac1x \]
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 115 — Exercises 13-15
Fill in the blanks with a calculator, keeping eight significant digits. The second point Q approaches P from the left here, since the x-values are below minus 1.
Something unexpected happens: the secant slope column is identical to the y column. The algebra underneath the table explains it. The rise, one over x plus 1, has x plus 1 over x as a single fraction, and dividing by the run, x plus 1, leaves exactly one over x.
So the secant slopes are one over x, and as x approaches minus 1 they approach minus 1. That is the tangent slope for Exercise 14, and the tangent line through (minus 1, minus 1) with slope minus 1 is y equal to minus x minus 2 for Exercise 15.
Estimation
\[ f(\varphi) = \cos(\pi\varphi), \qquad P(1.5, 0) \]
| φ | secant slope |
|---|---|
| 1.4 | 3.0901699 |
| 1.49 | 3.1410759 |
| 1.499 | 3.1415875 |
| 1.4999 | 3.1415926 |
Predict first
Estimate the slope of the tangent line at φ = 1.5 from the table.
Correct: π exactly
Why: The digits 3.14159 are hard to miss: the secant slopes are closing in on pi. Chapter 3 will confirm it, since the derivative of cos of pi phi is minus pi times sin of pi phi, and the sine of three pi over two is minus one. The tangent line for Exercise 12 is y equal to pi times phi minus 1.5.
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 114 — Exercises 10-12
Look at the digits before choosing. The secant slopes read 3.09, 3.141, 3.14158, 3.1415926. Anyone who has seen pi to a few places recognises it.
A table can only ever suggest a limit, never prove it, but a suggestion this strong is worth trusting provisionally. The limit is pi exactly, not a rounded decimal. In Chapter 3 you will be able to confirm it in one line with the derivative of cosine.
Notice the variable here is phi instead of x, and the book writes the second point as phi comma y. The letters change from problem to problem; the method, rise over run to a nearer and nearer point, does not.
Section
Part 3
Concept
average velocity — For a position s(t), the change in position over an interval divided by the length of the interval.
\[ v_{\text{ave}} = \frac{s(t) - s(a)}{t - a} \qquad (2.2) \]
Figure (svg): The height of the dropped rock, s(t) = −16t² + 64, for t from 0 to 2 seconds, falling from 64 to 0. From the point at t = 0.5, height 60, secants run to t = 1.5 and t = 1, and the tangent line at t = 0.5 is drawn; their slopes are −32, −24 and −16.
Put the position graph next to the secant formula and they are the same fraction with different letters. Every average velocity is the slope of a secant on the position graph.
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 109 — Definition and equation 2.2
Look at the figure: the curve is the height of a falling rock against time. Each coloured line joins the point at half a second to a later point. The slope of each line is the change in height divided by the elapsed time, which is exactly the average velocity over that interval.
The definition is the secant slope formula with new letters: s for position instead of f, and t for time instead of x. Everything you learned about secants carries over unchanged.
Notice the units. A slope on this graph is feet divided by seconds, which is a velocity. The slopes are negative because the rock's height is decreasing. Over the long interval from 0.5 to 1.5 seconds the average is minus 32 feet per second; over the shorter one it is minus 24; the tangent at half a second has slope minus 16.
Concept
A speedometer reports how fast you are going right now, over no interval at all. An interval of length zero gives zero over zero again, so the answer has to come from a limit.
instantaneous velocity — The value that the average velocities over the intervals from a to t, and from t to a, approach as t gets closer to a, provided such a value exists.
\[ v(a) = \text{the value } \frac{s(t) - s(a)}{t - a} \text{ approaches as } t \to a \]
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 110 — Definition of instantaneous velocity
A speedometer does not report an average over a trip. It reports how fast you are going now. That is a velocity at an instant, and an instant has no length, so the average velocity formula would divide by zero.
The definition gets around this the same way the tangent did. Compute averages over shorter and shorter intervals, on both sides of the instant, and take the value they approach. If both sides approach the same value, that value is the instantaneous velocity.
The phrase 'provided such a value exists' matters. Later in the chapter you will meet functions where the two sides disagree, or where the values never settle. For an object moving smoothly, like every example in this section, the averages always close in on one number.
Notation
Annotate
On: \( v(a) = \lim_{t \to a} \frac{s(t) - s(a)}{t - a} \)
Compare this formula with the tangent slope formula from Part 1. They are identical except for the letters. The instantaneous velocity is the slope of the tangent line to the position graph.
The second annotation handles a small detail. The book allows intervals on either side of a: from a forward to t, or from t forward to a. The formula works for both because when t comes first, the top and bottom both change sign together, so the fraction still gives the average over that interval.
The last annotation is the physics. Velocity is the rate of change of position, and the derivative is the rate of change of a function. They are the same idea, which is why calculus and mechanics were invented together.
Step zero
\[ s(t) = -16t^2 + 64, \qquad \text{interval } [0.49, \, 0.5] \]
Discussion prompt
Example 2.2 asks for the rock's average velocity on this interval. Before you calculate: which two values of s do you need, what do you divide by, and should the answer be positive or negative?
Answer this before touching a calculator. Deciding what you need and what the answer should look like takes ten seconds and catches most mistakes before they happen.
You need the height at each end of the interval: at 0.49 and at 0.5 seconds. The change in height goes on top, the length of the interval, one hundredth of a second, underneath.
Predicting the sign is the valuable part. The rock is falling, so its height is decreasing, so any change in height over a forward interval is negative. If your computed velocity comes out positive, you know immediately that the subtraction was done in inconsistent orders.
Worked example
A rock is dropped from 64 feet; its height after t seconds, for t from 0 to 2, is given below. Find the average velocities on the two intervals and guess the velocity at t equal to 0.5.
\[ s(t) = -16t^2 + 64, \qquad \text{(a) } [0.49, \, 0.5], \qquad \text{(b) } [0.5, \, 0.51] \]
Height at t equal to 0.5
Why: Square, multiply by minus 16, add 64.
\[ s(0.5) = -16(0.25) + 64 = 60 \]
Height at t equal to 0.49
Why: The square of 0.49 is 0.2401.
\[ s(0.49) = -16(0.2401) + 64 = 60.1584 \]
Average velocity on (a)
Why: Change in height over one hundredth of a second.
\[ v_{\text{ave}} = \frac{60 - 60.1584}{0.5 - 0.49} = \frac{-0.1584}{0.01} = -15.84 \]
Height at t equal to 0.51
Why: The square of 0.51 is 0.2601.
\[ s(0.51) = -16(0.2601) + 64 = 59.8384 \]
Average velocity on (b)
Why: Again over one hundredth of a second.
\[ v_{\text{ave}} = \frac{59.8384 - 60}{0.51 - 0.5} = \frac{-0.1616}{0.01} = -16.16 \]
Guess the instantaneous velocity
Why: The two averages sit on either side of minus 16.
\[ -16.16 < v(0.5) < -15.84 \quad\Longrightarrow\quad v(0.5) \approx -16 \text{ ft/s} \]
Check with the algebra of a general interval
Why: Over the interval from 0.5 to 0.5 plus h, the average simplifies to minus 16 times 1 plus h. Putting h equal to minus 0.01 and 0.01 reproduces both answers, and h going to zero gives exactly minus 16.
\[ \frac{s(0.5 + h) - s(0.5)}{h} = \frac{-16(h + h^2)}{h} = -16(1 + h) \to -16 \; \checkmark \]
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 110 — Example 2.2
This is the velocity version of Example 2.1. Two short intervals, one on each side of half a second, and two average velocities. The arithmetic is easier if you find the heights first and only then divide.
On the interval just before half a second the average is minus 15.84 feet per second; on the interval just after, minus 16.16. The rock is speeding up as it falls, so the later interval has the larger speed. The instantaneous velocity at half a second is trapped between them, and minus 16 is the natural guess.
The check confirms the guess exactly. Over the interval from 0.5 to 0.5 plus h, the average simplifies to minus 16 times 1 plus h. Put h equal to minus one hundredth and one hundredth and you get the two computed values; let h shrink to zero and you get minus 16 exactly.
Trap
Asked for the rock's velocity at half a second:
\[ v = \frac{s(0.5)}{0.5} = \frac{60}{0.5} \]
\[ = 120 \text{ ft/s} \]
Wrong. A height divided by a time is not a velocity.
Velocity is a change in position divided by a change in time. The rock's height, 60 feet, says where it is, not how fast it moves; and a falling rock must have a negative velocity.
\[ \frac{s(0.51) - s(0.5)}{0.51 - 0.5} = -16.16 \]
Dividing where the rock is by the time elapsed looks like a velocity, and it even has the right units. But it answers the wrong question. It would be the average velocity from time zero only if the rock had started at height zero, which it did not.
Velocity is always about change: how much the position changed, over how much time. The rock's height of 60 feet at half a second is a location, and says nothing about motion. A rock could be sitting still at a height of 60 feet.
The sign is a second warning. The rock is falling, so its velocity must be negative. Any method that gives a positive velocity for a falling object has gone wrong somewhere. Check signs and differences every time.
Worked example
Estimate the instantaneous velocity at t equal to 2 from the average over the interval from 2 to 2.001.
\[ s(t) = t^3, \qquad [2, \, 2.001] \]
Position at both ends
Why: Cube each time.
\[ s(2) = 8, \qquad s(2.001) = 2.001^3 = 8.012006001 \]
Change in position
Why: Subtract.
\[ s(2.001) - s(2) = 0.012006001 \]
Divide by the elapsed time
Why: One thousandth of a second.
\[ v_{\text{ave}} = \frac{0.012006001}{0.001} = 12.006001 \]
Estimate
Why: A very short interval gives a very good estimate.
\[ v(2) \approx 12 \]
Figure (svg): The average velocity of s(t) = t cubed over the interval from 2 to 2 + h, which equals 12 + 6h + h², plotted against h from −0.6 to 1.1. It is a smooth curve with an open circle at h = 0, height 12.
Check by expanding the cube
Why: Over the interval from 2 to 2 plus h the average is 12 plus 6h plus h squared. At h equal to 0.001 that is the value above, and as h shrinks it approaches 12.
\[ \frac{(2 + h)^3 - 8}{h} = \frac{12h + 6h^2 + h^3}{h} = 12 + 6h + h^2 \to 12 \; \checkmark \]
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 110 — Checkpoint 2.2
A single very short interval, one thousandth of a second, gives a very good estimate. Cube 2.001 carefully: it is 8.012006001, and the digits matter because you are about to divide a small difference by a small number.
The average velocity is 12.006001, so the instantaneous velocity at 2 is about 12.
The graph shows the average velocity as a function of the interval length h. It is defined for every h except zero, where there is a hole, and both sides of the curve head into that hole at height 12. The check expands the cube to show why: the average is 12 plus 6h plus h squared, and the extra terms vanish as h shrinks. The limit is exactly 12.
Worked example
\[ s(t) = 200 - 4.9t^2, \qquad \text{velocity at } t = 5 \]
Position at t equal to 5
Why: Twenty-five times 4.9 is 122.5.
\[ s(5) = 200 - 122.5 = 77.5 \]
Average on the interval from 4.99 to 5
Why: The square of 4.99 is 24.9001.
\[ s(4.99) = 77.98951, \qquad v_{\text{ave}} = \frac{77.5 - 77.98951}{0.01} = -48.951 \]
Average on the interval from 5 to 5.01
Why: The square of 5.01 is 25.1001.
\[ s(5.01) = 77.00951, \qquad v_{\text{ave}} = \frac{77.00951 - 77.5}{0.01} = -49.049 \]
The two shorter intervals
Why: Same method with a thousandth of a second.
\[ [4.999, 5]: \; -48.9951, \qquad [5, 5.001]: \; -49.0049 \]
Guess the instantaneous velocity (Exercise 17)
Why: The averages close in on minus 49 from both sides.
\[ v(5) \approx -49 \text{ m/s} \]
Check with a general interval
Why: The average from 5 to 5 plus h simplifies to minus 49 minus 4.9h; h equal to minus 0.01 gives minus 48.951, as computed.
\[ \frac{s(5 + h) - s(5)}{h} = \frac{-49h - 4.9h^2}{h} = -49 - 4.9h \to -49 \; \checkmark \]
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 115 — Exercises 16-17
This is the same method with metric units and gravity written as 4.9 rather than 16. The book asks for eight significant digits, which is enough to see the pattern clearly.
From the left the averages are minus 48.951 and minus 48.9951; from the right, minus 49.049 and minus 49.0049. They close in on minus 49 from both sides, symmetrically, so the ball is falling at 49 metres per second after five seconds.
The check writes the average over the interval from 5 to 5 plus h as minus 49 minus 4.9 h. Every entry in the table is that expression at one particular h, and the limit as h goes to zero is minus 49. When a table and a formula agree like this you can be confident in both.
Worked example
\[ h(t) = 15t - 4.9t^2, \qquad \text{velocity at } t = 1 \]
Height at t equal to 1
Why: Fifteen minus 4.9.
\[ h(1) = 15 - 4.9 = 10.1 \]
Average on the interval from 1 to 1.05
Why: Height at 1.05 first.
\[ h(1.05) = 10.34775, \qquad v_{\text{ave}} = \frac{10.34775 - 10.1}{0.05} = 4.955 \]
The shorter intervals
Why: Same method.
| interval | average velocity (m/s) |
|---|---|
| [1, 1.05] | 4.955 |
| [1, 1.01] | 5.151 |
| [1, 1.005] | 5.1755 |
| [1, 1.001] | 5.1951 |
\[ v_{\text{ave}} \to 5.2 \]
Figure (svg): The stone's height h(t) = 15t − 4.9t², an arch from t = 0 to about t = 3.06 seconds, peaking near 11.5 metres. The tangent line at t = 1, height 10.1, is dashed; its slope is 5.2.
Check with a general interval
Why: Over a short interval from 1 onward the average is 5.2 minus 4.9 times the interval length, which reproduces every row of the table and tends to 5.2.
\[ \frac{h(1 + \Delta t) - h(1)}{\Delta t} = \frac{5.2\,\Delta t - 4.9(\Delta t)^2}{\Delta t} = 5.2 - 4.9\,\Delta t \to 5.2 \; \checkmark \]
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 115 — Exercises 18-19
This time the object goes up and comes back down, as the arch in the figure shows. At one second it is still rising, so the velocity should be positive but smaller than the launch speed of 15 metres per second.
The averages over the shrinking intervals, 4.955, 5.151, 5.1755, 5.1951, all come from the right-hand side, and they climb toward 5.2. So at one second the stone is rising at about 5.2 metres per second.
The check writes the average over a short interval after 1 as 5.2 minus 4.9 times the interval length. That explains why these one-sided averages approach from below: every one of them is 5.2 minus something positive. The dashed tangent in the figure has slope exactly 5.2.
Fill the middle
\[ h(t) = 600 + 78.4t - 4.9t^2, \qquad h(9) = 908.7, \qquad h(9.01) = 908.60151 \]
Fill in the blanks
v_-9.849} = \frac______ = \frac______ = ___ \text___
Why: The rocket lost 0.09849 metres of height in one hundredth of a second, so the average velocity is minus 9.849 metres per second. The other intervals give minus 9.751, minus 9.8049 and minus 9.7951, so at nine seconds the rocket is falling at about 9.8 metres per second.
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 115 — Exercises 20-21
Type the blank before checking. The heights at 9 and 9.01 seconds are given, so the only work is a subtraction and a division by one hundredth.
The rocket's height fell by about a tenth of a metre in a hundredth of a second, so the average velocity is minus 9.849 metres per second. At nine seconds the rocket is already on its way down.
If you work out the other three intervals from the exercise, the averages are minus 9.751, minus 9.8049 and minus 9.7951. Two approach from above and two from below, and they pinch minus 9.8. Exercise 21 asks for exactly that estimate.
Sorting
Sort into buckets
Sort each quantity: is it an average over an interval, or a value at one instant?
For each item, ask one question: is there an interval of positive length involved, or a single moment? That decides the bucket every time.
The car trip, the difference quotient over a hundredth of a second, and the sprinter's time are all averages, however short the interval. A secant slope over a tiny interval is still a secant slope.
The speedometer reading, the velocity at exactly half a second, and the tangent slope at five seconds all refer to one instant. Each is defined as a limit of averages. It is striking that a speedometer shows a number every driver understands, yet defining that number properly needs the idea of a limit.
Error analysis
Annotate
On: \( v_{\text{ave}} = \frac{s(0.5) - s(0.49)}{0.49 - 0.5} = \frac{-0.1584}{-0.01} = 15.84 \)
Before revealing the annotations, try to spot why a falling rock came out with a positive velocity. The heights are right, and so is the interval length.
The problem is the order of subtraction. On top, the later time comes first; underneath, the earlier time comes first. Reversing only one of the two differences flips the sign of the whole fraction.
The rule is to subtract in the same order in both places. Later minus earlier over later minus earlier, or earlier minus later over earlier minus later: both give the correct minus 15.84. The quickest safeguard is the one from the step-zero slide: decide the sign before you compute, and distrust any answer that disagrees.
Section
Part 4
Concept
Figure (svg): A curve y = f(x) rising to a peak and falling, with the region between it and the x-axis shaded from x = a to x = b. A question mark sits in the shaded region.
How much area lies between the graph of f and the x-axis from a to b? Areas like this measure real quantities: the work done by a varying force, the distance covered at a varying speed.
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 110 — Figure 2.8
The figure poses the second great problem. The shaded region is bounded by a curve on top, the x-axis underneath, and two vertical lines. What is its area?
Geometry has formulas for regions with straight edges and for circles, and nothing else. A region under a general curve is not on that list, and no amount of cutting it into triangles will make the curved edge straight.
Areas like this are not just geometry exercises. If the curve is a speed graph, the area is the distance travelled; if it is a varying force, the area is the work done. The same mathematics answers all of these, which is why the area problem is worth a whole chapter later in the course.
Concept
Figure (svg): The same curve and interval from a to b, now covered by twelve thin rectangles of equal width, each as tall as the curve at its left edge. Together they nearly fill the region.
Cut the interval into thin strips and stand a rectangle on each, as tall as the curve somewhere in its strip. Add up the rectangles.
\[ \text{area} \approx f(x_1)\,\Delta x + f(x_2)\,\Delta x + \cdots + f(x_n)\,\Delta x \]
As the widths shrink toward zero, the totals approach the area. The limit of such sums is the definite integral, and its study is integral calculus.
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 111 — Figure 2.9
The idea is the same as for the tangent: approximate with something you can compute. Here that means rectangles. Stand a thin rectangle on each strip of the interval, as tall as the curve at some point in the strip, and add up their areas.
In the figure each rectangle takes its height from the left edge of its strip. On the rising part of the curve the rectangles fall a little short; on the falling part they poke a little above. The errors are the thin slivers, and they shrink as the rectangles get narrower.
As the widths approach zero, the totals approach the exact area. That limit is called the definite integral, and it is the subject of Chapter 5. The sum written on the slide is only a preview of the notation.
Worked example
Estimate the area between the x-axis and the parabola below, from 0 to 3, using the three rectangles of Figure 2.10.
\[ f(x) = x^2 + 1, \qquad [0, 3], \qquad \Delta x = 1 \]
Read off each height
Why: Each rectangle is as tall as the curve at its left end.
\[ f(0) = 1, \qquad f(1) = 2, \qquad f(2) = 5 \]
Area of each rectangle
Why: Height times width, and every width is 1.
\[ 1 \cdot 1 = 1, \qquad 2 \cdot 1 = 2, \qquad 5 \cdot 1 = 5 \]
Add
Why: The estimate is the total.
\[ 1 + 2 + 5 = 8 \text{ unit}^2 \]
Figure (svg): The parabola f(x) = x² + 1 over x from 0 to 3, with three rectangles of width 1 whose heights are the function values at their left ends: 1, 2 and 5. Each rectangle's top-left corner touches the curve.
Check that each height is on the curve and which way the error goes
Why: Each top-left corner is a point of the parabola, and because the parabola rises on this interval every rectangle sits below it, so 8 is an underestimate.
\[ 0^2 + 1 = 1, \quad 1^2 + 1 = 2, \quad 2^2 + 1 = 5 \quad\Longrightarrow\quad 8 < \text{area} \]
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 111 — Example 2.3 and Figure 2.10
This example uses just three rectangles, each one unit wide, to estimate the area under the parabola from 0 to 3. With width 1, each rectangle's area equals its height, so the estimate is simply the sum of three function values.
The heights come from the left end of each strip: the function values at 0, 1 and 2, which are 1, 2 and 5. The total is 8 square units.
The check is the important habit. Each rectangle's top-left corner sits on the curve, which confirms the heights. And because the parabola rises across the whole interval, the left end is the lowest point of each strip, so every rectangle lies under the curve. The estimate 8 is therefore too small, and you should say so.
Prediction
\[ f(x) = x^2 + 1 \text{ on } [0, 3], \qquad \text{heights at the right ends} \]
Predict first
Checkpoint 2.3 uses three rectangles whose heights are taken at the RIGHT end of each strip. Compared with the true area, the total will be…
Correct: More than the true area
Why: The parabola rises from 0 to 3, so on each strip the right end is the highest point. A rectangle that tall covers the whole strip under the curve and then some. Every rectangle overshoots, so the total overestimates the area.
Make your prediction by picturing a single strip, say from 1 to 2. The curve climbs from 2 to 5 across it. A rectangle as tall as the right end is 5 units high across the whole strip.
Such a rectangle covers everything under the curve on that strip, plus a wedge above the curve on the left. That is true on every strip because the function rises everywhere on the interval, so every rectangle overshoots and the total is an overestimate.
The width does not change the direction of the error, only its size. Left heights under, right heights over, for any rising function; the reverse for a falling one.
Worked example
Estimate the same area with the three rectangles drawn in Checkpoint 2.3, whose heights are taken at the right ends.
\[ f(x) = x^2 + 1, \qquad [0, 3], \qquad \Delta x = 1 \]
Read off each height
Why: Right ends of the strips: 1, 2 and 3.
\[ f(1) = 2, \qquad f(2) = 5, \qquad f(3) = 10 \]
Area of each rectangle
Why: Width 1 each.
\[ 2 \cdot 1 = 2, \qquad 5 \cdot 1 = 5, \qquad 10 \cdot 1 = 10 \]
Add
Why: An overestimate this time.
\[ 2 + 5 + 10 = 17 \text{ unit}^2 \]
Figure (svg): The same parabola over x from 0 to 3, now with three rectangles whose heights are the function values at their right ends: 2, 5 and 10. Each rectangle's top-right corner touches the curve, and every rectangle pokes above it.
Check against the exact area
Why: Chapter 5 will compute the area exactly as 12. It lies between the two estimates, as the pictures promise.
\[ 8 < 12 < 17 \; \checkmark \]
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, pp. 111-112 — Checkpoint 2.3
The picture in Checkpoint 2.3 takes each height at the right end of its strip. Reading the heights off the grid gives 2, 5 and 10, the function values at 1, 2 and 3, and the total is 17.
Put this next to Example 2.3. The same region, the same number of rectangles, and the estimates are 8 and 17. The true area is somewhere between them, and the two estimates together tell you more than either one alone.
The check uses a value you will be able to compute yourself in Chapter 5: the exact area is 12. It does lie in the bracket. The bracket is wide because three rectangles are crude. The next slides tighten it.
Trap
After Example 2.3:
\[ \text{area} = 1 + 2 + 5 = 8 \]
Wrong. Eight is an estimate, and here a known underestimate.
Rectangles never match a curved top exactly. With left heights on a rising curve the total is too small; with right heights, too big. Write the estimate as a bracket, and remember that only the limit is exact.
\[ 8 < \text{area} < 17 \]
It is easy to finish a rectangle computation and write 'area equals 8'. The equals sign claims too much. A finite collection of rectangles cannot match a curved top, so there is always an error, and here you even know its direction.
The accurate statement is a bracket: the left total is below the area and the right total above it. That is honest, useful, and it tells you how good the estimate is: the true value is within the width of the bracket.
Only the limit of the rectangle totals, as the widths go to zero, is exactly the area. This is the same lesson as the secants: no single approximation is the answer; the answer is what the approximations approach.
Worked example
Halve the width and redo both estimates for the same area.
\[ f(x) = x^2 + 1, \qquad [0, 3], \qquad \Delta x = \tfrac12 \]
Left-end heights
Why: At 0, 0.5, 1, 1.5, 2 and 2.5.
\[ 1, \; 1.25, \; 2, \; 3.25, \; 5, \; 7.25 \]
Left total
Why: Add the heights, then multiply by the width one half.
\[ L_6 = \tfrac12(19.75) = 9.875 \]
Right-end heights
Why: At 0.5, 1, 1.5, 2, 2.5 and 3.
\[ 1.25, \; 2, \; 3.25, \; 5, \; 7.25, \; 10 \]
Right total
Why: Same width.
\[ R_6 = \tfrac12(28.75) = 14.375 \]
Figure (svg): Two panels of f(x) = x² + 1 over x from 0 to 3, each with six rectangles of width one half. Left panel: heights at left ends, all under the curve, total 9.875. Right panel: heights at right ends, all above the curve, total 14.375.
Check the gap with a shortcut
Why: The two lists of heights share every entry but one at each end, so the gap between the totals is the width times the total rise of f, from 1 to 10. That gives 4.5, matching the subtraction.
\[ R_6 - L_6 = \tfrac12\bigl(f(3) - f(0)\bigr) = \tfrac12(9) = 4.5 = 14.375 - 9.875 \; \checkmark \]
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 111 — the rectangles of Figure 2.9, refined
Halving the width means doubling the number of rectangles to six. There are six heights to find for each estimate, and each total is the sum of the heights times the width one half.
The left total rises from 8 to 9.875 and the right total falls from 17 to 14.375. Both moved toward 12, and the bracket shrank from a width of 9 to a width of 4.5.
The check reveals why the gap halved. The left and right lists of heights overlap in every entry except the first height of the left list and the last height of the right list. So the difference between the totals is the width times the total rise of the function, 9. With width one, the gap was 9; with width one half, it is 4.5. Halve the width again and the gap halves again.
Picture it
Figure (svg): Dots for the left-end totals and the right-end totals for f(x) = x² + 1 on [0, 3], using n = 3, 6, 9 up to 30 rectangles. The right totals fall from 17 toward 12, the left totals rise from 8 toward 12, and a dashed line marks 12.
Keep adding rectangles. The left totals climb, the right totals fall, and the gap between them, 9 times the width, shrinks to nothing. The single number they squeeze is the exact area, 12.
The figure plots the left and right totals for more and more rectangles, three at a time up to thirty. The upper dots are right totals, all above 12; the lower dots are left totals, all below 12.
The two sequences squeeze toward each other, because the gap between them is 9 times the width, and the width shrinks toward zero. Whatever the area is, it is caught between every pair of dots, so it can only be 12.
This squeeze is the strongest kind of evidence a table of approximations can give: one sequence approaching from above, one from below, and the gap shrinking to nothing. You will see the same idea formalised as the squeeze theorem in Section 2.3.
Tweak it
Parameter explorer
The graph shows the tops of n left-end rectangles for f(x) = x² + 1 on [0, 3]. Slide n from 1 to 40. What happens to the staircase, and what does its total approach?
\[ L_{n} = \sum_{i=0}^{{n}-1} f\left(\tfrac{3i}{{n}}\right)\cdot\tfrac{3}{{n}} \]
Start at n equal to 1: a single rectangle of height 1, the function's value at 0, which badly underestimates the area. Move to 3 and you see the three steps of Example 2.3.
Keep sliding. The steps get narrower and more numerous, and the staircase starts to look like the parabola itself. The readout shows the sum being computed: n heights, each multiplied by the width three over n.
At forty rectangles the staircase is hard to tell apart from the curve, and the total is within a few tenths of 12. No setting of the slider ever gives exactly 12; only the limit as n grows without bound does. That is the area problem solved by a limit.
Ranking
Put in order
Order these values for the area under x² + 1 on [0, 3], from smallest to largest.
Why: The values are 8, 9.875, 12, 14.375 and 17. Left totals on a rising curve are too small and grow as the rectangles narrow; right totals are too big and shrink. The exact area sits between every left total and every right total.
Drag the five values into order before checking. You do not need to recompute anything; the pictures from the last few slides tell you the order.
Left totals are too small and right totals too big, because the parabola rises. More rectangles means a smaller error, so six left rectangles beat three, and six right rectangles beat three. The exact area lies between every left and every right total.
In numbers: 8, 9.875, 12, 14.375, 17. Notice that the two six-rectangle estimates sit much closer to 12 than the three-rectangle ones, in line with the gap halving when the width halves.
Section
Part 5
Concept
| tangent (and velocity) | area | |
|---|---|---|
| what you can compute | a secant slope, or an average velocity | a total of rectangle areas |
| how you refine it | move x toward a, or shrink the time interval | shrink the rectangle widths |
| what is exact | the value the slopes approach | the value the totals approach |
| its name | the derivative | the definite integral |
Neither answer is ever reached by an approximation. Each is defined as the number the approximations approach: a limit. That is why the rest of Chapter 2 is about limits.
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, pp. 109-111 — taking a limit, in both problems
Read the table column by column, then row by row. The columns are the two problems. The rows are the same three stages in each: something you can compute, a way to make it better, and the exact value those improvements approach.
The tangent slope and the area look like unrelated questions, one about steepness and one about accumulation. But in both, the exact answer is never produced by any single computation. It is defined as the value the approximations approach.
That shared step is the limit, and it is why Chapter 2 exists. Before you can compute derivatives quickly or integrals exactly, you need a precise idea of what 'the value they approach' means and reliable ways to find it.
Matching
Match the pairs
Why: Three of the four quantities are limits; the secant slope is the ordinary computation the tangent slope is built from. Velocity and tangent slope are the same limit with different letters, which is why both lead to the derivative.
Match each quantity before checking. The one to be careful with is the secant slope: it is the only item on the left that is an ordinary finite computation.
The tangent slope is the limit of secant slopes; the instantaneous velocity is the limit of average velocities; the area is the limit of rectangle totals. The secant slope is simply rise over run.
Notice that the first two right-hand entries are the same process with different letters. Velocity is the tangent slope of the position graph, which is why both lead to the derivative. The third is genuinely different, and leads to the integral.
Picture it
Figure (svg): A curved surface z = f(x, y) floating above a square in the xy-plane. Vertical edges join the four corners of the square to the surface, enclosing a solid; the square floor is dashed and the surface is drawn as a mesh.
Nothing forces calculus to stay in two dimensions. A rock fired from a catapult or an aeroplane moves in three, and the area problem has a big sibling: the volume under a surface. That is multivariable calculus, the calculus of functions of two or more variables.
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 112 — Figure 2.11
So far every function has had one input and a graph in the plane. Many real questions need more. The height of a landscape depends on two coordinates; the position of an aeroplane changes in three directions.
The figure shows a surface above a square floor, with the solid between them. Its volume is the area problem one dimension up: instead of thin rectangles under a curve, you would stack thin boxes under a surface and take a limit.
This is multivariable calculus, the subject of Calculus III. It uses exactly the ideas of this course, limits, derivatives and integrals, applied to functions of two or more variables. Everything starts with the one-variable limit you are about to study.
Intuition
Each rectangle estimate is a finite sum. The exact area is what those sums approach as the number of terms grows without bound.
\[ L_3 = 8, \quad L_6 = 9.875, \quad L_{12} = 10.906, \quad \ldots \quad \to 12 \]
Asking what a sum of ever more terms approaches is the starting point of infinite series, one of the main topics of Calculus II. Limits, derivatives, integrals and series all grow from the same idea you met on this page.
OpenStax Calculus Volume 1, §2.1 A Preview of Calculus §2.1, p. 106 — learning objective 2.1.7
Each rectangle estimate is a finite sum, and the exact area is the value that longer and longer sums approach. That is a question about sums with ever more terms.
Push that question one step further and you get infinite series: what does it mean to add infinitely many numbers, and when does the answer make sense? The same limit idea answers it. An infinite sum is defined as the value its partial sums approach.
This is the last learning objective of the section: limits, derivatives and integrals lead naturally to infinite series and to multivariable calculus. You do not need any of that yet. It is here so that you can see where the road goes.
Real world
Discussion prompt
A phone's GPS records your position once a second and displays a speed. A car's odometer adds up distance from the speedometer. Which one is doing the tangent problem, which the area problem, and what approximation is each really making?
Write your answer before revealing. Think about what each device actually measures and what it has to do with that measurement to produce the number on the screen.
A GPS unit only knows your position at separate moments. The speed it displays is the change in position over the last interval divided by the length of the interval: an average velocity, a secant slope. It is a good estimate of your instantaneous speed only because the interval is short.
The odometer does the opposite job. It takes the speed at each moment, multiplies by a short time, and adds the pieces: a rectangle sum under the speed graph. One device is working on the tangent problem, the other on the area problem, and in Chapter 5 you will see they are inverse operations.
Section
Part 6
Pattern
Figure (svg): A two-row flow diagram. Top row, the tangent problem: secant slope through two points, then let x approach a, then the tangent slope. Bottom row, the area problem: total of rectangle areas, then let the width shrink to zero, then the exact area. Both rows pass through a middle column headed 'take a limit'.
This is the procedure behind every computation in the lesson. The diagram shows the tangent problem across the top and the area problem across the bottom, both passing through the same middle box.
When you estimate from a table, read the trend of the whole column, not only the last row. Compute from both sides when you can, because agreement from above and below is much stronger evidence than a one-sided sequence.
When you have a formula, simplify before letting the variable move. The zero over zero at the point itself is not an answer; it is a signal that a factor needs cancelling. Finish with a check: an algebraic formula for the approximation, a units check, or a bracket.
Check
Check your understanding
For f(x) = x², what is the slope of the secant line through (1, 1) and (3, 9)?
Answer: A
Why: Rise over run: nine minus one is eight, three minus one is two, and eight over two is four. It is the average rate of change of x squared between 1 and 3.
The secant slope is rise over run between the two given points: eight over two, which is four.
The most tempting wrong answer is 2, because it is the tangent slope at (1, 1) and you have seen it many times in this lesson. But the question asks about one particular secant, through a point quite far away, and its slope is the average rate of change over the whole stretch from 1 to 3.
Check
Check your understanding
An object has position s(t) = t². What is its average velocity on the interval [1, 1.1]?
Answer: B
Why: The position changes from 1 to 1.21, a change of 0.21, over 0.1 seconds, so the average velocity is 0.21 divided by 0.1, which is 2.1.
The change in position is 1.21 minus 1, which is 0.21, and the interval lasts 0.1 seconds. The average velocity is 2.1.
Two of the wrong answers come from using a position where a change in position belongs. The third, 2, is the instantaneous velocity at t equal to 1, which the averages approach but which no single average equals.
Check
Check your understanding
Estimate the area under f(x) = x² on [0, 2] with four rectangles of equal width, each as tall as f at its RIGHT end.
Answer: C
Why: The width is one half, and the right-end heights are 0.25, 1, 2.25 and 4, which add to 7.5. Times the width one half, the estimate is 3.75, an overestimate since x squared rises.
Four rectangles across an interval of length 2 have width one half. The right-end heights are the squares of 0.5, 1, 1.5 and 2, which are 0.25, 1, 2.25 and 4. Their sum, 7.5, times the width gives 3.75.
The most common slip is forgetting to multiply by the width, which gives 7.5. The other distractors are the left-end estimate, 1.75, and the exact area, about 2.67, which lies between the left and right estimates as it should.
Explain it to yourself
Discussion prompt
In two or three sentences, explain why the slope of the tangent at x = a cannot be found by putting x = a into the secant slope formula, and what you do instead.
Write your answer in your own words before revealing. If you can explain this clearly, you have understood the central idea of the chapter.
A good answer has two parts. The first says why substitution fails: at the point itself there is only one point, so the rise and the run are both zero, and zero over zero gives no information. The second says what to do instead: work with nearby points, where the secant slope is an honest number, and find the value those numbers approach.
If you mentioned simplifying first, cancelling the factor that makes the run zero, you have also anticipated the main technique of Section 2.3.
Exit ticket
\[ f(x) = x^2 + x, \qquad a = 2 \]
Discussion prompt
Find the slope of the secant through x = 2 and x = 2.1; simplify the secant slope for a general x; state the tangent slope at x = 2; and write the tangent line.
This puts the whole tangent method into one problem. Work all four parts before revealing.
The function values are 6 at 2 and 6.51 at 2.1, so the secant slope is 0.51 over 0.1, which is 5.1. For a general x, the numerator x squared plus x minus 6 factors as x minus 2 times x plus 3, and cancelling leaves x plus 3. As x approaches 2 that approaches 5, so the tangent slope is 5.
Point-slope form through (2, 6) with slope 5 gives y equal to 5x minus 4. The simplified formula also checks your first answer: x plus 3 at x equal to 2.1 is 5.1, exactly the secant slope you computed directly.
Recap
| problem | approximation | exact answer |
|---|---|---|
| tangent slope at a | secant slope (f(x) − f(a)) / (x − a) | the value it approaches as x → a: f′(a) |
| velocity at t = a | average velocity (s(t) − s(a)) / (t − a) | the value it approaches as t → a |
| area under f on [a, b] | total of rectangle areas | the value it approaches as the widths → 0 |
\[ f'(a) = \lim_{x \to a} \frac{f(x) - f(a)}{x - a} \]
Next, Section 2.2 makes the phrase 'the value it approaches' precise: the limit of a function, found from tables, graphs and, soon, algebra.
Stewart, Calculus: Early Transcendentals 8e, §2.1 The Tangent and Velocity Problems §2.1, pp. 78-82 — the tangent and velocity problems in Stewart
Three problems, one method. The slope of a curve at a point, the velocity of an object at an instant, and the area under a curve can each be approximated by something you can compute, and each is defined exactly as the value those approximations approach.
For the first two, the approximation is a difference quotient: a change in output over a change in input, a secant slope. For the third, it is a total of rectangle areas. The slope limit is the derivative, and the area limit is the integral.
Section 2.2 now takes the phrase 'the value it approaches' and turns it into a precise idea, the limit of a function, with ways to estimate it from tables and graphs and to recognise when it does not exist.
Want this taught 1-on-1? Alexander tutors Calculus I — $55/session, free consultation.