2.4 Continuity

The three-part definition of continuity at a point and the three ways it fails, the classification of discontinuities as removable, jump or infinite, continuity on an interval with one-sided continuity at endpoints, the algebra of continuous functions and continuity of composites, and the Intermediate Value Theorem as the course's first existence theorem.

Subject: Calculus I · 70 slides · symbolic lesson

Open the interactive version of this deck

What this lesson covers

The lesson, slide by slide

1. Continuity

Title

Calculus I · Section 2.4

When the limit and the value finally agree

2. What this lesson gives you

Objectives

Section 2.3 let you compute limits by substituting, but only for some functions. This lesson names exactly which ones, and what goes wrong for the rest.

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, pp. 155-168 — learning objectives 2.4.1 to 2.4.5

In Section 2.3 you learned to evaluate many limits by plugging in the number. It worked for polynomials, it failed for zero over zero, and it was never quite clear which functions deserve the shortcut. This lesson answers that question with one word, continuity, and turns the word into three checks you can run on any formula.

You will also learn to describe how a function breaks when it does break, because the kind of break tells you whether it can be fixed. Then you will widen the idea from a single point to a whole interval, and learn a theorem that lets a limit slip inside a continuous function.

The lesson ends with the Intermediate Value Theorem. It is the first theorem in the course that proves something exists without telling you what it is, and it is how a calculator finds a root it cannot solve for.

3. Before anything new: when does substitution work?

Warm-up

Discussion prompt

Compute these three limits from Section 2.3. In which of them does plugging in the number give the right answer, and why not in the others?

Write down all three answers and your reason before you reveal. The first limit is a polynomial, and plugging in 3 gives 10 at once.

The second and third are the interesting ones. Plugging in gives zero over zero in both, which is not a number, and yet the limits exist: 4 and 1. So the value at the point and the limit at the point are not always the same thing. Sometimes the value is missing entirely.

That gap between the value and the limit is exactly what this lesson is about. A function for which the two always agree is called continuous, and for such a function substitution is not a trick, it is a theorem.

4. Continuity at a point

Section

Part 1

5. Drawing without lifting the pencil

Picture it

Figure (svg): Two panels. Left: a smooth wavy curve drawn in one stroke across the whole window. Right: a line that stops at x equals 1 with a hollow circle, and restarts higher up with a filled dot, so the pen has to be lifted.

On the left the pen never leaves the paper. On the right it must jump at one point, and that single point is where continuity fails.

The informal idea: a function is continuous if you can trace its graph without lifting your pencil. The rest of this part turns that picture into three checks you can carry out on a formula.

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 155 — introduction

Look at the two panels. On the left you could trace the whole curve in one motion. On the right, at x equal to 1, the graph stops at a hollow circle and starts again higher up, so your pencil has to leave the paper.

That is the intuitive meaning of continuous: no breaks. It is a good picture to carry around, but it is not something you can check on a formula you cannot draw, and it does not say what counts as a break. The next three slides look at three different ways a graph can break at a point, and each one produces a condition.

Notice that continuity is about individual points. The right-hand graph is perfectly unbroken everywhere except at x equal to 1, and that is the only place where anything will fail.

6. Failure one: a hole where the value should be

Concept

Figure (svg): A straight line rising left to right with a hollow circle at x equals a: the graph has a hole and no point is plotted at a.

The line heads for a definite height, but at a itself there is no point. Condition one fails: f(a) is undefined.

The graph heads straight for a point and then there is nothing there. Whatever else continuity means, the function must have a value at the point.

\[ \text{(i) } f(a) \text{ is defined} \]

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 156 — Figure 2.32

Here the graph approaches a definite height from both sides, but at a itself nothing is plotted. The hollow circle marks a missing point. You cannot trace through a point that is not there.

So the first requirement is almost too obvious to state: the function has to have a value at a. If a is outside the domain, because a denominator is zero there or a square root would be of a negative number, then the function cannot be continuous at a, full stop.

Keep in mind that this is only the first gate. Plenty of functions have a value at a and are still broken there, as the next two pictures show.

7. Failure two: the two sides disagree

Concept

Figure (svg): Two line pieces: the left piece ends at x equals a with a filled dot, the right piece starts higher at x equals a with a hollow circle. The pieces do not meet.

The value exists (the filled dot), but the two sides aim at different heights, so there is no single limit. Condition two fails.

Here the value exists, but the graph arrives from the left at one height and leaves to the right from another. So the second requirement is that the two sides agree on a single limit.

\[ \text{(ii) } \lim_{x\to a} f(x) \text{ exists} \]

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 156 — Figure 2.33

This time there is a value at a, the filled dot. But the left piece of the graph arrives at one height and the right piece leaves from another. Walking along the graph from the left, you would reach the dot, and then have to jump up to continue.

In the language of Section 2.2, the left-hand limit and the right-hand limit are different, so the two-sided limit does not exist. That gives the second requirement: the limit at a must exist.

Remember that the limit exists only when both one-sided limits exist and are equal. When the formula for a function changes at a, as it does for a piecewise function, you will usually find the limit by computing the two sides separately.

8. Failure three: the value is in the wrong place

Concept

Figure (svg): A straight line with a hollow circle at x equals a, and a filled dot directly above it, well off the line.

Both the value and the limit exist, and they are different numbers. Condition three fails.

Now there is a value and there is a limit, and they are different numbers. The last requirement ties the two together.

\[ \text{(iii) } \lim_{x\to a} f(x) = f(a) \]

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 156 — Figure 2.34

Now both earlier conditions hold. There is a value at a, the red dot, and the graph approaches a single height from both sides, the hollow circle. Yet the graph is still broken, because the value is not where the graph is heading.

This is the subtle case, and it is why the definition needs a third condition: the limit must equal the value. The limit is built only from inputs near a, never from a itself, so nothing forces it to agree with the value unless you demand it.

Put the three pictures side by side in your head: a hole, a gap, a misplaced dot. Each one fails exactly one of the three conditions, and together they show why none of the three can be dropped.

9. Reading the definition of continuity

Notation

Annotate

On: \( f \text{ continuous at } a \iff \lim_{x\to a} f(x) = f(a) \)

  • Condition (i): the right side must be a number. If a is outside the domain, the test fails before it starts.
  • Condition (ii): the left side must exist, which means both one-sided limits exist and are equal.
  • Condition (iii): the two numbers are the same. This is the only condition that links the value to the approach.
  • Continuity is a property of a function AT a point. A function can be continuous at one point and not at the next.

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 157 — Definition: continuous at a point

The whole definition fits in one equation, and each annotation shows which of the three conditions is hiding in which part of it. For the equation to make sense at all, the right side must be a number, which is condition one, and the left side must be a number, which is condition two. The equals sign is condition three.

Why state three conditions when one equation says it all? Because when continuity fails, you want to say how it fails, and the three conditions fail in genuinely different ways. The book's examples are arranged so that each one fails a different condition.

Notice also the phrase at a. Continuity is checked one point at a time. Later in the lesson you will collect points into intervals, but the basic definition is always about a single input.

10. Continuity is the licence to substitute

Intuition

The limit is built from inputs near a, never at a. The value is the output at a alone. In general those are two unrelated numbers, which is why Section 2.2 kept them apart.

Continuity is precisely the statement that, at this point, they happen to agree. So for a continuous function the hard object, the limit, can be replaced by the easy one, the value:

\[ f \text{ continuous at } a \;\Longrightarrow\; \lim_{x\to a} f(x) = f(a) \]

Every time you evaluated a limit by plugging in, you were quietly using continuity. From now on you will say so.

This slide reframes something you already do. The limit at a and the value at a are computed from completely different information: the limit from nearby inputs, the value from a alone. There is no reason in general for them to agree.

Continuity is the special situation in which they do agree. When a function is continuous at a, the limit, which is hard to compute, equals the value, which is easy. That is exactly what you were doing when you evaluated a polynomial limit by plugging in.

From now on, when you substitute into a limit, you should be able to say why it is allowed: because the function is continuous at that point. The next part of the lesson gives you theorems that establish continuity for whole families of functions at once, so the justification is usually one short phrase.

11. The three checks, in order

Concept

Figure (svg): A flow diagram of three checks in a row: is f(a) defined, does the limit exist, does the limit equal f(a). A yes moves right to the next check; a no at any check drops down to 'not continuous'. Three yeses reach 'continuous'.

Stop at the first no. Each check is only worth doing once the one before it has passed.

The book's problem-solving strategy is the definition run in order, stopping at the first failure. When the formula changes at a, find the limit through its two one-sided limits.

\[ \lim_{x\to a^-} f(x) = \lim_{x\to a^+} f(x) \;\Longrightarrow\; \lim_{x\to a} f(x) \text{ exists} \]

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 157 — Problem-Solving Strategy: Determining Continuity at a Point

Follow the arrows in the diagram. You check the conditions in order, and the first no ends the job: the function is not continuous at a, and you name the condition that failed. Only if all three answers are yes is the function continuous.

The order matters for efficiency. There is no point computing a limit to compare with a value that does not exist, and no point comparing a limit that does not exist with anything.

The second check is where the work usually is. When the formula for the function changes at a, as it does at the seam of a piecewise rule, compute the limit from the left and the limit from the right separately, then see whether they agree. The next four worked examples run this procedure on four different functions, and each one stops at a different place.

12. Example 2.26: zero over zero at x = 2

Worked example

Using the definition, decide whether this function is continuous at 2.

\[ f(x) = \frac{x^2-4}{x-2}, \qquad a = 2 \]

Check condition (i): try to compute f(2)

Why: Substitute 2 into the numerator and the denominator separately.

\[ f(2) = \frac{2^2 - 4}{2 - 2} = \frac{0}{0} \]

Read the result

Why: Zero over zero is not a number, so f(2) is undefined.

\[ f(2) \text{ undefined} \]

Conclude

Why: Condition (i) fails; there is no need to look at conditions (ii) or (iii).

\[ f \text{ is discontinuous at } 2 \]

Figure (svg): The graph of x squared minus 4 over x minus 2: the straight line y equals x plus 2, with a hollow circle at the point (2, 4).

Everywhere except 2 the graph is the line y = x + 2. At 2 the formula gives 0/0, so the point is missing.

Check against the graph

Why: Cancelling the common factor shows the graph is the line x + 2 with the point above 2 missing, exactly a hole.

\[ \frac{(x-2)(x+2)}{x-2} = x + 2 \quad (x \ne 2) \]

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 157 — Example 2.26 and Figure 2.35

This example stops at the first check. When you substitute 2, both the top and the bottom become zero, and zero over zero is not a number. So the function has no value at 2 and cannot be continuous there.

You might object that the limit at 2 exists, and indeed it does, as you saw in Section 2.3. But continuity needs the value too. The limit existing is not enough on its own.

The check at the end connects the verdict to the picture. Away from 2, the fraction simplifies to x plus 2, so the graph is a straight line with a single point missing. That is exactly the hole from the first failure picture. You will come back to this function in Part 2 to say what kind of break it is.

13. Example 2.27: a piecewise function at its seam

Worked example

Using the definition, decide whether this function is continuous at 3.

\[ f(x) = \begin{cases} -x^2 + 4 & x \le 3 \\ 4x - 8 & x > 3 \end{cases} \]

Check condition (i)

Why: At 3 itself the first line applies, because it includes x equal to 3.

\[ f(3) = -(3^2) + 4 = -5 \]

Compute the limit from the left

Why: Inputs just below 3 use the first line.

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

Compute the limit from the right

Why: Inputs just above 3 use the second line.

\[ \lim_{x\to 3^+} f(x) = 4(3) - 8 = 4 \]

Compare the two sides

Why: They disagree, so the two-sided limit does not exist and condition (ii) fails.

\[ -5 \ne 4 \;\Longrightarrow\; \lim_{x\to 3} f(x) \text{ DNE} \]

Figure (svg): A downward parabola, minus x squared plus 4, drawn up to x equals 3 where it ends in a filled dot at height minus 5, and a steep line 4x minus 8 starting at x equals 3 with a hollow circle at height 4.

The left piece arrives at minus 5 and the right piece leaves from 4. Different one-sided limits: the two-sided limit does not exist.

Check with nearby values

Why: Just below 3 the function is about minus 5; just above, about 4. A jump of 9 in an arbitrarily short step.

\[ f(2.99) = -4.9401, \qquad f(3.01) = 4.04 \]

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 158 — Example 2.27 and Figure 2.36

Piecewise functions are where continuity is most often tested, because the seam is the one place a break can hide. Here the value at 3 comes from the first line, since that line includes 3, and it is minus 5. So the first check passes.

For the limit you need the two sides separately, because different formulas apply on each side. From the left the parabola heads to minus 5. From the right the line heads to 4. They disagree, so the two-sided limit does not exist and the second check fails.

The picture shows the parabola ending in a filled dot and the line starting from a hollow circle nine units higher. The numerical check makes the jump vivid: move from 2.99 to 3.01, a step of two hundredths, and the output leaps by about nine.

14. Example 2.28: sine x over x, patched at 0

Worked example

Using the definition, decide whether this function is continuous at 0.

\[ f(x) = \begin{cases} \dfrac{\sin x}{x} & x \ne 0 \\ 1 & x = 0 \end{cases} \]

Check condition (i)

Why: The second line defines the value at 0.

\[ f(0) = 1 \]

Check condition (ii)

Why: Near 0 the first line applies, and this is the special trigonometric limit from Section 2.3.

\[ \lim_{x\to 0} f(x) = \lim_{x\to 0}\frac{\sin x}{x} = 1 \]

Check condition (iii)

Why: Compare the two numbers.

\[ f(0) = 1 = \lim_{x\to 0} f(x) \]

Conclude

Why: All three conditions hold.

\[ f \text{ is continuous at } 0 \]

Figure (svg): The graph of sine x over x from minus 10 to 10, a central hump of height 1 with decaying ripples on each side, and a filled dot exactly at (0, 1) sitting on the curve.

The patch value f(0) = 1 lands exactly where the curve was heading, so there is no hole left: the function is continuous at 0.

Check numerically

Why: Values near 0 are already within two thousandths of the patch value 1.

\[ \frac{\sin 0.1}{0.1} \approx 0.99833, \qquad \frac{\sin 0.01}{0.01} \approx 0.99998 \]

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 159 — Example 2.28

This one passes all three checks, which makes it a good model of a complete continuity argument. The value at 0 is 1, because the second line of the definition says so. The limit at 0 is the famous sine x over x limit from Section 2.3, which is also 1. The two agree.

It is worth seeing why this function was defined in two pieces. Sine x over x on its own is undefined at 0, so it would fail the first check. Someone chose the value at 0 to be exactly the limit, and that choice filled the hole. You will meet this idea again when you learn about removable discontinuities.

The graph confirms it: the patch point sits right on top of the central hump, with no gap. The numbers confirm it too, creeping up to 1 from below as x shrinks.

15. Checkpoint 2.21: a value set below the peak

Worked example

Decide whether f is continuous at 1; if not, name the condition that fails.

\[ f(x) = \begin{cases} 2x + 1 & x < 1 \\ 2 & x = 1 \\ -x + 4 & x > 1 \end{cases} \]

Check condition (i)

Why: The middle line gives the value at 1.

\[ f(1) = 2 \]

Left-hand limit

Why: Inputs below 1 use the first line.

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

Right-hand limit

Why: Inputs above 1 use the third line.

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

Condition (ii) holds

Why: The sides agree.

\[ \lim_{x\to 1} f(x) = 3 \]

Condition (iii) fails

Why: The limit and the value are different numbers.

\[ 3 \ne 2 = f(1) \]

Figure (svg): A peaked graph: the line 2x plus 1 rising to a hollow circle at (1, 3), the line minus x plus 4 falling away from the same hollow circle, and a filled dot at (1, 2) just below the peak.

The two pieces meet, at height 3, but the function's value at 1 was set to 2. Limit and value disagree.

Check the diagnosis

Why: Change only the value to f(1) = 3 and all three conditions hold, which confirms that condition (iii) alone was at fault.

\[ f(1) := 3 \;\Longrightarrow\; \lim_{x\to 1} f(x) = 3 = f(1) \]

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 159 — Checkpoint 2.21

Before you look at the steps, try running the three checks yourself. The middle line gives the value at 1, which is 2, so the first check passes. The two outer lines give the one-sided limits, and both come out to 3, so the limit exists and the second check passes.

The third check is where it fails: the limit is 3 and the value is 2. The graph has a peak whose tip has been moved down by one unit, leaving a hollow circle where the tip should be and a stray dot below it.

The check step confirms the diagnosis in a useful way. If you change only the value at 1 to 3, every condition holds. So the entire problem was the value, not the approach. That observation is the heart of the removable discontinuity, which is coming in the next part.

16. Trap: it is defined there, so it is continuous

Trap

The trap

A common one-line answer to Checkpoint 2.21:

\[ f(1) = 2 \text{ exists} \]

\[ \Longrightarrow f \text{ continuous at } 1 \]

Wrong. Condition (i) is one check out of three.

The fix

Having a value only gets you past the first check. The graph still has to arrive at that value from both sides. Here both sides arrive at 3, not 2.

\[ \lim_{x\to 1} f(x) = 3 \ne 2 \]

Defined is necessary, never sufficient.

This is the mistake of stopping after the first check. It is tempting because a function that is defined at a point feels complete there. But being defined only means there is a dot on the graph. It says nothing about whether the rest of the graph arrives at that dot.

In Checkpoint 2.21 the rest of the graph arrives at height 3 from both sides, and the dot is at height 2. Continuity fails. Whenever you write that a function is continuous, make sure your argument mentions a limit and compares it with the value. An argument that never mentions a limit cannot establish continuity.

17. Which condition fails first?

Sorting

Sort into buckets

Run the three checks in order at the point named. Sort each function by the FIRST condition that fails.

(i) f(a) undefined
(x² − 9)/(x − 3) at x = 3; 1/x at x = 0
(ii) no limit
|x|/x at x = 0, with f(0) defined as 1; x + 1 for x < 2, x − 1 for x ≥ 2, at x = 2
(iii) limit ≠ f(a)
x² for x ≠ 1, f(1) = 3, at x = 1; x² for x ≠ 2, f(2) = 0, at x = 2
i
(x² − 9)/(x − 3) and 1/x both have zero denominators at the point, so there is no value and the checks stop at once.
ii
Both have values at the point, but the sides disagree: |x|/x is −1 on the left of 0 and 1 on the right; at 2, x + 1 arrives at 3 while x − 1 leaves from 1.
iii
x² heads for 1 at x = 1 and for 4 at x = 2, but the values were set to 3 and 0: the limit exists and misses the value.

Work through the checks in order for each function and sort by the first check that fails. For each item, start by asking whether the formula can be evaluated at the point at all.

Two of the functions have a zero denominator at the point, so the first check already fails and you can stop. Two have a value, but the sides disagree: the absolute value of x over x is minus 1 on the left of 0 and 1 on the right, and the piecewise rule at 2 arrives at 3 from the left and leaves from 1 on the right.

The last two have a value and a limit that exists, but the value was set to something else. Those fail the third check. Notice how the same function, x squared, can fail at 1 or at 2 depending only on which single value has been tampered with.

18. Where functions break, and how

Section

Part 2

19. Polynomials and rational functions come continuous

Concept

Section 2.3 proved that you may substitute into a polynomial, and into a rational function wherever its denominator is not zero. In the new language, that is a continuity theorem.

\[ \lim_{x\to a} p(x) = p(a) \quad \text{for every polynomial } p \]

\[ \lim_{x\to a}\frac{p(x)}{q(x)} = \frac{p(a)}{q(a)} \quad \text{whenever } q(a) \ne 0 \]

Theorem 2.8 — Polynomials and rational functions are continuous at every point in their domains.

So for these functions the only candidates for a break are the zeros of the denominator. Everywhere else, no checking is needed.

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 160 — Theorem 2.8 and its proof

Here is the payoff of the limit laws. Section 2.3 proved that the limit of a polynomial at any point is its value there. Read in the language of this lesson, that is precisely the statement that every polynomial is continuous everywhere.

The same goes for a rational function, a polynomial over a polynomial, wherever the bottom is not zero. The limit of the quotient is the quotient of the values, which is the value of the quotient. So a rational function is continuous at every point of its domain.

The practical meaning is that you never need to run the three checks at an ordinary point of a polynomial or rational function. The only places worth examining are the zeros of the denominator. That turns the question of where a rational function is continuous into the much easier question of where its denominator vanishes.

20. Example 2.29: where a rational function breaks

Worked example

For what values of x is the function continuous?

\[ f(x) = \frac{x+1}{x-5} \]

Find where the denominator vanishes

Why: Set the bottom equal to zero.

\[ x - 5 = 0 \iff x = 5 \]

Apply Theorem 2.8

Why: A rational function is continuous on its domain, which is every real number except 5.

\[ f \text{ continuous for all } x \ne 5 \]

Figure (svg): The graph of x plus 1 over x minus 5: a hyperbola with a dashed vertical asymptote at x equals 5 and a horizontal asymptote at y equals 1; the curve is unbroken on each side of 5.

One break, at the one input outside the domain. Every other point of the graph is traced without lifting the pen.

Checkpoint 2.22 the same way

Why: A polynomial has no denominator to vanish.

\[ f(x) = 3x^4 - 4x^2: \quad \text{continuous for all real } x \]

Check the break at 5

Why: Just either side of 5 the function is huge and of opposite sign, so no limit exists there, confirming a genuine break.

\[ f(4.99) = -599, \qquad f(5.01) = 601 \]

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 160 — Example 2.29 and Checkpoint 2.22

With Theorem 2.8 in hand, this example takes two lines. The denominator vanishes only at 5, so the function is continuous at every other real number. No limits need to be computed.

The graph shows what that means. There is a vertical asymptote at 5, and on each side of it the curve is traced without lifting the pencil. The theorem does not say the function is continuous at 5; it is not even defined there.

Checkpoint 2.22 is the same idea with no denominator at all: a polynomial like three x to the fourth minus four x squared is continuous on the entire real line. The check at the end looks at values just either side of 5. They are huge and of opposite sign, so there is certainly no limit there, which confirms that 5 is a genuine break and not a removable one.

21. Three kinds of break

Picture it

Figure (svg): Three panels. Removable: a line with a hollow circle at a and a filled dot above it. Jump: a line ending in a filled dot at a and a second line starting higher with a hollow circle. Infinite: a curve shooting up to plus infinity on the left of a dashed asymptote at a and coming up from minus infinity on its right.

A hole can be filled with one point. A step and an asymptote cannot: the trouble is in how the graph approaches a, not in the value at a.

A discontinuity is described by what the graph does near the point. A hole, where the sides agree; a step, where they settle at different finite heights; an asymptote, where at least one side runs off to infinity.

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 160 — Figure 2.37

The book sorts the breaks you have seen into three types, and this picture shows one of each. On the left, a removable discontinuity: the graph has a hole, and the dot above it is a value in the wrong place. In the middle, a jump: the graph arrives at one height and leaves from another. On the right, an infinite discontinuity: a vertical asymptote.

What separates them is what the graph does as it approaches a, never what happens exactly at a. In the removable case the two sides agree on a height. In the jump case they settle on two different finite heights. In the infinite case at least one side does not settle at all but runs off to infinity.

The book adds a warning worth remembering: these are the three common types, not the only possible ones. You will see a function that fits none of them later in this part.

22. Reading the three definitions

Notation

Annotate

On: \( \lim_{x\to a} f(x) = L, \qquad \lim_{x\to a^-} f(x) \ne \lim_{x\to a^+} f(x), \qquad \lim_{x\to a^\pm} f(x) = \pm\infty \)

  • The two-sided limit exists, a real number L. The function is broken only because the value at a is missing or wrong.
  • Both one-sided limits exist as real numbers, finite on both sides, but they are different.
  • At least one one-sided limit is plus or minus infinity: the graph has a vertical asymptote at a. The word 'and/or' matters; one side is enough.
  • All three describe a function already known to be discontinuous at a. A continuous function has no 'kind'.

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 161 — Definition: removable, jump and infinite discontinuities

Step through the annotations. Each definition is stated in terms of limits, which is the key point: you classify a discontinuity by computing limits, not by looking at the value at a.

Removable means the two-sided limit exists as a real number. The function is discontinuous only because the value is missing or does not match. Jump means both one-sided limits exist as real numbers, finite on both sides, but disagree. Infinite means at least one one-sided limit is plus or minus infinity. The phrase and or matters: a single infinite side is already enough.

The last annotation is easy to miss. These words describe a function already known to be discontinuous at a. If the function is continuous there, it does not have a type of discontinuity at all.

23. Example 2.30: the hole at 2 is removable

Worked example

Example 2.26 showed this function is discontinuous at 2. Classify the discontinuity.

\[ f(x) = \frac{x^2-4}{x-2} \]

Set up the limit at 2

Why: Classification is decided by the limit, not the value.

\[ \lim_{x\to 2} f(x) = \lim_{x\to 2}\frac{x^2-4}{x-2} \]

Factor the numerator

Why: A difference of squares.

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

Cancel the common factor

Why: Legal because the limit never uses x equal to 2 itself.

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

Substitute

Why: x + 2 is a polynomial, continuous at 2.

\[ = 4 \]

Classify

Why: The function is discontinuous at 2 and the limit exists.

\[ \text{removable discontinuity at } x = 2 \]

Figure (svg): Dots at inputs 1.5, 1.8, 1.9, 1.99 on the left and 2.01, 2.1, 2.2, 2.5 on the right, plotted on the line y equals x plus 2; from both sides their heights crowd in on 4, while x equals 2 itself has only a hollow circle.

Both sides close in on the same height. A limit that exists, at a point where the value does not: a removable discontinuity.

Check with values from both sides

Why: Inputs a hundredth either side of 2 give outputs a hundredth either side of 4.

\[ f(1.99) = 3.99, \qquad f(2.01) = 4.01 \]

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 161 — Example 2.30

Example 2.26 showed that this function is broken at 2 because its value is missing. To say what kind of break it is, you compute the limit, and that is the familiar factor-and-cancel from Section 2.3.

The cancellation deserves a moment's thought. You may divide out x minus 2 because a limit at 2 never uses x equal to 2 itself, so x minus 2 is never zero in the calculation. After cancelling, you are left with x plus 2, a polynomial, and continuity lets you substitute. The limit is 4.

Since the function is discontinuous at 2 and the limit exists, the discontinuity is removable. The picture shows why the name fits: inputs creeping in from either side produce outputs creeping in on 4, and the only thing missing is a single point at height 4.

24. Example 2.31: what kind of break at 3?

Prediction

\[ f(x) = \begin{cases} -x^2 + 4 & x \le 3 \\ 4x - 8 & x > 3 \end{cases} \]

Predict first

Example 2.27 found the one-sided limits at 3: minus 5 from the left and 4 from the right. What kind of discontinuity is this?

  • Removable
  • Jump
  • Infinite
  • It cannot be classified

Correct: Jump

Why: Both one-sided limits exist as real numbers, minus 5 and 4, and they are different. That is exactly the definition of a jump discontinuity. It is not removable, because the two-sided limit does not exist, and it is not infinite, because neither side runs off to infinity.

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 161 — Example 2.31

Make your choice before revealing. All the work was done in Example 2.27; this question only asks you to read the one-sided limits against the three definitions.

Both sides settle on real numbers, minus 5 and 4, and those numbers are different. That is the definition of a jump discontinuity, word for word. It cannot be removable, because removable requires the two-sided limit to exist, and it does not. It cannot be infinite, because neither side runs off to infinity.

Notice that the value at 3, which is minus 5, played no part in the classification. The type of a discontinuity is decided entirely by the limits.

25. Example 2.32: an asymptote at x = −1

Worked example

Decide whether f is continuous at minus 1, and classify any discontinuity.

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

Check condition (i)

Why: The denominator is zero at minus 1.

\[ f(-1) = \frac{1}{0} \quad \text{undefined} \]

Conclude discontinuity

Why: Condition (i) fails.

\[ f \text{ is not continuous at } -1 \]

Limit from the left

Why: The top tends to 1; the bottom is a small negative number.

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

Limit from the right

Why: The top tends to 1; the bottom is a small positive number.

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

Classify

Why: At least one side is infinite (here both are).

\[ \text{infinite discontinuity at } x = -1 \]

Figure (svg): The graph of x plus 2 over x plus 1 with a dashed vertical asymptote at x equals minus 1: to its left the curve plunges down toward minus infinity, to its right it shoots up toward plus infinity.

Near minus 1 the top is close to 1 while the bottom shrinks to zero, negative on the left and positive on the right.

Check with values

Why: A step of a hundredth either side of minus 1 gives outputs near minus 99 and plus 101.

\[ f(-1.01) = -99, \qquad f(-0.99) = 101 \]

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 161 — Example 2.32

Here the first check fails immediately: at minus 1 the denominator is zero and the numerator is 1, so there is no value. That settles discontinuity. The classification needs the one-sided limits.

Near minus 1 the top is close to 1, a positive number, while the bottom is close to zero. Just to the left of minus 1 the bottom is a small negative number, so the quotient is a large negative number, heading to minus infinity. Just to the right the bottom is small and positive, so the quotient heads to plus infinity.

Both sides are infinite, so this is an infinite discontinuity, drawn as the vertical asymptote in the picture. Compare this with Example 2.26: both have a zero denominator at the point, but there the numerator was zero too, and the factors cancelled. A nonzero number over something shrinking to zero always blows up.

26. Checkpoint 2.23: a value set too high

Fill the middle

\[ f(x) = \begin{cases} x^2 & x \ne 1 \\ 3 & x = 1 \end{cases} \]

Fill in the blanks

The limit of f as x approaches 1 is 1, but f(1) is 3, so f is not continuous at 1. The kind of discontinuity is removable.

Why: Near 1 the function is x squared, which heads for 1. The value was set to 3, so condition (iii) fails. The two-sided limit exists, so the break is removable: redefining f(1) as 1 repairs it.

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 162 — Checkpoint 2.23

Fill in all three blanks before checking. Away from 1 the function is x squared, so as x approaches 1 the outputs approach 1. That is the limit.

The value at 1 was set separately, to 3. So the limit exists, but it does not equal the value, and the third condition fails. Because the limit exists, the break is removable.

It is worth saying what the repair would be: define the value at 1 to be 1 instead of 3, and the function becomes x squared everywhere, which is continuous. A removable discontinuity is always fixed by the same move, setting the value equal to the limit.

27. Match the one-sided limits to the break

Matching

Match the pairs

  • a. left → 2, right → 2, f(a) = 2
  • b. left → 2, right → 2, f(a) = 5
  • c. left → 2, right → 5
  • d. left → 2, right → +∞
  • w. continuous at a
  • x. removable discontinuity
  • y. jump discontinuity
  • z. infinite discontinuity

Why: Agreeing sides that also match the value: continuous. Agreeing sides that miss the value: removable. Finite sides that disagree: jump. One infinite side is already enough for an infinite discontinuity, even though the other side is a tame 2.

Each row gives you the information you would have after computing the one-sided limits, and sometimes the value. Pair each with its verdict before checking.

The first two rows have agreeing sides, both heading to 2. The difference is the value: when the value is also 2, the function is continuous; when the value is 5, the function has a removable discontinuity. The third row has finite sides that disagree, a jump.

The fourth row is the one that catches people. The left side is a perfectly tame 2, and only the right side runs off to infinity. That is still an infinite discontinuity, because the definition needs only one infinite side.

28. Find the error: a jump declared removable

Error analysis

Annotate

On: \( f(x) = \frac{|x-1|}{x-1}: \; f(1) \text{ undefined} \Rightarrow \text{removable; set } f(1) = 0 \)

  • True, and it shows f is discontinuous at 1. But a missing value says nothing about the KIND of break.
  • For x below 1, |x − 1| = −(x − 1), so f(x) = −1. The left-hand limit is −1.
  • For x above 1, |x − 1| = x − 1, so f(x) = 1. The right-hand limit is 1.
  • The sides are −1 and 1: a JUMP. No value at 1 can repair it; f(1) = 0 sits between the two pieces and matches neither.

Read the claim, then step through the annotations. The first observation is correct: the function has no value at 1, so it is discontinuous there. The error is in jumping from a missing value to the word removable.

To classify, you must compute the one-sided limits. Below 1, the quantity inside the absolute value is negative, so the absolute value flips its sign and the fraction equals minus 1. Above 1, the absolute value does nothing and the fraction equals 1. The graph is two flat pieces at heights minus 1 and 1.

Those sides disagree, so the break is a jump, and no value at 1 can repair it. The proposed value 0 sits halfway between the two pieces and agrees with neither. The general lesson: a missing value tells you there is a break, never what kind.

29. A break that fits none of the three boxes

Counterexample

Figure (svg): The graph of sine of one over x for x between minus 0.5 and 0.5: it oscillates between minus 1 and 1 faster and faster as x approaches 0, filling a vertical band at the origin.

No hole, no step, no asymptote: the graph never settles on any height from either side. Neither one-sided limit exists, and the values stay between minus 1 and 1.

Discussion prompt

The book warns that not every discontinuity is removable, jump or infinite. Using the graph, explain why sine of one over x (with any value you like at 0) is discontinuous at 0 but fits none of the three kinds.

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 160 — 'not all discontinuities fit neatly into these categories'

Write your explanation before revealing. The graph is the key evidence. As x shrinks toward 0, one over x grows without bound, so the sine completes more and more oscillations in less and less space. Near 0 the graph becomes a blur between minus 1 and 1.

The two sequences in the answer make the blur precise. Both sequences of inputs shrink to 0 from the right, but along one the function is always 1 and along the other it is always minus 1. So no single height is being approached, and the right-hand limit does not exist. The left side behaves the same way.

Now test each box. Removable needs the limit to exist; it does not. Jump needs finite one-sided limits; they do not exist. Infinite needs a side to run off to infinity; the values never leave the band from minus 1 to 1. So this discontinuity is of none of the three kinds, just as the book warned.

30. Why only a hole can be repaired

Intuition

To repair a break you are allowed to change one thing: the single value f(a). The limit is computed from inputs other than a, so no choice of f(a) can change it.

\[ \text{removable: set } f(a) = \lim_{x\to a} f(x) \]

For a jump, the two sides are fixed at different heights, and whatever height you choose for f(a) must disagree with at least one of them. For an infinite break there is no finite height to aim at at all.

\[ \text{Example 2.30 repaired: } f(2) := 4 \;\Longrightarrow\; f(x) = x + 2 \text{ everywhere} \]

This is the reason the classification is worth doing. To repair a function at a point you are allowed to change exactly one thing, the value at that point. But the limit is computed without ever looking at that value, so no choice of value can change the limit.

If the limit exists, you set the value equal to it and the function becomes continuous. Example 2.30 is repaired by setting the value at 2 to 4, and then the function is simply x plus 2.

If the sides disagree, as in a jump, then whatever height you pick for the value, at least one side is heading somewhere else. And if a side runs off to infinity, there is no finite height to aim at at all. So removable discontinuities are exactly the ones a single value can fix.

31. Continuity over an interval

Section

Part 3

32. Continuous from one side

Concept

Figure (svg): A step at x equals 2: the left piece rises to a hollow circle at height 1.5; the right piece starts with a filled dot at height 3 and continues from there. The filled dot belongs to the right piece.

The value f(a) is where the right-hand piece starts, so the right-hand limit equals f(a) and the left-hand limit does not. Continuous from the right, not from the left.

Sometimes only one side of the graph reaches the value. One-sided limits give one-sided continuity.

\[ \text{from the right at } a: \quad \lim_{x\to a^+} f(x) = f(a) \]

\[ \text{from the left at } a: \quad \lim_{x\to a^-} f(x) = f(a) \]

A function is continuous at a exactly when it is continuous from both sides there.

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 162 — Definition: continuity from the right and from the left

In the picture, the filled dot at a belongs to the right-hand piece. Coming from the right, the graph runs straight into that dot. Coming from the left, the graph arrives at the hollow circle, well below it.

One-sided limits let you describe this precisely. If the right-hand limit equals the value, the function is continuous from the right at a. If the left-hand limit equals the value, it is continuous from the left. This function is continuous from the right at a, and not from the left.

Being continuous at a in the ordinary sense is the same as being continuous from both sides at once, since that forces both one-sided limits to equal the value, and therefore each other. One-sided continuity on its own becomes important at the ends of an interval, which is the next slide.

33. What continuous on an interval means

Concept

Figure (svg): A curve drawn only over the interval from a equals 1 to b equals 4, with filled dots at both ends and nothing outside; arrows at a point rightward and at b point leftward to show the one side each endpoint uses.

Inside the interval every point needs the full two-sided test. At the ends there is nothing on the outside to approach from, so only the inward side is asked for.

continuous over an interval — On an open interval: continuous at every point of it. On a closed interval from a to b: continuous at every point strictly inside, continuous from the right at a, and continuous from the left at b.

\[ \lim_{x\to a^+} f(x) = f(a), \qquad \lim_{x\to b^-} f(x) = f(b) \]

Half-open intervals mix the two: continuity inside, plus one-sided continuity at whichever endpoint is included. The endpoint conditions guarantee that the pencil starts on the point above a and finishes on the point above b.

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 162 — continuity over an interval

On an open interval the definition is what you would expect: continuous at every point of it. The interesting case is a closed interval, one that includes its endpoints.

At an endpoint, the function may not exist on the outside at all, as the picture suggests: there is simply no graph to the left of a or to the right of b. So it would be unreasonable to ask for a two-sided limit there. Instead you ask for continuity from the inside only: from the right at a, and from the left at b.

The book puts the purpose nicely. These endpoint conditions make sure you can trace the graph from the point above a to the point above b without lifting your pencil. Without them, the graph could stop short of its endpoint value and jump to it at the last moment. Half-open intervals follow the same pattern at whichever end is included.

34. Example 2.33: the intervals of a rational function

Worked example

State the intervals over which the function is continuous.

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

Factor the denominator

Why: Common factor x.

\[ x^2 + 2x = x(x+2) \]

Find the zeros of the denominator

Why: Each factor gives one.

\[ x(x+2) = 0 \iff x = 0 \text{ or } x = -2 \]

Write the domain

Why: Every real number except those two.

\[ (-\infty, -2) \cup (-2, 0) \cup (0, +\infty) \]

Apply Theorem 2.8

Why: A rational function is continuous on its domain, so on each of these open intervals.

\[ \text{continuous on } (-\infty, -2), \; (-2, 0), \; (0, +\infty) \]

Figure (svg): The graph of x minus 1 over x squared plus 2x, with dashed vertical asymptotes at x equals minus 2 and x equals 0, cutting the graph into three unbroken pieces.

Three pieces, three intervals of continuity. The two inputs that make the denominator zero are the only breaks.

Check that neither break is removable

Why: The numerator is not zero at either point, so near each the bottom shrinks while the top stays near a nonzero number: the values blow up.

\[ f(-1.99) \approx 150.3, \qquad f(0.01) \approx -49.3 \]

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 162 — Example 2.33

This is Theorem 2.8 put to work. A rational function is continuous on its domain, so the only job is to find the domain, which means finding where the denominator vanishes.

Factoring the denominator as x times x plus 2 shows it is zero at 0 and at minus 2. Removing those two numbers splits the real line into three open intervals, and the function is continuous on each one. The picture shows three unbroken pieces separated by two asymptotes.

The check at the end confirms that neither break is removable. The numerator, x minus 1, is not zero at either point, so near each point you have a nonzero number over something shrinking to zero, and the values blow up. Just to the right of minus 2 the function is already above 150, and just to the right of 0 it is below minus 49.

35. Example 2.34: the upper semicircle on a closed interval

Worked example

State the interval over which the function is continuous.

\[ f(x) = \sqrt{4 - x^2} \]

Find the domain

Why: The expression under the root must not be negative.

\[ 4 - x^2 \ge 0 \iff -2 \le x \le 2 \]

Interior points

Why: For a strictly between minus 2 and 2, the limit laws let you substitute into the root.

\[ \lim_{x\to a}\sqrt{4-x^2} = \sqrt{4-a^2} = f(a) \]

Left endpoint, from the right only

Why: There is no function to the left of minus 2 to approach from.

\[ \lim_{x\to -2^+}\sqrt{4-x^2} = 0 = f(-2) \]

Right endpoint, from the left only

Why: Likewise at 2.

\[ \lim_{x\to 2^-}\sqrt{4-x^2} = 0 = f(2) \]

Conclude

Why: Continuous inside and one-sided continuous at both ends.

\[ f \text{ is continuous on } [-2, 2] \]

Figure (svg): The upper half of the circle of radius 2 centred at the origin, the graph of the square root of 4 minus x squared, with filled dots at its ends (minus 2, 0) and (2, 0).

At the ends the graph reaches the axis from one side only, and the one-sided limits there match the values 0. That is what makes the interval closed.

Check at a point near the end

Why: Close to 2 from the inside, the value is already small, heading to 0.

\[ f(1.99) = \sqrt{4 - 3.9601} = \sqrt{0.0399} \approx 0.1997 \]

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 162 — Example 2.34

The graph is the upper half of a circle of radius 2, so it runs from minus 2 to 2 and stops. That is why the natural answer is a closed interval, and it is a good example of the endpoint rule doing its job.

Strictly inside the interval, the limit laws for roots let you substitute, so the function is continuous at every interior point. At the endpoints there is no graph outside the interval, so you only need the inside limits. From the right at minus 2 and from the left at 2, the expression under the root shrinks to zero, and so does the function, matching the values.

The numerical check shows the approach to the right endpoint: at 1.99 the height is already only about 0.2, and it continues down to exactly 0 at 2. There is no jump at the end of the arc.

36. Checkpoint 2.24: a root with one endpoint

Fill the middle

\[ f(x) = \sqrt{x + 3} \]

Fill in the blanks

The root needs x + 3 to be at least zero, so the domain starts at x = -3. There f is continuous from the right only, and f is continuous on the interval [-3, ∞).

Why: x + 3 is at least zero exactly when x is at least minus 3. Inside, the limit laws let you substitute. At minus 3 the only approach is from the right, and the limit there is the square root of 0, which is f(−3). So f is continuous on the closed-at-one-end interval from minus 3 to infinity.

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 162 — Checkpoint 2.24

Fill in the blanks before checking. The root of x plus 3 only exists when x plus 3 is zero or positive, which means x is at least minus 3. So the domain has a left endpoint at minus 3 and runs to infinity on the right.

Inside the domain, the limit laws let you substitute, so the function is continuous at every point greater than minus 3. At minus 3 itself there is nothing to the left, so only continuity from the right is required, and the right-hand limit is the root of 0, which equals the value.

So the answer is continuity on the interval that is closed at minus 3 and open at infinity. Infinity never gets a square bracket, because it is not a number the function can be evaluated at.

37. Exercise 145: choose k so the pieces meet

Worked example

Find the value of k that makes f continuous over the interval where it is defined.

\[ f(x) = \begin{cases} 3x + 2 & x < k \\ 2x - 3 & k \le x \le 8 \end{cases} \]

Locate the only possible break

Why: Each piece is a polynomial, continuous on its own; trouble can only occur at the seam x = k.

\[ \text{test } x = k \]

Left-hand limit at k

Why: Inputs below k use the first line.

\[ \lim_{x\to k^-} f(x) = 3k + 2 \]

Value and right-hand limit at k

Why: The second line includes k itself.

\[ f(k) = \lim_{x\to k^+} f(x) = 2k - 3 \]

Demand that they agree

Why: Continuity at k requires all three to be the same number.

\[ 3k + 2 = 2k - 3 \]

Solve

Why: Subtract 2k and 2 from both sides.

\[ k = -5 \]

Figure (svg): Two lines, y equals 3x plus 2 and y equals 2x minus 3, drawn full length and faint, crossing at the point (minus 5, minus 13); the continuous function uses the first line left of the crossing and the second line from it up to 8.

The switch-over point k must be where the two lines cross, or the graph jumps. Two lines cross exactly once, at x = −5.

Check both sides at the seam

Why: Both lines give the same height at minus 5.

\[ 3(-5) + 2 = -13 = 2(-5) - 3 \]

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 165 — Exercise 145

This kind of question appears on nearly every calculus exam. Both pieces are lines, and lines are continuous, so the only point where continuity can fail is the seam, where the rule switches. Here the seam is at x equal to k, which is what you are choosing.

At the seam the left piece arrives at height three k plus 2, and the right piece, which includes k, starts at height two k minus 3. Continuity demands that these agree. Setting them equal gives a linear equation, and k is minus 5.

The picture makes the answer inevitable: the two lines cross exactly once, at the point minus 5, minus 13. Switching from one line to the other anywhere else would leave a vertical jump. The check confirms that both formulas give minus 13 at minus 5.

38. Slide k until the graph joins up

Tweak it

Parameter explorer

The graph is x² for x up to 3, and kx − 3 after 3. Slide k. For which value of k does the right-hand piece start exactly where the left-hand piece ends?

\[ f(x) = \begin{cases} x^2 & x \le 3 \\ {k}x - 3 & x > 3 \end{cases} \]

  • k — from 1 to 7: slope k

\[ 3k - 3 = 9 \iff k = 4 \]

Start with k at 2 and look at the seam at x equal to 3. The left piece, x squared, always ends at height 9. The right piece starts at height three k minus 3, which with k equal to 2 is only 3, so the graph drops.

Slide k upward and watch the right piece rise and steepen. For one value the two pieces meet exactly, and the graph can be traced in one stroke. That value is 4, because three times 4 minus 3 is 9. Push k beyond 4 and the right piece starts above the left one: a jump upward instead.

The equation under the slider is the whole method from the previous slide: set the height where the left piece ends equal to the height where the right piece begins, and solve.

39. Exercise 155: can any value of f(1) fix this?

Step zero

\[ f(x) = \begin{cases} 3x & x > 1 \\ x^3 & x < 1 \end{cases} \]

Discussion prompt

Before choosing a value k for f(1), what should you compute first, and what does it tell you here?

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 166 — Exercise 155

Answer before revealing. The instinct is to hunt for a clever value k, but a value can only fix a break that is removable, so the first thing to find out is what kind of break this is.

The one-sided limits settle it. From the left the function is x cubed, which heads to 1. From the right it is three x, which heads to 3. The sides disagree, so this is a jump.

No value can repair a jump: if you choose k equal to 1, the right side misses it; if you choose 3, the left side misses it; anything else misses both. So the answer is no, and the one-sided limits are the proof. This is the same reasoning as the slide on why only a hole can be repaired.

40. Trap: a two-sided limit at an endpoint

Trap

The trap

On the semicircle from Example 2.34:

\[ \lim_{x\to 2^+}\sqrt{4-x^2} \text{ DNE} \]

\[ \Rightarrow \text{ not continuous on } [-2, 2] \]

Wrong. The right-hand side of 2 is outside the interval.

The fix

At an endpoint of a closed interval, only the side that lies inside the interval is asked for. At b that is the left side:

\[ \lim_{x\to 2^-}\sqrt{4-x^2} = 0 = f(2) \]

That is what the closed interval definition was written for.

The wrong argument notices that the root of 4 minus x squared is undefined just to the right of 2, so the right-hand limit at 2 does not exist, and concludes the function is not continuous on the closed interval.

The premise is true, but it is irrelevant. The interval stops at 2, and the definition of continuity on a closed interval asks only for the one-sided limit from inside the interval. At the right endpoint that is the left-hand limit, which is 0 and matches the value.

This is exactly the situation the closed-interval definition was designed for. Functions defined by roots, logarithms and piecewise rules often stop at an endpoint, and the one-sided condition is what lets you still call them continuous there.

41. Continuous on which intervals?

Comparison

Comparison matrix

functionbreaks atcontinuous on
(x + 1)/(x − 5)x = 5(−∞, 5) and (5, ∞)
√(x + 3)none in its domain[−3, ∞)
√(4 − x²)none in its domain[−2, 2]
(x − 1)/(x² + 2x)x = −2 and x = 0(−∞, −2), (−2, 0), (0, ∞)
3x⁴ − 4x²none(−∞, ∞)

Fill in each blank before checking. For each function, first find the points where it can break: zeros of a denominator, or the edge of a square root's domain.

The first and fourth rows are rational functions, broken only where the denominator vanishes; removing those points leaves open intervals. The two root functions have no breaks inside their domains, and the domain has endpoints, so the answers are intervals closed at those endpoints.

The polynomial has no denominator and no restrictions, so it breaks nowhere and is continuous on the whole real line. Notice how Theorem 2.8 and the endpoint rule together answer every row without a single three-condition check.

42. Composites and trigonometric functions

Section

Part 4

43. The limit passes through a continuous outer function

Picture it

Figure (svg): Three horizontal number lines. Top: inputs x crowding in on a. Middle: the values g(x) crowding in on L. Bottom: the values f(g(x)) crowding in on f(L). Arrows carry each input down to its value and then down again.

The inner function herds its outputs toward L. Because f is continuous at L, it keeps them herded, toward f(L).

Composite Function Theorem (Theorem 2.9) — If the limit of g at a is L, and f is continuous at L, then the limit of f of g at a is f of L.

\[ \lim_{x\to a} f\big(g(x)\big) = f\Big(\lim_{x\to a} g(x)\Big) = f(L) \]

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 163 — Theorem 2.9

Follow the three number lines from top to bottom. On the top line, inputs crowd in on a. The inner function g carries them to the middle line, where their images crowd in on L; that is what the limit of g being L means.

The outer function f then carries those values to the bottom line. Because f is continuous at L, inputs close to L give outputs close to f of L, so the final values crowd in on f of L. That is the Composite Function Theorem: the limit of f of g equals f applied to the limit of g.

In practice this means you can move a limit inside a continuous function. To find the limit of the cosine of something, find the limit of the something and take its cosine. The theorem is what makes that move legitimate.

44. Reading the Composite Function Theorem

Notation

Annotate

On: \( \lim_{x\to a} g(x) = L,\; f \text{ continuous at } L \;\Longrightarrow\; \lim_{x\to a} f(g(x)) = f(L) \)

  • The inner function only needs a limit at a. It need not be continuous there, or even defined at a.
  • The outer function must be continuous at L, the point the inner values are heading to, not at a.
  • The limit moves inside f: take the inner limit first, then apply f to the answer.
  • If f is broken at L, the values f(g(x)) can land on the wrong side of the break (see the error slide below).

The first annotation is the surprising one. The inner function g does not need to be continuous at a, or even defined there. It only needs a limit. All the continuity is demanded of the outer function.

The second annotation is the one that causes errors. The outer function must be continuous at L, the place where the inner values are heading, not at a. You check the outer function at the inner limit.

The last annotation warns what happens if you skip that check. If f is broken at L, the inner values can approach L from a side where f takes a completely different value, and the limit lands in the wrong place. The error-analysis slide later in this part shows it happening with a real function.

45. Example 2.35: a limit slid inside cosine

Worked example

Evaluate the limit.

\[ \lim_{x\to \pi/2} \cos\left(x - \frac{\pi}{2}\right) \]

Name the inner and outer functions

Why: The inner function shifts x; the outer takes a cosine.

\[ g(x) = x - \frac{\pi}{2}, \qquad f(u) = \cos u \]

Find the inner limit

Why: A polynomial in x, so substitute.

\[ \lim_{x\to \pi/2}\left(x - \frac{\pi}{2}\right) = 0 \]

Check the outer function is continuous there

Why: Section 2.3 showed the cosine limit at 0 equals cos 0.

\[ \lim_{u\to 0}\cos u = 1 = \cos 0 \]

Apply Theorem 2.9

Why: Move the limit inside the cosine.

\[ \lim_{x\to \pi/2}\cos\left(x - \frac{\pi}{2}\right) = \cos\left(\lim_{x\to \pi/2}\left(x - \frac{\pi}{2}\right)\right) \]

Evaluate

Why: Cosine of 0.

\[ = \cos 0 = 1 \]

Figure (svg): The graph of cosine of x minus pi over 2, which is the same wave as sine x, for x from 0 to pi, peaking at the point (pi over 2, 1), which is marked.

The inner expression x − π/2 heads for 0, and cosine is continuous at 0, so the limit is simply cos 0 = 1.

Check numerically

Why: Take x a hundredth away from pi over 2 on each side.

\[ \cos(-0.01) = \cos(0.01) \approx 0.99995 \]

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 163 — Example 2.35

Start by seeing the function as a composite: an inner function that shifts x by pi over 2, and an outer cosine. The inner limit is easy because the inner function is a polynomial, and it is 0.

Before moving the limit inside, check the hypothesis. The outer function must be continuous at 0, the inner limit. Section 2.3 showed that the limit of cosine at 0 is 1, which is cos 0, so cosine is continuous there. Now Theorem 2.9 lets the limit pass inside, and the answer is cos 0, which is 1.

The picture shows why the answer makes sense: shifting a cosine right by pi over 2 gives the sine wave, and the sine wave reaches its peak of 1 exactly at pi over 2. The numerical check agrees, because nearby inputs give outputs within a twenty-thousandth of 1.

46. Checkpoint 2.25: sine of x minus pi

Prediction

\[ \lim_{x\to \pi}\sin(x - \pi) \]

Predict first

What is the limit?

  • 0
  • 1
  • −1
  • It does not exist

Correct: 0

Why: The inner expression x minus pi tends to 0, and sine is continuous at 0, so by the composite function theorem the limit is sin 0, which is 0. The answer 1 comes from confusing this with the sine x over x limit; here there is no division by x.

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 163 — Checkpoint 2.25

Commit to an answer first. The structure is the same as the previous example: an inner shift and an outer trigonometric function.

The inner expression, x minus pi, tends to 0 as x tends to pi. Sine is continuous at 0, so the limit moves inside and the answer is sine of 0, which is 0.

If you chose 1, you were probably thinking of the limit of sine x over x. That limit has a division by x that makes it zero over zero; this one has no division at all, so it is just the sine of a number heading to 0. Always look at the exact form before reaching for a remembered answer.

47. Why cosine is continuous everywhere

Concept

Theorem 2.10 says every trigonometric function is continuous on its domain. The proof for cosine at any point a writes x as a plus a small shift, then uses the angle-sum identity:

\[ \lim_{x\to a}\cos x = \lim_{x\to a}\cos\big((x-a) + a\big) \]

\[ = \lim_{x\to a}\big(\cos(x-a)\cos a - \sin(x-a)\sin a\big) \]

\[ = \cos(0)\cos a - \sin(0)\sin a \]

\[ = 1 \cdot \cos a - 0 \cdot \sin a = \cos a \]

The third line is Theorem 2.9: the shift x minus a tends to 0, and sine and cosine are continuous at 0. Sine works the same way, and tangent, secant and the rest are quotients of the two, continuous wherever their denominators are not zero.

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 163 — Theorem 2.10 and its proof

Section 2.3 only established that sine and cosine behave well at 0. This proof extends that to every real number a, and it is a lovely use of the Composite Function Theorem.

The trick in the first line is to write x as a plus the small shift x minus a. The angle-sum identity then splits cosine of x into pieces involving cosine and sine of that small shift, multiplied by the constants cos a and sin a. As x approaches a, the shift approaches 0, and because sine and cosine are continuous at 0, the pieces approach cos 0 and sin 0, which are 1 and 0.

What survives is cos a, so the limit equals the value and cosine is continuous at a. Sine works the same way, and every other trigonometric function is a quotient of sine and cosine, so it is continuous wherever its denominator is not zero. That is Theorem 2.10.

48. Building continuity from known pieces

Concept

Sums, differences, products and quotients (where the bottom is not zero) of continuous functions are continuous, by the limit laws; composites are continuous by Theorem 2.9. So continuity can be read off a formula:

\[ \lim_{x\to 3}\sqrt{x^2 + 7} = \sqrt{\lim_{x\to 3}(x^2+7)} = \sqrt{16} = 4 \]

\[ \frac{\sin x}{x^2 - 4}: \quad \text{continuous except where } x^2 - 4 = 0, \text{ i.e. } x = \pm 2 \]

The first is a root of a polynomial, continuous where the inside is positive. The second is a quotient of two continuous functions, broken only where its denominator vanishes.

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 167 — Exercises 161 and 166 use the same reasoning

You now have a small library of continuous functions: polynomials, rational functions, roots on their domains, and the trigonometric functions. The limit laws say that adding, subtracting, multiplying and dividing continuous functions gives continuous functions, as long as you never divide by zero. Theorem 2.9 adds composition.

So most continuity questions reduce to reading the formula. The square root of x squared plus 7 is a root of a polynomial that is always positive, so it is continuous everywhere, and its limit at 3 is found by substituting: the root of 16, which is 4.

The quotient of sine x by x squared minus 4 is continuous wherever the denominator is not zero, so everywhere except 2 and minus 2. The only real work is finding where a denominator vanishes or a root's inside turns negative. Exercises 161 and 166 in the book are this reasoning exactly.

49. Find the error: sliding a limit through a break

Error analysis

Annotate

On: \( \lim_{x\to 0}\lfloor \cos x \rfloor = \Big\lfloor \lim_{x\to 0}\cos x \Big\rfloor = \lfloor 1 \rfloor = 1 \)

  • Correct: cos x tends to 1 as x tends to 0.
  • The error. Theorem 2.9 needs the OUTER function continuous at L = 1, and the floor function jumps at 1.
  • For x near 0 but not 0, cos x is slightly LESS than 1, so its floor is 0. The limit is 0, not 1.
  • Check continuity of the outer function at the inner limit before moving the limit inside.

\[ 0 < |x| < \tfrac{\pi}{2} \;\Rightarrow\; 0 < \cos x < 1 \;\Rightarrow\; \lfloor \cos x \rfloor = 0 \]

The floor function rounds down to the nearest whole number, so its graph is a staircase that jumps at every integer. Read the argument and look for the step that needs a hypothesis it does not have.

The inner limit is fine: cos x tends to 1. The error is moving the limit inside the floor. Theorem 2.9 requires the outer function to be continuous at the inner limit, which is 1, and the floor function jumps exactly at 1.

The truth comes from looking at the actual values. For x near 0 but not equal to 0, cos x is slightly less than 1, and the floor of a number slightly less than 1 is 0. So the limit is 0, not 1. The inner values approach 1 from below, landing on the lower step of the staircase. Always check the outer function at the inner limit.

50. The Intermediate Value Theorem

Section

Part 5

51. A continuous function cannot skip a value

Concept

Figure (svg): A continuous curve over the interval from a to b: it rises to a hump, dips, then climbs steeply. Dashed lines mark the heights f(a), f(b) and a height z between them; the curve crosses height z at an input c between a and b.

Any height z between f(a) and f(b) is met at least once. The curve may wander above and below on the way, but it cannot skip z.

Intermediate Value Theorem (Theorem 2.11) — Let f be continuous over a closed, bounded interval from a to b. If z is any real number between f(a) and f(b), then there is a number c in that interval with f(c) equal to z.

The most useful case takes z to be 0: a continuous function that changes sign on an interval has a zero inside it.

\[ f(a) < 0 < f(b) \;\Longrightarrow\; f(c) = 0 \text{ for some } c \in (a, b) \]

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 164 — Theorem 2.11 and Figure 2.38

Look at the picture. The curve is continuous from a to b, so you can trace it without lifting the pencil. It starts at height f of a and ends at height f of b. Pick any height z in between, the dashed line. The theorem says the curve must cross that line somewhere; c marks the place.

The curve is allowed to wander: here it rises, dips below the starting height, and then climbs. It might cross the line at z several times. What it cannot do is get from below z to above z without meeting it, because that would require a break.

The most common use takes z equal to 0. If a continuous function is negative at one end of an interval and positive at the other, it has a zero in between. That turns the problem of solving an equation into the much easier problem of finding a sign change.

52. Reading the Intermediate Value Theorem

Notation

Annotate

On: \( f \text{ continuous on } [a,b],\; z \text{ between } f(a), f(b) \;\Longrightarrow\; \exists\, c \in [a,b]: f(c) = z \)

  • The engine of the theorem. Without it, the graph can hop over z (Example 2.38).
  • Closed and bounded: the endpoints belong, so f(a) and f(b) are genuine values of a connected piece of graph.
  • At least one c. The theorem says nothing about how many, and nothing about where exactly.
  • It only produces values. It never proves a value is missed (Example 2.37).

Each annotation marks a part of the statement you must not lose. Continuity on the whole interval is the engine; without it the graph could jump over z, and the theorem would be false. Example 2.38 shows exactly that.

The interval is closed and bounded, so the endpoints belong to it and the values at them are part of the same unbroken piece of graph.

The conclusion says there exists a c, which means at least one. It does not say how many, and it does not say where. Finally, the theorem works in one direction only: it tells you that certain values are hit, and it never tells you that a value is missed. Example 2.37 is built around that last point.

53. Example 2.36: x minus cos x has a zero

Worked example

Show that the function has at least one zero.

\[ f(x) = x - \cos x \]

Establish continuity

Why: A difference of a polynomial and a cosine, both continuous everywhere.

\[ f \text{ continuous on } (-\infty, +\infty) \]

Evaluate at a convenient left point

Why: At 0 the cosine is 1.

\[ f(0) = 0 - \cos 0 = -1 < 0 \]

Evaluate at a convenient right point

Why: At pi over 2 the cosine is 0.

\[ f\left(\frac{\pi}{2}\right) = \frac{\pi}{2} - \cos\frac{\pi}{2} = \frac{\pi}{2} > 0 \]

Apply the IVT with z = 0

Why: f is continuous on the closed interval and 0 lies between minus 1 and pi over 2.

\[ \exists\, c \in \left[0, \tfrac{\pi}{2}\right]: \; f(c) = 0 \]

Figure (svg): The graph of x minus cos x from minus 1 to 2, rising steadily; a dot below the axis at (0, minus 1), a dot above the axis at (pi over 2, pi over 2), and the crossing near 0.739 marked on the x-axis.

Below the axis at 0, above it at π/2, and no breaks in between: the graph has to cross somewhere in that interval.

Check by finding it

Why: Bisection pins the zero at about 0.739, inside the interval, and there the two terms balance.

\[ c \approx 0.7391: \quad \cos(0.7391) \approx 0.7391 \]

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 164 — Example 2.36

The equation x equals cos x cannot be solved with algebra, yet this example proves it has a solution. The method is to find two inputs where the function has opposite signs.

First establish continuity: x is a polynomial, cos x is continuous everywhere, and their difference is continuous. Then choose inputs where cosine is easy. At 0 the function is minus 1, negative. At pi over 2 it is pi over 2, about 1.571, positive. Zero lies between them, so the IVT guarantees a zero in the interval from 0 to pi over 2.

The picture shows the steady rise across the axis, and the check finds the crossing numerically at about 0.739, where x and cos x agree to four decimal places. The theorem only promised existence; the calculation confirms the promise.

54. Checkpoint 2.26: a cubic crossing on [0, 1]

Worked example

Show that the function has a zero over the interval from 0 to 1.

\[ f(x) = x^3 - x^2 - 3x + 1 \]

Continuity

Why: A polynomial, continuous everywhere, so on the closed interval.

\[ f \text{ continuous on } [0, 1] \]

Left endpoint

Why: Every term but the constant vanishes.

\[ f(0) = 1 > 0 \]

Right endpoint

Why: Add the four terms.

\[ f(1) = 1 - 1 - 3 + 1 = -2 < 0 \]

Apply the IVT

Why: Zero lies between minus 2 and 1.

\[ \exists\, c \in [0, 1]: \; f(c) = 0 \]

Figure (svg): The cubic x cubed minus x squared minus 3x plus 1 from minus 2 to 3, with the interval from 0 to 1 highlighted: the curve is at 1 when x is 0 and at minus 2 when x is 1, crossing the axis near 0.311.

The IVT only promises the crossing inside [0, 1]. The cubic has other roots outside that interval; the theorem says nothing about them.

Check by halving once

Why: At the midpoint the value is already negative, so the zero is in the left half, consistent with the root near 0.311.

\[ f(0.5) = 0.125 - 0.25 - 1.5 + 1 = -0.625 < 0 \]

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 165 — Checkpoint 2.26

Try it yourself before looking at the steps. The function is a polynomial, so it is continuous on any interval you like, including this one. Evaluate at the endpoints: at 0 only the constant survives, giving 1; at 1 the terms add to minus 2.

Positive at one end, negative at the other, continuous in between: the IVT gives a zero in the interval from 0 to 1. That is the whole required answer.

The picture adds context. The cubic has three real roots, and only one of them is inside the highlighted interval. The theorem is local: it tells you about the interval you chose and nothing else. The check halves the interval once and finds the function already negative at one half, so the zero is in the left half, consistent with the actual root near 0.311.

55. Exercise 152: trap the solution of cos t = t³

Worked example

Prove that the equation has a solution, then find an interval of length 0.01 containing one.

\[ \cos t = t^3 \iff g(t) = \cos t - t^3 = 0 \]

Signs at the ends of [0, 1]

Why: g is continuous (cosine minus a polynomial).

\[ g(0) = 1 > 0, \qquad g(1) = \cos 1 - 1 \approx -0.4597 < 0 \]

Conclude a solution exists

Why: The IVT with z = 0.

\[ \exists\, t \in [0, 1]: \; \cos t = t^3 \]

Halve: test 0.5

Why: Positive, so the sign change is in the right half.

\[ g(0.5) \approx 0.7526 > 0 \;\Rightarrow\; [0.5, 1] \]

Halve: test 0.75

Why: Still positive.

\[ g(0.75) \approx 0.3098 > 0 \;\Rightarrow\; [0.75, 1] \]

Halve: test 0.875

Why: Negative, so keep the left half.

\[ g(0.875) \approx -0.0289 < 0 \;\Rightarrow\; [0.75, 0.875] \]

Figure (svg): The graph of cos t minus t cubed from 0 to 1.1, falling from 1 through zero near t equals 0.865. Below the axis, four bars show the bracketing intervals shrinking: from 0 to 1, 0.5 to 1, 0.75 to 1, and 0.75 to 0.875.

Each halving keeps the half whose endpoints still have opposite signs. Every bar contains the root; each is half as long as the one above.

Check an interval of length 0.01

Why: Evaluating at hundredths inside [0.75, 0.875] finds the sign change between 0.86 and 0.87.

\[ g(0.86) \approx 0.0164 > 0, \quad g(0.87) \approx -0.0137 < 0 \;\Rightarrow\; t \in [0.86, 0.87] \]

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 166 — Exercise 152

The first move is to turn the equation into a function whose zeros are its solutions: subtract t cubed from cos t. This g is continuous, being a cosine minus a polynomial.

At 0 the function is 1, positive; at 1 it is cos 1 minus 1, negative. So the IVT gives a solution in the interval from 0 to 1. Then comes bisection: test the midpoint, and keep whichever half still has a sign change. The bars in the picture show the interval shrinking from length 1 to one half, a quarter, an eighth.

Each halving costs one evaluation and cuts the uncertainty in half, so seven halvings would get below one hundredth. The last step shortcuts that by testing hundredths inside the current interval and finding the sign change between 0.86 and 0.87. That interval of length 0.01 is exactly what the exercise asked for, and it is how calculators solve equations.

56. Exercise 153: which interval holds the solution?

Estimation

\[ 2^x = x^3 \iff k(x) = 2^x - x^3 = 0 \]

Predict first

Use k(1.25) ≈ 0.4253, k(1.375) ≈ −0.0059 and k(1.5) ≈ −0.5466. What can the IVT tell you about the two intervals [1.25, 1.375] and [1.375, 1.5]?

  • A solution in [1.25, 1.375]; nothing either way about [1.375, 1.5]
  • A solution in each interval
  • A solution in [1.375, 1.5] only
  • No solution in [1.375, 1.5]

Correct: A solution in [1.25, 1.375]; nothing either way about [1.375, 1.5]

Why: k is continuous and changes sign on the first interval (from 0.4253 down to minus 0.0059), so the IVT guarantees a solution there; bisection puts it near 1.3735. On the second interval both endpoint values are negative, so the IVT makes no promise, but it does not forbid a solution either. Only the first option says exactly what the theorem licenses.

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 166 — Exercise 153

Choose an answer before revealing, and pay attention to exactly what the theorem is entitled to say.

On the first interval the function goes from positive to negative, so the IVT guarantees a solution there. On the second interval both endpoint values are negative. The IVT needs the target value, here 0, to lie between the endpoint values, and it does not, so the theorem makes no promise about that interval.

But no promise is not the same as a proof of absence. The option that says there is no solution in the second interval claims more than the theorem gives. The correct answer states what the IVT guarantees on the first interval and nothing about the second. Notice how close the solution is to the shared endpoint: k at 1.375 is only a few thousandths below zero.

57. Example 2.37: both ends positive, and still a zero

Worked example

f is continuous on [0, 2], with f(0) and f(2) both positive. Can the IVT be used to conclude f has no zeros in [0, 2]?

\[ f(0) > 0, \quad f(2) > 0 \quad \overset{?}{\Longrightarrow} \quad \text{no zeros} \]

Read what the IVT promises

Why: It guarantees that values between f(0) and f(2) are hit. It makes no statement about values outside that range.

\[ z \text{ between } f(0), f(2) \;\Rightarrow\; z \text{ is hit} \]

Test the claim on a specific function

Why: The book's choice: a parabola with its vertex on the axis.

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

Evaluate the endpoints

Why: Both positive.

\[ f(0) = 1 > 0, \qquad f(2) = 1 > 0 \]

Evaluate the middle

Why: A zero, in spite of the positive endpoints.

\[ f(1) = 0 \]

Figure (svg): The parabola (x minus 1) squared from minus 0.5 to 2.5, with dots at (0, 1) and (2, 1), both above the axis, and the vertex (1, 0) touching the axis between them.

Positive at both ends, yet a zero in between. The IVT is silent here, and silence is not a 'no'.

Check the answer

Why: One continuous function meets every hypothesis and has a zero, so the conclusion cannot follow. The answer is no.

\[ \text{No: the IVT cannot prove the absence of zeros} \]

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, pp. 164-165 — Example 2.37

This example is about reading the theorem in the wrong direction. The IVT says that values between f of a and f of b are hit. It does not say that values outside that range are missed. With both endpoint values positive, zero is not between them, so the theorem simply has nothing to say about zeros.

To prove that the conclusion cannot be drawn, it is enough to find one function that satisfies all the hypotheses and still has a zero. The parabola x minus 1, squared, does the job: it is 1 at both ends and touches zero at 1, as the picture shows.

Keep this example as your reminder that the IVT is a one-way theorem. It can prove that a solution exists. It can never prove that one does not.

58. Example 2.38: a sign change with no zero

Worked example

For f(x) = 1/x, the value at minus 1 is negative and the value at 1 is positive. Can we conclude that f has a zero between minus 1 and 1?

\[ f(-1) = -1 < 0, \qquad f(1) = 1 > 0 \]

Check the IVT's hypothesis

Why: The function must be continuous on the whole closed interval.

\[ f(0) = \frac{1}{0} \quad \text{undefined} \]

Conclude the hypothesis fails

Why: Zero lies inside the interval, so f is not continuous on it.

\[ f \text{ not continuous on } [-1, 1] \]

So the theorem does not apply

Why: No conclusion either way comes from the IVT.

\[ \text{IVT: no conclusion} \]

Figure (svg): The graph of 1 over x from minus 1 to 1: the left branch drops from (minus 1, minus 1) toward minus infinity, the right branch comes down from plus infinity to (1, 1); a dashed asymptote at x equals 0 separates them and the graph never touches the x-axis.

The sign changes, but the graph jumps the axis by breaking at 0 instead of crossing it. Continuity is the hypothesis that forbids this.

Check directly

Why: One over x is never zero, because a fraction is zero only when its numerator is, and here the numerator is 1. The sign change happens across the asymptote.

\[ \frac{1}{x} = 0 \text{ has no solution} \]

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 165 — Example 2.38

Here the sign change looks like it should force a zero: the function is minus 1 at minus 1 and 1 at 1. But before using the IVT you must check its hypothesis, and it fails. One over x is undefined at 0, which lies inside the interval, so the function is not continuous there.

Without continuity the theorem does not apply, and the picture shows what goes wrong. The graph passes from negative to positive values by breaking at the asymptote, not by crossing the axis. One over x is never zero, since a fraction is zero only when its numerator is, and this numerator is always 1.

Put Examples 2.37 and 2.38 side by side. The first shows the conclusion can hold without a sign change. The second shows a sign change is useless without continuity.

59. Exercise 150: the value 10 is skipped

Anomaly

\[ h(x) = \begin{cases} 3x^2 - 4 & x \le 2 \\ 5 + 4x & x > 2 \end{cases}, \qquad h(0) = -4, \; h(4) = 21 \]

Figure (svg): The piecewise function 3x squared minus 4 up to x equals 2, ending with a filled dot at (2, 8), and 5 plus 4x after 2, starting at a hollow circle (2, 13). A dashed horizontal line at height 10 passes through the gap between 8 and 13 without touching the graph.

The graph leaps from 8 to 13 at x = 2, stepping right over every height in between, 10 included.

Predict first

h(0) is below 10 and h(4) is above 10, yet no x in [0, 4] gives h(x) = 10. Does this contradict the IVT?

  • No: h is not continuous on [0, 4]
  • Yes: the IVT has failed
  • No: 10 is not between −4 and 21
  • No: the IVT needs h(0) and h(4) to have opposite signs

Correct: No: h is not continuous on [0, 4]

Why: At 2 the left piece reaches 12 minus 4, which is 8, while the right piece starts from 5 plus 8, which is 13. That is a jump from 8 to 13, stepping over 10. The IVT requires continuity on the whole interval, so it simply does not apply to h.

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 166 — Exercise 150

Choose an answer before revealing. It looks like a contradiction: the function is below 10 at 0, above 10 at 4, and never equal to 10 in between.

The resolution is in the picture. At 2 the first piece ends at height 8 and the second piece starts at height 13. The graph jumps over every height from 8 to 13, including 10. That jump is a discontinuity, so the function is not continuous on the interval from 0 to 4, and the IVT does not apply to it.

A theorem is only contradicted by an example that meets all its hypotheses and still fails its conclusion. This function fails a hypothesis, so it contradicts nothing. That is a useful habit in general: whenever a theorem seems to fail, look for the hypothesis that is missing.

60. Trap: a sign change means exactly one root

Trap

The trap

Exercise 165 states: if f is continuous and f(a), f(b) have opposite signs, then

\[ f(x) = 0 \text{ has exactly one} \]

\[ \text{solution in } [a, b] \]

False. The IVT says at least one.

The fix

A continuous curve can cross the axis three times, or five, between endpoints of opposite sign. For example:

\[ f(x) = x^3 - x \text{ on } [-2, 2] \]

\[ f(-2) = -6, \; f(2) = 6 \]

\[ \text{zeros at } -1, \; 0, \; 1 \]

Existence, not uniqueness.

The IVT says at least one, and this trap quietly strengthens it to exactly one. Exercise 165 in the book asks you to spot the difference.

A single counterexample settles it. The cubic x cubed minus x is continuous, negative at minus 2 and positive at 2, and it has three zeros in between, at minus 1, 0 and 1. The IVT is perfectly happy with that; it only ever promised one.

To prove there is exactly one root you need extra information, for example that the function is always increasing on the interval. You will get tools for that from derivatives in Chapter 4. For now, when you use the IVT, write at least one, and mean it.

61. Same place, same time on the mountain

Real world

Figure (svg): Height on the trail plotted against clock time from 6 to 18 hours for two days: the Saturday climb rises from 0 to 1200 metres, the Sunday descent falls from 1200 to 0. The two curves cross once, marked with a dot.

The difference between the two curves starts negative and ends positive, so it is zero somewhere: the same spot at the same clock time.

Discussion prompt

A hiker climbs a trail from 6 am to 6 pm on Saturday, camps at the top, and walks down the same trail from 6 am to 6 pm on Sunday, at whatever speeds she likes. Prove there is a spot she passes at exactly the same clock time on both days.

OpenStax Calculus Volume 1, §2.4 Continuity §2.4, p. 166 — compare Exercise 151

Try the proof before revealing; it is a famous puzzle, and the IVT makes it almost effortless. The key idea is the same one Exercise 151 uses: compare the two days by subtracting them.

Let d be the height on Saturday minus the height on Sunday, at the same clock time. At 6 am she is at the bottom on Saturday and at the top on Sunday, so d is negative. At 6 pm it is the other way round, so d is positive. Heights change continuously in time, so d is continuous, and the IVT gives a time when d is zero.

At that time she is at the same height, on the same trail, so at the same spot, on both days. The picture shows the crossing of the two height curves. Notice that you never need to know her speeds: continuity alone forces the meeting.

62. Put an IVT argument in order

Ranking

Put in order

Order the steps of a complete proof that an equation has a solution in [a, b].

  1. Rewrite the equation as f(x) = 0
  2. State why f is continuous on [a, b]
  3. Evaluate f(a) and f(b) and note their opposite signs
  4. Apply the IVT with z = 0
  5. Conclude: some c in [a, b] solves the equation

Why: Moving everything to one side turns 'two things are equal' into 'one thing is zero', which is the form the theorem handles. Continuity is stated before the theorem is used, because it is the hypothesis Example 2.38 showed cannot be skipped.

Drag the steps into order before checking. The first step is the one people forget: the theorem is about a function hitting a value, so rewrite the equation as a single function equal to zero.

Next comes continuity, stated explicitly, because it is the hypothesis that fails in Examples 2.38 and Exercise 150. Only then evaluate the endpoints and point out the sign change, apply the theorem, and state the conclusion in terms of the original equation.

Written in this order, the argument is complete and graders cannot fault it. Skipping the continuity line is the most common way to lose marks on an otherwise correct IVT proof.

63. Putting it together

Section

Part 6

64. Pattern: deciding continuity and using it

Pattern

Figure (svg): A left-to-right chain of four stages. First find where the formula can break: zero denominators, the seams of a piecewise rule, the edges of a root's domain. Then test only those points with the three conditions. Then classify each break from its one-sided limits. Finally, on a closed interval with no breaks, the IVT may be used.

Only a handful of points can break. Find them first, test only them, and the rest of the domain is continuous by theorem.
  1. List the candidate points: zeros of a denominator, the seams of a piecewise rule, the edges of a root's domain. Everywhere else is continuous by Theorems 2.8 to 2.10.
  2. Test each candidate with the three conditions, in order, stopping at the first failure.
  3. Classify a break from its one-sided limits: equal means removable, finite and unequal means jump, infinite means infinite.
  4. On an interval, ask for two-sided continuity inside and one-sided continuity at an included endpoint.
  5. Before using the IVT, state continuity on the closed interval, then find a sign change.

This is the routine for any continuity question. The diagram reads left to right, and the first stage does most of the work: find the handful of places a formula can break. Zero denominators, the seams of piecewise rules and the edges of a root's domain are the usual suspects. Every other point is continuous by the theorems in this lesson.

At each suspect point, run the three checks in order. If the function fails, compute the one-sided limits and classify the break. If a question asks about an interval, remember that endpoints need only the inside limit.

The last stage uses continuity rather than testing it. Once you know a function is continuous on a closed interval, a sign change at the ends proves a root inside, and repeated halving traps it as tightly as you like.

65. Check: which condition fails?

Check

Check your understanding

For f(x) = (x² − 1)/(x − 1) with f(1) defined as 5, what is true at x = 1?

  • A. Continuous, because f(1) is defined
  • B. Removable discontinuity: the limit is 2, not 5 (correct)
  • C. Jump discontinuity: the sides are 2 and 5
  • D. Infinite discontinuity: the denominator is 0

Answer: B

Why: Cancelling x − 1 leaves x + 1, so both one-sided limits are 2. The value was set to 5, so condition (iii) fails with the limit existing: a removable discontinuity, repaired by setting f(1) = 2.

Why A tempts people
Being defined passes only condition (i); the limit 2 must also equal the value 5, and it does not.
Why C tempts people
5 is the value, not a one-sided limit. Both sides approach 2, so there is no jump.
Why D tempts people
The zero in the denominator cancels with a zero in the numerator; the values near 1 stay near 2 and do not blow up.

Work it out before choosing. The value at 1 is given as 5, so the first check passes. For the limit, factor the numerator as a difference of squares and cancel, leaving x plus 1, which heads to 2.

The limit exists and does not equal the value, so the third condition fails and the discontinuity is removable. Setting the value at 1 to 2 would repair it. The jump option mistakes the value for a one-sided limit, and the infinite option forgets that the zero on the bottom is cancelled by a zero on top.

66. Check: the composite function theorem

Check

Check your understanding

Which limit can be found by moving the limit inside the outer function?

  • A. lim as x → 0 of ⌊cos x⌋
  • B. lim as x → 3 of √(x² + 7) (correct)
  • C. lim as x → 0 of 1/(x²)
  • D. lim as x → 1 of |x − 1|/(x − 1)

Answer: B

Why: The inner limit is 16, and the square root is continuous at 16, so the limit is √16 = 4. In each other case the outer function is not continuous at the inner limit, or there is no inner limit to use.

Why A tempts people
The floor function jumps at 1, the inner limit, so the limit cannot move inside; the true answer is 0, not 1.
Why C tempts people
The inner limit is 0, and 1/u is undefined at 0, so the theorem does not apply; the limit is infinite.
Why D tempts people
This is not a continuous function of a single inner expression at the limit point; its sides are −1 and 1, so the limit does not exist.

For each option, find the inner limit and ask whether the outer function is continuous there. Only then may the limit move inside.

In the square root option the inner limit is 16, and the square root is continuous at 16, so the limit is 4. The floor option fails because the floor function jumps at 1, the inner limit; that was the error-analysis slide. The reciprocal option fails because the outer function is undefined at the inner limit 0. The absolute value option has no limit at all, since its sides are minus 1 and 1.

67. Check: what the IVT guarantees

Check

Check your understanding

f is continuous on [−1, 1] (Exercise 164 uses f(x) = cos x − sin x − x), with f(−1) ≈ 2.382 and f(1) ≈ −1.301. Which conclusion is justified?

  • A. f(x) = 2 has exactly one solution in [−1, 1]
  • B. f(x) = 2 has at least one solution in [−1, 1] (correct)
  • C. f(x) = 3 has a solution in [−1, 1]
  • D. f has no zero in [−1, 1]

Answer: B

Why: 2 lies between −1.301 and 2.382, and f is continuous on the closed interval, so the IVT gives at least one c with f(c) = 2. It says nothing about how many.

Why A tempts people
The IVT gives existence, never uniqueness. (This f happens to be decreasing, but the theorem alone does not show that.)
Why C tempts people
3 is not between f(1) and f(−1), so the IVT promises nothing about it.
Why D tempts people
0 is between the endpoint values, so the IVT guarantees a zero, the opposite of this claim.

Look at the two endpoint values and ask which target values lie between them. The value 2 does, since it is between about minus 1.3 and about 2.38. So the IVT guarantees at least one solution of f equals 2.

The exactly-one option overstates the theorem; the value 3 is outside the range of the endpoint values, so the theorem is silent about it; and the no-zero option contradicts the theorem, since 0 is between the endpoint values too. This is Exercise 164 from the book, which asks whether the statement is true: it is.

68. Explain why continuity is essential

Explain it to yourself

Discussion prompt

In two or three sentences, explain why the IVT needs continuity on the closed interval, using Example 2.38 or Exercise 150 as your evidence.

Write your explanation before revealing. A good answer names the mechanism, not just the rule. The theorem works because a continuous graph cannot change height without passing through every height in between.

Both counterexamples in the lesson show what a break allows. One over x goes from negative to positive by leaping across an asymptote, never touching zero. The function in Exercise 150 steps from 8 straight to 13 at a jump, never taking the value 10.

If your answer says something like the theorem requires continuity, push it one step further: say what the graph would be able to do without it. That is the understanding that lets you spot a misuse of the theorem on sight.

69. Exit ticket

Exit ticket

\[ f(x) = \begin{cases} \dfrac{x^2 - 9}{x - 3} & x \ne 3 \\ k & x = 3 \end{cases} \]

Discussion prompt

(1) Which value of k makes f continuous at 3? (2) With that k, use the IVT to show that f(x) = 5 has a solution in [0, 3].

This one problem uses both halves of the lesson. First make the function continuous, then use that continuity.

Away from 3 the fraction cancels to x plus 3, so the limit at 3 is 6, and choosing k equal to 6 fills the hole. The discontinuity was removable, and now the function is simply x plus 3 everywhere, continuous on the interval from 0 to 3.

At 0 the function is 3 and at 3 it is 6, and 5 lies between them, so the IVT gives some c in the interval with f of c equal to 5. Here you can even find it: x plus 3 equals 5 when x is 2. If you had left k as any other value, the IVT would not have applied, because the function would be broken at the endpoint 3.

70. Recap

Recap

ideastatementuse it to
continuity at af(a) defined, limit exists, limit = f(a)substitute into a limit
removablelimit exists (value missing or wrong)repair: set f(a) = limit
jumpone-sided limits finite and differentno repair possible
infinitea one-sided limit is ±∞locate vertical asymptotes
closed intervalinside: two-sided; ends: one-sidedset up the IVT
composite theoremlim f(g(x)) = f(lim g(x)) if f continuous thereslide a limit inside
IVTcontinuous on [a, b], z between f(a), f(b) ⇒ f(c) = zprove a root exists, then trap it

\[ f \text{ continuous at } a \iff \lim_{x\to a} f(x) = f(a) \]

Next, Section 2.5 makes the word 'approaches' precise, and continuity will then have an exact definition too.

Stewart, Calculus: Early Transcendentals 8e, §2.5 Continuity §2.5, pp. 114-125 — the same material in Stewart

Continuity at a point means three things: a value, a limit, and agreement between them. When any of them fails, the one-sided limits tell you the kind of break: removable if the limit exists, jump if the sides are finite and different, infinite if a side runs off to infinity. Only the removable kind can be repaired, by setting the value equal to the limit.

On an interval, interior points need two-sided continuity and included endpoints need it only from the inside. Polynomials, rational functions, roots and trigonometric functions are continuous wherever they are defined, and the Composite Function Theorem lets a limit move inside a continuous outer function.

The Intermediate Value Theorem turns continuity into existence: a continuous function on a closed interval hits every value between its endpoint values. Next, Section 2.5 gives the precise definition of a limit, which will put all of this on exact footing.

Sources

  1. OpenStax Calculus Volume 1, §2.4 Continuity — Strang & Herman, OpenStax / Rice University, CC BY-NC-SA 4.0, pp. 155-168
  2. Stewart, Calculus: Early Transcendentals 8e, §2.5 Continuity — James Stewart, Cengage Learning, 2016, pp. 114-125

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

Book on Wyzant · Text (657) 465-8108