← back to MATH 101
12 worked examples23 exercises6 figuresbölüm 2 / 14

02Continuity, asymptotes, and limits at infinity

At 06:00 the balcony thermometer read $-2$ °C; at 14:00 it read $11$ °C. Nobody watched it in between, and still you can be certain that at some instant it read exactly $0$ °C. Now count the people in the building instead: $3$ at 06:00, $40$ at 14:00, and it may well never have been exactly $20$, because a bus can unload seventeen of them at once.

By the end of this section you can write the two line argument that proves $x^{3}=x+1$ has a solution between $1$ and $2$ without finding it, name the single property of the thermometer that makes the argument work, and say exactly why the same argument is illegal for the people counter.

In 60 seconds

Continuity is the licence to replace $\lim_{x\to a}f(x)$ by $f(a)$; this section says where that licence holds, what to do at the points where it does not, and what the graph is doing at the two ends, where there is no $f(a)$ at all.

Repairable break
$\lim_{x\to a}f(x)=L\ \text{exists}\ \Rightarrow\ \text{set}\ f(a):=L$

the two sides agree on a finite number but $f(a)$ is missing or parked elsewhere

$f\ \text{continuous on}\ [a,b],\ N\ \text{between}\ f(a),f(b)\ \Rightarrow\ \exists c\in(a,b):f(c)=N$

"show that the equation has a solution", with no way to solve it

Powers die at infinity
$\lim_{x\to\pm\infty}\frac{1}{x^{n}}=0\quad(n>0)$

after dividing top and bottom by the dominant power of $x$

Cancel before you classify
$\frac{P(x)}{Q(x)}=\frac{(x-a)\tilde P(x)}{(x-a)\tilde Q(x)}\ \Rightarrow\ \text{hole at}\ x=a,\ \text{not an asymptote}$

listing vertical asymptotes of a rational function

Three most common mistakes
  1. Calling $x=a$ a vertical asymptote because the denominator vanishes there. Reduce the fraction first: a factor that dies in the numerator too leaves a .

  2. Using the Intermediate Value Theorem on an interval that contains a break. $\tan$ is $1$ at $\pi/4$ and $-1$ at $3\pi/4$ and is never $0$ in between.

  3. Throwing away the small terms under a root. $\sqrt{x^{2}+x}-x$ does not tend to $0$; it tends to $\tfrac12$.

Weights this term: Midterm 1 28%, Midterm 2 28%, Final 28%, quizzes 10%, homework 6%. The two midterms also set the FZ line — under 40 points out of 200 and you cannot sit the final.

How much time do you have?
10 minutes

You leave with the classification of a break, the asymptote recipe and the three mistakes that cost the most marks — enough to attempt every standard question, not enough to defend a proof.

In 60 seconds, Three ways a graph can break, and the one you can repair, Cancel first: holes, vertical asymptotes and the direction of each, Formula card
45 minutes

Add the theorem that gets asked in words rather than in symbols, and the far away picture. This is the honest minimum for a midterm question that says "show that there is a solution".

In 60 seconds, Three ways a graph can break, and the one you can repair, Proving a solution exists without solving anything, What the graph does far away, Full exam-style question, Practice C · exam level
the full read

Everything in order, including the two blocks people skip and then lose marks on: endpoints of a , and where the continuity check sits in a composition.

everything, top to bottom, then Practice D · interleaved last
By the end of this section
  1. Classify a break at a point as removable, jump or infinite, and repair the removable one by naming the value $f(a)$ has to take.

  2. Decide on which interval a function built from standard families is continuous, endpoints included, without ever writing a limit.

  3. Move a limit through a continuous outer function and say at which number the continuity of that outer function was checked.

  4. Prove that an equation has a solution in a given interval with the Intermediate Value Theorem, stating the hypothesis you verified and the interval you had to avoid.

  5. Compute a limit as $x\to\pm\infty$ by dividing by the dominant power, keeping the sign of $\sqrt{x^{2}}=\vert x\vert$ right, and report the .

  6. Produce the complete asymptote map of a rational function: holes with their coordinates, vertical asymptotes with both one sided limits, and the .

Syllabus coverage
1.6

Calculating limits using the limit laws

The laws were computed with in the previous section. Here they are used twice more: as the theorem that lets you declare whole families of functions continuous without writing a single limit, and as the divide by the dominant power move that keeps them usable when $x$ runs away.

covered
1.8

Continuity: classification of breaks, , compositions, and the Intermediate Value Theorem

Four blocks: the three kinds of break and the repair, continuity on an interval with the endpoint rule, continuity of a composition, and the Intermediate Value Theorem with the interval trap that kills half the attempts.

covered
extra

Limits at infinity, horizontal asymptotes, and the slant case

This week's line names 1.6 and 1.8 only, so the end behaviour of a graph is not part of this week's reading; the plan returns to it later in the term, with the curve sketching block. It is here because it is the other half of the same picture: once you can say where a graph breaks, the remaining question about its shape is what it does at the two ends.

off_syllabus
Recall first
The three part test at a point

$f$ is continuous at $a$ when $f(a)$ is defined, $\lim_{x\to a}f(x)$ exists, and the two are the same number.

Everything in this section is either a way of failing this test or a consequence of passing it.

Existence through the two sides

$\lim_{x\to a}f(x)=L$ exactly when $\lim_{x\to a^{-}}f(x)=\lim_{x\to a^{+}}f(x)=L$.

Classifying a break is nothing but reading these two numbers and comparing them.

The limit laws

Limits pass through sums, differences, products, constant multiples, powers and roots, and through quotients provided the denominator's limit is not $0$.

The whole algebra of continuous functions is these laws with $L=f(a)$ written in.

$\tfrac{0}{0}$ is an instruction, not a value

When substitution gives $\tfrac{0}{0}$, rewrite: factor and cancel, or multiply by the conjugate.

Every removable break and every parameter fitting question in this section starts with that rewrite.

Infinite limits

$\lim_{x\to a^{+}}f(x)=\infty$ says the values grow past every bound; it is a way of saying the limit does not exist, with the reason attached.

It is the definition of a vertical asymptote and the third kind of break.

The Squeeze Theorem

If $g\le f\le h$ near the point and $g,h$ have the same limit $L$ there, then $\lim f=L$ too.

One interleaved question needs it far from the origin, where $\sin x$ is still trapped between $-1$ and $1$.

Try it yourself first (2 questions)
1§02.0 — the value and the limit are different questions●○○○○

Two questions before the section starts. Getting them wrong is not a problem; it only tells you which of the recalls above to read slowly.

Given
  • $f(x)=\dfrac{x^{2}-9}{x-3}$, with no separate value assigned at $x=3$

Find
  1. (a) What is $f(3)$?

Hint 1/4

Read the question again: it asks for the value of the function at $3$, not for what the function approaches near $3$.

Hint 2/4

A quotient produces a number at $x=a$ only when the denominator is nonzero there. Cancelling a factor changes the formula, and a formula is not allowed to change the domain it came with.

Hint 3/4

At $x=3$ the denominator $x-3$ is $0$, so the recipe never gets to divide.

Hint 4/4

Therefore $f(3)$ is not defined — while $\lim_{x\to3}f(x)=6$, which is a different question with a different answer.

Show solution
Try to evaluate
$f(3)=\frac{9-9}{3-3}=\frac{0}{0}$

$\tfrac00$ is not a number, so no value is produced

Separate value from limit
$\frac{x^{2}-9}{x-3}=x+3\quad(x\neq3)$

the simplified formula agrees with $f$ everywhere except at the one point in question

$\lim_{x\to3}f(x)=6,\qquad\boxed{f(3)\ \text{undefined}}$

the limit exists and the value does not, which is exactly the removable case

Answer $$f(3)\ \text{is undefined};\ \lim_{x\to3}f(x)=6$$
Check

Evaluate at $x=2.999$ and $x=3.001$: $5.999$ and $6.001$. Both are close to $6$ and neither of them is $f(3)$, because the function has nothing to say at $3$.

2§02.0 — a first ●○○○○

The second warm up. If this one is comfortable, the far away block will be quick.

Given
  • $\displaystyle g(x)=\frac{3x+1}{x-4}$

Find
  1. (a) Find $\displaystyle\lim_{x\to\infty}g(x)$.

Hint 1/4

Substitution is not available: both parts grow without bound. The question is which of them grows faster, and by how much.

Hint 2/4

Divide numerator and denominator by the highest power of $x$ present, here $x$ itself, and use $\dfrac{1}{x}\to0$.

Hint 3/4

$\dfrac{3x+1}{x-4}=\dfrac{3+\tfrac{1}{x}}{1-\tfrac{4}{x}}$, and both fractions $\tfrac1x,\tfrac4x$ die.

Hint 4/4

The limit is $\dfrac{3}{1}=3$.

Show solution
Divide by the dominant power
$\frac{3x+1}{x-4}=\frac{3+\tfrac{1}{x}}{1-\tfrac{4}{x}}$

dividing by $x$ keeps both halves finite; dividing by $x^{2}$ would send everything to $0$ and lose the answer

Let the small pieces die
$\frac{1}{x}\to0,\quad\frac{4}{x}\to0\Rightarrow\lim_{x\to\infty}g(x)=\frac{3}{1}=\boxed{3}$

a constant over a growing power dies, which is the one new fact this block needs

Answer $$3$$
Check

At $x=1000$: $\dfrac{3001}{996}\approx3.013$, already within $0.02$ of $3$.

Notation
symbolreads asmeanswatch out
$f\ \text{continuous at}\ a$

f is continuous at a

$f(a)$ exists and $\lim_{x\to a}f(x)=f(a)$

It is a statement about one point, not about the formula. The same formula can be continuous at $3$ and break at $2$.

$x\to\infty$

x increases past every bound

we are describing the tail of the graph, not a point on it

$\infty$ is not a number, so you may not substitute it. Divide first, then let the pieces die.

$y=L$

the line y equals L

a horizontal asymptote when $f(x)\to L$ as $x\to\infty$ or as $x\to-\infty$

The graph is allowed to cross it, even infinitely often. The asymptote is about the tail, not a barrier.

$x=a$

the line x equals a

a vertical asymptote when at least one one sided limit at $a$ is $\pm\infty$

Report the two sides separately; they often carry different signs.

$\sqrt{x^{2}}=\vert x\vert$

the square root of x squared is the absolute value of x

$\vert x\vert=x$ for $x\ge0$ and $\vert x\vert=-x$ for $x<0$

This is where the sign of a limit at $-\infty$ comes from; forgetting it turns $-3$ into $3$.

$[a,b]\ \text{vs}\ (a,b)$

closed interval versus open interval

$[a,b]$ contains its endpoints, $(a,b)$ does not

The Intermediate Value Theorem needs continuity on the closed one and hands back a point of the open one.

Three ways a graph can break, and the one you can repair

The previous section left a test that answers yes or no. An exam asks which kind of no, and whether it can be undone.

Solvable with what we have
  • Compare the two one sided limits with $f(a)$ and answer yes or no.

  • Turn a $\tfrac{0}{0}$ into a number by factoring or by the conjugate.

  • Say that $x=a$ is a vertical asymptote when the values run away there.

Not solvable yet
  • Answer "classify the discontinuity", which is how the question is printed.

  • Decide whether one well chosen value of $f(a)$ mends the function or whether nothing will.

  • Choose the constant $k$ in a piecewise definition so that the two pieces meet.

Take $f(x)=\dfrac{x^{2}-4}{x-2}$ with $f(2)=1$. The test fails at $2$: $\lim_{x\to2}f(x)=4$ while $f(2)=1$. So we write "discontinuous at $2$" and stop.

Why it fails

That is a doctor who says "ill" and stops. A graph missing one dot and a graph torn in two both earn the word discontinuous, and every repair question depends on which one you are holding.

DefinitionDefinition: the three kinds of break at a point
Conditions
  • $f$ is defined on both sides of $a$; at $a$ itself it may or may not be defined

  • each one sided limit either settles on a finite number or runs to $\pm\infty$

$$\boxed{\begin{aligned}&\textbf{removable}:\ \lim_{x\to a}f(x)=L\ \text{exists, but }f(a)\ \text{is missing or}\ \neq L\\&\textbf{jump}:\ \lim_{x\to a^{-}}f(x),\ \lim_{x\to a^{+}}f(x)\ \text{finite and different}\\&\textbf{infinite}:\ \text{at least one of the two is}\ \pm\infty\end{aligned}}$$

Ask the two sides. If they agree on a finite number, the only thing wrong is the dot at $a$, and you may move it: removable. If both are finite but disagree, the graph steps: jump. If either side runs away, the graph has a vertical asymptote there: infinite.

Looks like this, but is not

A candidate for removable: $s(x)=\sin\!\left(\frac{1}{x}\right)$ for $x\neq0$ never leaves the band between $-1$ and $1$, so nothing runs away and there is no visible step. Set $s(0):=0$ and the repair looks done.

Neither one sided limit exists: as $x$ closes in on $0$ the values sweep the band over and over, so there is no single number for $s(0)$ to be. The three names assume each side either settles or runs away; this one does neither.

Classifying the break of (x²−x−6)/(x−3) at x = 3

The formula refuses to produce a number at $x=3$. The question on an exam is never whether it refuses, but which of the three refusals this is.

Given
  • $f(x)=\dfrac{x^{2}-x-6}{x-3}$ for $x\neq3$, and $f$ is not defined at $3$

Find

The type of the break at $x=3$, and the value that repairs it if there is one.

Solution
Make substitution legal again
$\frac{x^{2}-x-6}{x-3}=\frac{(x-3)(x+2)}{x-3}=x+2\quad(x\neq3)$

the factor $x-3$ is exactly what kills the top and the bottom at the same time; cancelling it changes the formula at one single point, and the limit never looks at that point

Read both sides off the reduced formula
$\lim_{x\to3^{-}}(x+2)=5,\qquad\lim_{x\to3^{+}}(x+2)=5$

$x+2$ is a polynomial, so on each side substitution is legal

$\lim_{x\to3}f(x)=5$

the two sides agree on a finite number, which is the first branch of the classification

Compare with the value and name the break
$f(3)\ \text{does not exist}$

the original quotient divides by $0$ there, so there is no third number to compare

$\boxed{\text{removable; put }f(3):=5}$

one dot is missing and the limit says where it belongs, so filling it in is the whole repair

Answer $$\text{removable break at }x=3,\ \text{repaired by }f(3):=5$$
Check

Evaluate the original quotient, not the reduced one, at $x=2.99$ and $x=3.01$: it gives $4.99$ and $5.01$. Both sit within $0.01$ of $5$, which is what a repairable break looks like from the outside.

The entire diagnosis was one cancellation. What survives the cancellation is the height the missing dot should have had.

Two breaks in one formula: (x−1)/(x²−1)

Both suspicious points come from the same denominator, and they get opposite verdicts.

Given
  • $g(x)=\dfrac{x-1}{x^{2}-1}$, defined for $x\neq\pm1$

Find

The type of break at $x=1$ and at $x=-1$, with the one sided limits that justify each.

Solution
Reduce once, and remember what you cancelled
$\frac{x-1}{x^{2}-1}=\frac{x-1}{(x-1)(x+1)}=\frac{1}{x+1}\quad(x\neq1)$

the restriction $x\neq1$ has to travel with the reduced formula; dropping it is how the hole disappears from the answer

The point where the cancelled factor sat
$\lim_{x\to1}\frac{1}{x+1}=\frac{1}{2}$

the reduced formula is a rational function whose denominator is $2$ at $x=1$, so substitution is legal

$\Rightarrow\ \text{removable, repaired by }g(1):=\tfrac12$

finite limit, missing value: the first branch of the classification

The point the denominator still owns
$\lim_{x\to-1^{-}}\frac{1}{x+1}=-\infty,\qquad\lim_{x\to-1^{+}}\frac{1}{x+1}=+\infty$

just left of $-1$ the number $x+1$ is a tiny negative, just right of it a tiny positive, and $1$ divided by a tiny number is huge

$\boxed{\text{infinite break at }x=-1;\ x=-1\ \text{is a vertical asymptote}}$

at least one side runs away, which is the third branch, and that is what the word asymptote records

Answer $$x=1:\ \text{removable, } g(1):=\tfrac12.\qquad x=-1:\ \text{infinite, vertical asymptote}$$
Check

Two test values: $g(-1.01)=-100$ and $g(-0.99)=100$. The signs match the two one sided verdicts and the sizes match the word "runs away", while $g(0.999)=0.50025$ sits calmly next to $\tfrac12$.

One cancellation answered both questions: the factor that died produced the hole, the factor that survived produced the asymptote.

Checkpoint
§02.1 — classifying a break in thirty seconds●●○○○

Thirty seconds, no writing. The formula below is the standard way an exam hides a jump inside an absolute value.

Given
  • $f(x)=\dfrac{\vert x-2\vert}{x-2}$ for $x\neq2$, and $f(2)=0$

Find
  1. (a) Which kind of break sits at $x=2$, and can any value of $f(2)$ repair it?

Hint 1/4

Do not compute anything yet. Ask what the formula does just to the left of $2$ and just to the right of it, where the absolute value has already made up its mind.

Hint 2/4

$\vert x-2\vert=x-2$ when $x>2$ and $\vert x-2\vert=-(x-2)$ when $x<2$; a break is removable only when the two one sided limits agree on one finite number.

Hint 3/4

For $x>2$ the quotient is $\frac{x-2}{x-2}=1$; for $x<2$ it is $\frac{-(x-2)}{x-2}=-1$. The stated value is $f(2)=0$.

Hint 4/4

The sides give $-1$ and $1$, so the break is a jump and no choice of $f(2)$ can close a gap of $2$.

Show solution
Rewrite each side without the bars
$x>2:\ \frac{\vert x-2\vert}{x-2}=\frac{x-2}{x-2}=1$

for $x>2$ the inside is positive, so the bars do nothing

$x<2:\ \frac{\vert x-2\vert}{x-2}=\frac{-(x-2)}{x-2}=-1$

for $x<2$ the inside is negative, so the bars flip the sign

Compare the sides
$\lim_{x\to2^{-}}f(x)=-1\neq1=\lim_{x\to2^{+}}f(x)$

both are finite and they disagree, which is the definition of a jump

$\boxed{\text{jump; not repairable}}$

a single value $f(2)$ can equal one side or the other, never both

Answer $$\text{jump of size }2\ \text{at }x=2$$
Check

The graph is two horizontal rays, at heights $-1$ and $1$. Any horizontal line you draw meets at most one of them, so no dot placed at $x=2$ can be on both.

⚠ Repairing a jump with the midpoint

both sides are finite, so it feels as though a number is merely missing and the average is the fair choice

wrong$f(2):=\tfrac{(-1)+1}{2}=0\ \Rightarrow\ f\ \text{continuous at }2$
right$\lim_{x\to2^{-}}f\neq\lim_{x\to2^{+}}f\ \Rightarrow\ \text{no value of }f(2)\ \text{works}$
⚠ Losing the restriction that comes with a cancellation

the cancelled factor leaves the page and then leaves memory

wrong$\frac{x-1}{x^{2}-1}=\frac{1}{x+1}\ \text{for all }x$
right$\frac{x-1}{x^{2}-1}=\frac{1}{x+1}\quad(x\neq1)$

Continuity on an interval, and the families you may simply quote

One point at a time is not how the question is printed; it is printed about an interval, and it expects an answer that never writes a limit.

TheoremTheorem: continuity is inherited, and endpoints are tested from one side
Conditions
  • $f$ and $g$ are continuous at $a$

  • for the quotient, additionally $g(a)\neq0$

$$\boxed{\begin{aligned}&f\pm g,\ fg,\ cf\ \text{continuous at }a,\ \text{and}\ f/g\ \text{too if}\ g(a)\neq0\\&f\ \text{continuous on}\ [a,b]\iff\text{continuous at every}\ x\in(a,b),\ \lim_{x\to a^{+}}f=f(a),\ \lim_{x\to b^{-}}f=f(b)\end{aligned}}$$

Anything assembled from continuous pieces with $+,-,\times,\div$ is continuous wherever the pieces are and the denominator is not zero. On a closed interval the two endpoints are half tested: only the side that reaches into the interval counts.

Proof

Each line is a limit law with $L=f(a)$ written in. The sum law says $\lim(f+g)=\lim f+\lim g=f(a)+g(a)=(f+g)(a)$, and that last equality is the definition of continuity for $f+g$. The quotient law carries its hypothesis along unchanged, which is where $g(a)\neq0$ comes from.

Looks like this, but is not

$\sqrt{x-2}$ at $x=2$ looks like a break: the graph simply starts there, there is no left hand side at all, and $\lim_{x\to2}$ in the two sided sense does not exist.

The domain is $[2,\infty)$ and $2$ is its left endpoint, so the only test that applies is $\lim_{x\to2^{+}}\sqrt{x-2}=0=\sqrt{2-2}$, which passes. A missing side is not a break; a disagreeing side is.

familycontinuous onthe trap

polynomials

all of $\mathbb{R}$

none — this is the free one

rational $P/Q$

every $x$ with $Q(x)\neq0$

a zero of $Q$ is not automatically an asymptote

$\sqrt[n]{\ \cdot\ }$, $n$ even

wherever the inside is $\ge0$

the endpoint of the domain is tested from one side only

$\sin x$, $\cos x$

all of $\mathbb{R}$

$\tan x$ is not: it breaks at $\tfrac{\pi}{2}+k\pi$

$\vert x\vert$

all of $\mathbb{R}$

continuous everywhere, and still the classic source of jumps once you divide by it

Read the middle column as a promise. If the function in front of you is assembled from these with $+,-,\times,\div$ and composition, you may write "continuous on its domain" and spend the rest of your time on the domain itself, which is where the marks are.

Where is √(x+3)/(x²−4) continuous?

Nothing here needs a limit. The answer is read off the domain, and the only care needed is at the point where the domain stops.

Given
  • $h(x)=\dfrac{\sqrt{x+3}}{x^{2}-4}$

Find

The set of points at which $h$ is continuous, endpoints included.

Solution
Find where the formula produces a number at all
$x+3\ge0\iff x\ge-3$

an even root refuses negative input, so this is the first restriction

$x^{2}-4\neq0\iff x\neq\pm2$

a quotient refuses a zero denominator, and here nothing cancels because the numerator is not a polynomial with those factors

$\text{domain}=[-3,-2)\cup(-2,2)\cup(2,\infty)$

the two restrictions cut the half line at $-2$ and at $2$

Name the reason instead of computing a limit
$\sqrt{x+3}\ \text{continuous on}\ [-3,\infty),\quad x^{2}-4\ \text{continuous on}\ \mathbb{R}$

a root of a polynomial and a polynomial are both on the quotable list

$\Rightarrow\ \frac{\sqrt{x+3}}{x^{2}-4}\ \text{continuous wherever}\ x^{2}-4\neq0$

the quotient rule for continuity, whose only hypothesis is the nonzero denominator

Say what happens at the endpoint of the domain
$\lim_{x\to-3^{+}}h(x)=\frac{0}{5}=0=h(-3)$

there is no left side at $-3$, so only the right hand test is asked, and it passes

$\boxed{h\ \text{is continuous at every point of its domain}}$

for a function assembled this way that sentence is the complete answer, and it is worth full marks

Answer $$\text{continuous on}\ [-3,-2)\cup(-2,2)\cup(2,\infty),\ \text{right continuous at}\ -3$$
Check

Spot check the two ends of the claim: $h(-2.9)=\frac{\sqrt{0.1}}{4.41}\approx0.072$, an entirely ordinary number, while at $x=-2$ the denominator is $0$ and there is no value to compare a limit with. Both match what the answer says.

"Continuous on its domain" is a sentence you may write for anything built from the quotable families — and the marks then sit in the domain, not in the continuity.

Checkpoint
§02.2 — what a closed interval still asks for●●○○○

A graded proof lost one line, and it is always the same line.

Given
  • $f$ is defined at every point of $[0,3]$

  • $f$ is continuous at every point of the open interval $(0,3)$

  • $\lim_{x\to3^{-}}f(x)=f(3)$

Find
  1. (a) Which single statement is still needed before "$f$ is continuous on $[0,3]$" may be written?

Hint 1/4

Continuity on a closed interval is a list of conditions, one per point. Ask which points of $[0,3]$ the three given lines have not yet spoken about.

Hint 2/4

On $[a,b]$: continuity at every interior point, plus $\lim_{x\to a^{+}}f=f(a)$ at the left end and $\lim_{x\to b^{-}}f=f(b)$ at the right end. Endpoints are tested from the inside only.

Hint 3/4

The three given lines cover the interior and the right endpoint $3$. The left endpoint $0$ has not appeared yet.

Hint 4/4

The missing line is the right hand test at $0$: $\lim_{x\to0^{+}}f(x)=f(0)$.

Show solution
Split the interval into interior and ends
$[0,3]=\{0\}\cup(0,3)\cup\{3\}$

the definition treats these three pieces differently, which is why the answer hides in one of them

Match the given lines to the pieces
$(0,3):\ \text{given},\qquad 3:\ \text{given}$

two of the three pieces are already paid for

$\boxed{0:\ \lim_{x\to0^{+}}f(x)=f(0)}$

the left endpoint is tested from inside the interval, so the right hand limit is the one that has to match

Answer $$\lim_{x\to0^{+}}f(x)=f(0)$$
Check

Sanity test with a function that satisfies the three given lines and fails the missing one: $f(x)=1$ for $x=0$ and $f(x)=x$ for $0 < x\le3$. It is continuous on $(0,3)$, it passes the test at $3$, and it is visibly broken at $0$ — so the given three cannot have been enough.

⚠ Demanding a two sided limit at an endpoint

the definition at an interior point is the one everybody memorised, and it gets applied everywhere

wrong$\text{on}\ [a,b]:\ \lim_{x\to a}f(x)=f(a)$
right$\text{on}\ [a,b]:\ \lim_{x\to a^{+}}f(x)=f(a)$
⚠ Calling a function discontinuous at a point outside its domain

the graph is missing there, and missing looks like broken

wrong$\sqrt{x-2}\ \text{is discontinuous at}\ x=1$
right$\sqrt{x-2}\ \text{is undefined at}\ x=1;\ \text{continuity is not asked there}$

Compositions: where the continuity check actually sits

Exam functions come in layers — a root over a quotient, a cosine of a fraction — and the rule for layers has one hypothesis that is checked in a place most people do not look.

TheoremTheorem: moving a limit through a continuous outer function
Conditions
  • $\lim_{x\to a}g(x)=L$ exists

  • $f$ is continuous at $L$ — at $L$, not at $a$

  • nothing is required of $g$ at $a$; it may be undefined there

$$\boxed{\lim_{x\to a}f\bigl(g(x)\bigr)=f\Bigl(\lim_{x\to a}g(x)\Bigr)=f(L)}$$

If the outer function is continuous at the number the inner one is heading for, you may push the limit inside the outer function and evaluate. In particular, a composition of continuous functions is continuous.

Proof

Continuity of $f$ at $L$ says $f(u)$ is close to $f(L)$ whenever $u$ is close to $L$. The inner limit says $g(x)$ is close to $L$ whenever $x$ is close to $a$. Chain the two sentences and you get $f(g(x))$ close to $f(L)$ whenever $x$ is close to $a$, which is the claim.

Looks like this, but is not

Let $f(u)=\dfrac{u}{\vert u\vert}$ for $u\neq0$ with $f(0)=0$, and $g(x)=x^{2}$. Since $g(x)\to0$ as $x\to0$, it is tempting to write $\lim_{x\to0}f(g(x))=f(0)=0$.

$f$ is not continuous at $0$: it equals $-1$ on one side and $1$ on the other. And $g(x)=x^{2}>0$ for every $x\neq0$, so $f(g(x))=1$ for every $x\neq0$ and the true limit is $1$. Pushing a limit inside is a privilege the outer function earns by being continuous at $L$.

A limit under a square root: lim of √((x²−1)/(x−1)) as x → 1

The inside is not even defined at $x=1$. That turns out to be irrelevant, and knowing why is the whole point of this block.

Given
  • $F(x)=\sqrt{\dfrac{x^{2}-1}{x-1}}$ for $x>1$ and for $x<1$ with $x\neq1$

Find

$\lim_{x\to1}F(x)$, with the reason each move is allowed.

Solution
Settle the inner function first
$\frac{x^{2}-1}{x-1}=\frac{(x-1)(x+1)}{x-1}=x+1\quad(x\neq1)$

the inner function has a removable break at $1$, and a removable break has a limit, which is all the outer function will ask for

$\lim_{x\to1}\frac{x^{2}-1}{x-1}=2$

substitution in $x+1$ is legal, so the inner limit is $L=2$

Check the outer function at the inner limit, not at x = 1
$\sqrt{\ \cdot\ }\ \text{is continuous at}\ u=2$

$2>0$, so the square root is continuous there; this is the only hypothesis the rule has, and it is checked at $2$, never at $1$

Carry the limit through
$\lim_{x\to1}\sqrt{\frac{x^{2}-1}{x-1}}=\sqrt{\lim_{x\to1}\frac{x^{2}-1}{x-1}}=\sqrt{2}$

with the outer function continuous at the inner limit, the limit may be moved inside

$\boxed{\sqrt{2}\approx1.41421}$

a number, even though the expression under the root is undefined at the point in question

Answer $$\sqrt{2}$$
Check

Take $x=1.001$: the inside is $2.001$ and the root is $1.41457$, which agrees with $\sqrt{2}=1.41421$ in its first four digits — and the same test at $x=0.999$ gives $1.41386$, closing in from the other side.

The inner function was not continuous at $1$; it is not even defined there. Only its limit mattered, and only the outer function's continuity at that limit.

Checkpoint
§02.3 — where the continuity of the outer function is checked●●○○○

No formulas, only the four numbers an exam gives you when it wants to see whether you know which one the rule uses.

Given
  • $\lim_{x\to2}g(x)=5$

  • $g(2)=7$

  • $f$ is continuous at $5$, and $f(5)=-3$

  • $f(7)=4$

Find
  1. (a) Find $\lim_{x\to2}f\bigl(g(x)\bigr)$.

Hint 1/4

Do not compute; decide which of the four given numbers the rule is entitled to use. The inner function delivers a number to the outer one — which number is that?

Hint 2/4

If $\lim_{x\to a}g(x)=L$ and $f$ is continuous at $L$, then $\lim_{x\to a}f(g(x))=f(L)$. The continuity hypothesis is about $L$.

Hint 3/4

Here $a=2$, $L=5$, and continuity of $f$ is given exactly at $5$, with $f(5)=-3$. The numbers $g(2)=7$ and $f(7)=4$ are the decoys.

Hint 4/4

So the limit is $f(5)=-3$.

Show solution
Read off the inner limit
$L=\lim_{x\to2}g(x)=5$

this is the number the outer function will be asked about

Verify the one hypothesis and conclude
$f\ \text{continuous at}\ 5$

given, and it is given precisely at $L$, which is the point of the question

$\lim_{x\to2}f(g(x))=f(5)=\boxed{-3}$

the rule now applies and no other datum is needed

Answer $$-3$$
Check

Test the reasoning against a function that fits the data: $g(x)=5+(x-2)$ for $x\neq2$ with $g(2)=7$. Then $f(g(x))$ takes values of $f$ at inputs near $5$, never at $7$, so the answer cannot depend on $f(7)$.

⚠ Evaluating the outer function at $a$

$x\to a$ is written on the limit sign, so $a$ is the number in front of your eyes when the outer function asks for input

wrong$\lim_{x\to a}f\bigl(g(x)\bigr)=f(a)$
right$\lim_{x\to a}f\bigl(g(x)\bigr)=f\Bigl(\lim_{x\to a}g(x)\Bigr)$
⚠ Demanding continuity of the inner function too

the rule for a composition of continuous functions is the one people memorise, and it is stronger than what limits need

wrong$\text{need }g\ \text{continuous at }a\ \text{and}\ f\ \text{continuous at }L$
right$\text{need only}\ \lim_{x\to a}g=L\ \text{and}\ f\ \text{continuous at }L$

Proving a solution exists without solving anything

So far continuity has been a licence to substitute. Now it becomes a tool that answers a question substitution cannot touch: does this equation have a solution.

TheoremTheorem: Intermediate Value Theorem
Conditions
  • $f$ is continuous on the closed interval $[a,b]$ — endpoints included, and with no break anywhere inside

  • $N$ is strictly between $f(a)$ and $f(b)$; in particular $f(a)\neq f(b)$

$$\boxed{f\ \text{continuous on}\ [a,b],\ N\ \text{between}\ f(a)\ \text{and}\ f(b)\ \Longrightarrow\ \exists\,c\in(a,b)\ \text{with}\ f(c)=N}$$

An unbroken curve that starts below a horizontal line and ends above it has to touch that line somewhere in between. The special case $N=0$ is the one exams use: a sign change forces a root.

Proof

The textbook does not prove this one either, and neither will we: the proof rests on the completeness of the real numbers, the statement that the number line has no gaps. What is worth keeping is that the theorem collapses the moment continuity fails, and the counterexample below shows how fast.

Looks like this, but is not

The people counter from the opening: $3$ of them in the building at 06:00, $40$ at 14:00, and $20$ is a number between $3$ and $40$. Same shape of argument, so surely there was a moment with exactly $20$ people inside.

The count is not continuous: a bus unloads seventeen people at once and the function jumps from $19$ to $36$ without taking any value in between. The thermometer cannot do that, which is the only reason the same sentence works for temperature and fails here.

Showing x³ = x + 1 has a solution between 1 and 2, then trapping it

There is a formula for the roots of a cubic and nobody wants to use it. The theorem gives the existence in two lines, and halving gives the digits.

Given
  • $f(x)=x^{3}-x-1$, which is $0$ exactly when $x^{3}=x+1$

  • the interval $[1,2]$

Find

First: a proof that a solution exists in $(1,2)$. Then: an interval of width at most $0.15$ that contains it.

Solution
Check the hypothesis before using it
$f\ \text{is a polynomial}\Rightarrow f\ \text{continuous on}\ [1,2]$

polynomials are on the quotable list, so this line is free — but it is the line the grader looks for, and leaving it out costs marks even when the answer is right

Produce the sign change
$f(1)=1-1-1=-1,\qquad f(2)=8-2-1=5$

the two endpoint values are what the theorem compares $N$ with

$-1<0<5$

so $N=0$ lies strictly between $f(1)$ and $f(2)$, which is the second hypothesis

$\Rightarrow\ \exists c\in(1,2):f(c)=0$

the theorem now applies, and this single line is the whole proof

Halve the interval to trap the root
$f(1.5)=3.375-1.5-1=0.875>0\Rightarrow c\in(1,1.5)$

the sign at the midpoint tells you which half still has a sign change; the other half is discarded, not because there is no root there but because nothing forces one

$f(1.25)=1.953125-1.25-1=-0.296875<0\Rightarrow c\in(1.25,1.5)$

the sign flipped, so the root is in the right half this time

$f(1.375)=2.599609-1.375-1=0.224609>0\Rightarrow \boxed{c\in(1.25,1.375)}$

three halvings cut the width from $1$ to $0.125$, and each one costs a single evaluation

Answer $$\text{a solution exists in }(1,2),\ \text{and in fact in }(1.25,1.375)$$
Check

Two fresh test points that the halving never used: $f(1.3)=2.197-1.3-1=-0.103<0$ and $f(1.35)=2.460375-1.35-1=0.110375>0$. They put the root in $(1.3,1.35)$, which sits inside the bracket found above — an independent confirmation of both claims.

Three evaluations bought a bracket of width $0.125$; every further halving costs one more evaluation and halves the width again.

The theorem never produced the root. It produced the right to hunt for one — and the hunt is what turns "somewhere in $(1,2)$" into digits.

Checkpoint
§02.4 — how many solutions the theorem promises●●○○○

The single most expensive misreading of this theorem, in one sentence.

Given
  • $f$ is continuous on $[0,4]$

  • $f(0)=-2$ and $f(4)=6$

Find
  1. (a) True or false: it follows that $f$ has exactly one zero in $(0,4)$.

Hint 1/4

Separate two different claims: that a zero exists, and that only one does. Which of the two does the theorem actually make?

Hint 2/4

The Intermediate Value Theorem is an existence statement: it produces at least one $c$ with $f(c)=N$. It says nothing about how many such $c$ there are.

Hint 3/4

With $f(0)=-2<0<6=f(4)$ and $f$ continuous, the theorem gives at least one zero in $(0,4)$.

Hint 4/4

So the statement is false: at least one, not exactly one.

Show solution
Ask what the theorem promises
$\exists c\in(0,4):f(c)=0$

an existence quantifier, with no uniqueness attached anywhere in the statement

Kill the stronger claim with one example
$f(1)=1,\ f(2)=-1,\ f(3)=1\ \text{(joined by straight segments to the endpoints)}$

this function is continuous, starts at $-2$, ends at $6$, and its sign changes four times

$\Rightarrow\ \boxed{\text{false}}$

one counterexample is enough to sink a universal claim

Answer $$\text{false}$$
Check

The counterexample only used sign changes, so it survives any smoothing: replace the segments by any continuous curve through the same points and the four crossings are still forced.

⚠ Reading "at least one" as "exactly one"

the picture people draw has a single crossing, and the picture gets remembered instead of the statement

wrong$f(a) < 0 < f(b)\Rightarrow f\ \text{has exactly one zero in}\ (a,b)$
right$f(a) < 0 < f(b)\Rightarrow f\ \text{has at least one zero in}\ (a,b)$
⚠ Using the theorem across a vertical asymptote

the endpoint values look perfect, and the break sits in the middle where nobody evaluates anything

wrong$\tan\tfrac{\pi}{4}=1,\ \tan\tfrac{3\pi}{4}=-1\Rightarrow\exists c:\tan c=0$
right$\tan\ \text{is not continuous on}\ \left[\tfrac{\pi}{4},\tfrac{3\pi}{4}\right]\ \text{(break at}\ \tfrac{\pi}{2})$

What the graph does far away

Every limit so far has been taken near a fixed point $a$. Now nothing is fixed: $x$ itself runs away, and we ask what the graph settles on.

DefinitionDefinition and tool: limits at infinity, horizontal asymptotes
Conditions
  • $f$ is defined on an interval that runs out to $\infty$ (or to $-\infty$) — otherwise the question is not asked

  • $n>0$ in the power rule below

$$\boxed{\begin{aligned}&\lim_{x\to\infty}\textcolor{#1f6feb}{f(x)}=L\ \text{or}\ \lim_{x\to-\infty}\textcolor{#1f6feb}{f(x)}=L\ \Longrightarrow\ \textcolor{#6e7781}{y=L}\ \text{is a horizontal asymptote}\\&\lim_{x\to\pm\infty}\frac{1}{x^{n}}=0\quad(n>0)\end{aligned}}$$

A horizontal asymptote is a height the graph settles on far out. The second line is the only new fact needed to compute one: a constant over a growing power dies. Everything else is the limit laws you already have.

Looks like this, but is not

A horizontal asymptote feels like a barrier: the graph gets close to $y=L$ and, being asymptotic, never touches it. For $y=0$ and $f(x)=\dfrac{\sin x}{x}$ that picture predicts a graph that stays strictly above or below the axis far out.

$\dfrac{\sin x}{x}$ is squeezed between $-\tfrac{1}{x}$ and $\tfrac{1}{x}$, so its limit at $\infty$ is $0$ and $y=0$ is a genuine horizontal asymptote — and it also equals $0$ at every $x=k\pi$, crossing the asymptote infinitely often. The asymptote describes the tail, it does not fence it.

A limit at minus infinity with a root: √(9x²+2x)/(4x+1)

Everything about this one is routine except a single minus sign, and that minus sign is what the question is for.

Given
  • $\displaystyle p(x)=\frac{\sqrt{9x^{2}+2x}}{4x+1}$, and $x\to-\infty$

Find

$\displaystyle\lim_{x\to-\infty}p(x)$.

Solution
Pull the dominant power out of the root
$\sqrt{9x^{2}+2x}=\sqrt{x^{2}}\sqrt{9+\tfrac{2}{x}}=\textcolor{#d1690a}{\vert x\vert}\sqrt{9+\tfrac{2}{x}}$

$\sqrt{x^{2}}$ is $\vert x\vert$, never $x$; writing $x$ here is the one move that decides the sign of the whole answer

$x<0\Rightarrow\textcolor{#d1690a}{\vert x\vert=-x},\ \text{so}\ \sqrt{9x^{2}+2x}=-x\sqrt{9+\tfrac{2}{x}}$

we are running to $-\infty$, so $x$ is negative from here on and the bars resolve to a minus

Divide top and bottom by the dominant power
$\frac{\sqrt{9x^{2}+2x}}{4x+1}=\frac{-x\sqrt{9+\tfrac{2}{x}}}{x\left(4+\tfrac{1}{x}\right)}=\frac{-\sqrt{9+\tfrac{2}{x}}}{4+\tfrac{1}{x}}$

dividing by $x$ rather than by $x^{2}$ is the choice that makes both parts finite; $x^{2}$ would send everything to $0$ and tell us nothing

Let the small pieces die
$\frac{2}{x}\to0,\qquad\frac{1}{x}\to0$

each is a constant over a power of $x$, and those go to $0$ at both ends

$\lim_{x\to-\infty}p(x)=\frac{-\sqrt{9}}{4}=\boxed{-\frac{3}{4}}$

the surviving numbers are the leading coefficients, carrying the minus produced by $\vert x\vert$

Answer $$-\frac{3}{4}$$
Check

Numerical check at $x=-1000$: the root is $\sqrt{8\,998\,000}\approx2999.667$ and the denominator is $-3999$, giving $-0.750104$ — the sign and the three digits both match $-\tfrac34$.

Whenever a root of an even power meets $x\to-\infty$, write the bars first and resolve them second. That habit is worth more marks than the rest of the computation put together.

The difference that refuses to vanish: √(x²+6x) − x as x → ∞

Both terms run to $\infty$ and their difference looks like $\infty-\infty$, which is a question, not an answer.

Given
  • $\displaystyle q(x)=\sqrt{x^{2}+6x}-x$, and $x\to\infty$

Find

$\displaystyle\lim_{x\to\infty}q(x)$, and the horizontal asymptote it produces.

Solution
Turn the difference into a quotient
$\sqrt{x^{2}+6x}-x=\frac{\left(\sqrt{x^{2}+6x}-x\right)\left(\sqrt{x^{2}+6x}+x\right)}{\sqrt{x^{2}+6x}+x}=\frac{6x}{\sqrt{x^{2}+6x}+x}$

multiplying by the conjugate is what removes the root; a difference of two huge numbers carries no information, a quotient does

Divide by the dominant power
$\frac{6x}{\sqrt{x^{2}+6x}+x}=\frac{6x}{x\left(\sqrt{1+\tfrac{6}{x}}+1\right)}=\frac{6}{\sqrt{1+\tfrac{6}{x}}+1}$

here $x>0$, so $\vert x\vert=x$ and the bars cost nothing — the opposite of the previous example, and worth noticing

Read off the limit
$\frac{6}{x}\to0\Rightarrow\lim_{x\to\infty}q(x)=\frac{6}{1+1}=\boxed{3}$

the surviving denominator is $2$, not $1$: the root contributes a $1$ of its own

Answer $$3,\ \text{so}\ \textcolor{#6e7781}{y=3}\ \text{is a horizontal asymptote as}\ x\to\infty$$
Check

Numerical check at $x=1000$: $\sqrt{1\,006\,000}\approx1002.9955$, minus $1000$ leaves $2.9955$. Not $0$, and heading for $3$ — which also kills the tempting answer $0$ outright.

Half the marks in this family are lost by writing $\sqrt{x^{2}+6x}\approx x$ and concluding $0$. The approximation is fine; the error it hides is exactly the number being asked for.

Checkpoint
§02.5 — the sign of a limit at minus infinity●●●○○

Thirty seconds. The whole question is which of four numbers carries the right sign.

Given
  • $\displaystyle r(x)=\frac{5x+2}{\sqrt{4x^{2}+1}}$, and $x\to-\infty$

Find
  1. (a) Find $\displaystyle\lim_{x\to-\infty}r(x)$.

Hint 1/4

Both parts grow without bound, so the answer is a ratio of growth rates. Decide first what the root behaves like when $x$ is a large negative number.

Hint 2/4

$\sqrt{4x^{2}+1}=\vert x\vert\sqrt{4+\tfrac{1}{x^{2}}}$, and $\vert x\vert=-x$ for $x<0$. Then divide numerator and denominator by $x$.

Hint 3/4

With $x<0$: numerator over $x$ is $5+\tfrac{2}{x}$, denominator over $x$ is $-\sqrt{4+\tfrac{1}{x^{2}}}$.

Hint 4/4

Letting the small terms die: $\dfrac{5}{-2}=-\dfrac52$.

Show solution
Resolve the root
$\sqrt{4x^{2}+1}=\vert x\vert\sqrt{4+\tfrac{1}{x^{2}}}=-x\sqrt{4+\tfrac{1}{x^{2}}}\quad(x<0)$

on the far left $\vert x\vert=-x$, and this is the only place a sign can enter

Divide by x and finish
$r(x)=\frac{5+\tfrac{2}{x}}{-\sqrt{4+\tfrac{1}{x^{2}}}}$

dividing top and bottom by $x$ turns both halves into constants plus dying terms

$\to\frac{5}{-2}=\boxed{-\frac{5}{2}}$

the dying terms leave the leading coefficients and the minus behind

Answer $$-\frac{5}{2}$$
Check

At $x=-100$: numerator $-498$, denominator $\sqrt{40001}\approx200.0025$, ratio $\approx-2.49$ — the sign and the size both agree with $-\tfrac52$.

⚠ Turning $\sqrt{x^{2}}$ into $x$ on the far left

the identity $\sqrt{x^{2}}=x$ is true for every $x$ anybody ever practises with, because those are all positive

wrong$\sqrt{4x^{2}+1}\approx2x\quad(x\to-\infty)$
right$\sqrt{4x^{2}+1}=\vert x\vert\sqrt{4+\tfrac{1}{x^{2}}}\approx-2x\quad(x\to-\infty)$
⚠ Dropping the small term under a root

$6x$ really is negligible next to $x^{2}$, so the approximation feels safe — but the two huge terms then cancel and the neglected part is all that is left

wrong$\sqrt{x^{2}+6x}-x\approx x-x=0$
right$\sqrt{x^{2}+6x}-x=\frac{6x}{\sqrt{x^{2}+6x}+x}\to3$

Cancel first: holes, vertical asymptotes and the direction of each

The far away picture is fixed; what is left is the handful of points where the formula refuses to produce a number, and whether each refusal is a missing dot or an explosion.

RuleRule: reduce, then classify each zero of the denominator
Conditions
  • $P$ and $Q$ are polynomials

  • the fraction has been reduced to lowest terms first

$$\boxed{\begin{aligned}&\tilde Q(a)=0,\ \tilde P(a)\neq0\ \Rightarrow\ \textcolor{#6e7781}{x=a}\ \text{is a vertical asymptote}\\&(x-a)\ \text{cancelled completely}\ \Rightarrow\ \text{hole at}\ \left(a,\textcolor{#d1690a}{\tfrac{\tilde P(a)}{\tilde Q(a)}}\right)\end{aligned}}$$

Reduce the fraction first. A zero of what is left downstairs, with something nonzero upstairs, makes the values explode: that is an asymptote. A factor that died on both floors leaves one missing dot, whose height is what the reduced formula gives.

Looks like this, but is not

$\dfrac{x^{2}-9}{x-3}$ has a denominator that vanishes at $3$, which is the standard signal for a vertical asymptote, so $x=3$ goes on the list.

The numerator vanishes there too: the fraction is $x+3$ for every $x\neq3$, and its limit at $3$ is $6$. Nothing explodes; there is a hole at $(3,6)$. The signal is a vanishing denominator in a fraction already reduced.

The complete map of h(x) = (x³ + x² − 6x)/(x² − 9)

This is the shape of the asymptote question on a midterm: one function, and every special point has to be named with the limit that justifies it.

Given
  • $h(x)=\dfrac{x^{3}+x^{2}-6x}{x^{2}-9}$

Find

Holes with their coordinates, vertical asymptotes with both one sided limits, and the behaviour as $x\to\pm\infty$.

Solution
Factor and reduce
$h(x)=\frac{x(x^{2}+x-6)}{(x-3)(x+3)}=\frac{x(x+3)(x-2)}{(x-3)(x+3)}$

factor before deciding anything; the whole classification is a comparison of factors

$=\textcolor{#d1690a}{\frac{x(x-2)}{x-3}}\quad(x\neq-3)$

the factor $x+3$ died on both sides, so it cannot produce an asymptote — but the restriction stays

The cancelled factor gives a hole
$\lim_{x\to-3}h(x)=\frac{(-3)(-5)}{-6}=-\frac{5}{2}$

the reduced formula is continuous at $-3$, so its value there is the limit

$\Rightarrow\ \text{hole at}\ \left(-3,-\tfrac52\right)$

a single missing dot, not an asymptote, because nothing grows

The surviving factor gives the asymptote, with a direction on each side
$x\to3:\ \text{numerator}\ x(x-2)\to3\neq0,\ \text{denominator}\to0$

a nonzero number over something shrinking to zero is what makes values explode

$\lim_{x\to3^{-}}h(x)=\textcolor{#d1690a}{-\infty},\qquad\lim_{x\to3^{+}}h(x)=\textcolor{#d1690a}{+\infty}$

just left of $3$ the denominator is a small negative and the numerator is positive, so the quotient is a large negative; just right of it the denominator flips sign

$\Rightarrow\ \textcolor{#6e7781}{x=3}\ \text{is a vertical asymptote}$

reported with both sides, because the two are different

Far away: divide, do not guess
$\frac{x(x-2)}{x-3}=\frac{x^{2}-2x}{x-3}=x+1+\frac{3}{x-3}$

the degree on top is one more than the degree below, so there is no horizontal asymptote; long division exposes what there is instead

$h(x)-(x+1)=\frac{3}{x-3}\to0\quad(x\to\pm\infty)$

the gap between the graph and the line $y=x+1$ dies, which is exactly what a means

$\boxed{\text{hole }\left(-3,-\tfrac52\right),\ \text{VA }x=3,\ \text{slant }y=x+1}$

three answers, each with the limit that earns it

Answer $$\text{hole at}\ \left(-3,-\tfrac52\right),\quad \textcolor{#6e7781}{x=3}\ \text{vertical asymptote},\quad y=x+1\ \text{slant asymptote}$$
Check

Three numerical probes, none of them reused from the computation: $h(2.9)=-26.1$ and $h(3.1)=34.1$ confirm the two directions at the asymptote, and $h(100)=101.03$ sits $0.03$ above the line $y=x+1$ at $x=100$, exactly the $\tfrac{3}{x-3}$ the division predicted.

One factorisation carried all four answers; nothing here needed a second idea.

Checkpoint
§02.6 — which line is the vertical asymptote●●○○○

Thirty seconds. Two of the four candidates are produced by the same denominator, and only one of them survives.

Given
  • $\displaystyle u(x)=\frac{x^{2}-4}{x^{2}-x-2}$

Find
  1. (a) Which line is a vertical asymptote of $u$?

Hint 1/4

Do not test the denominator's zeros one by one yet. First ask whether the numerator shares any of them.

Hint 2/4

Reduce the fraction to lowest terms; a zero of the reduced denominator that is not a zero of the reduced numerator gives a vertical asymptote, a cancelled factor gives a hole.

Hint 3/4

$x^{2}-4=(x-2)(x+2)$ and $x^{2}-x-2=(x-2)(x+1)$, so the reduced form is $\dfrac{x+2}{x+1}$ for $x\neq2$.

Hint 4/4

The reduced denominator vanishes at $-1$ only, and the reduced numerator is $1$ there, so $x=-1$ is the vertical asymptote.

Show solution
Factor and reduce
$u(x)=\frac{(x-2)(x+2)}{(x-2)(x+1)}=\frac{x+2}{x+1}\quad(x\neq2)$

the common factor is the whole story: it decides which candidate is a hole

Classify the two candidates
$x=2:\ \lim_{x\to2}u(x)=\frac{4}{3}\ \Rightarrow\ \text{hole}$

the limit is finite, so nothing explodes there

$x=-1:\ \text{numerator}\to1\neq0,\ \text{denominator}\to0\ \Rightarrow\ \boxed{x=-1}$

nonzero over zero is the signature of a vertical asymptote

Answer $$x=-1$$
Check

Probe the two points: $u(1.99)=\tfrac{3.99}{2.99}\approx1.3344$, near $\tfrac43$, while $u(-0.99)=101$ — only one of the two candidates makes the values explode.

⚠ Listing every zero of the original denominator

the denominator is where asymptotes come from, and reducing takes an extra minute that nobody has in an exam

wrong$\frac{x^{2}-9}{x-3}:\ x=3\ \text{is a vertical asymptote}$
right$\frac{x^{2}-9}{x-3}=x+3\ (x\neq3):\ \text{hole at}\ (3,6)$
⚠ Giving one infinity for both sides

the graph explodes, and one symbol feels like enough to say so

wrong$\lim_{x\to3}\frac{x(x-2)}{x-3}=+\infty$
right$\lim_{x\to3^{-}}\frac{x(x-2)}{x-3}=-\infty,\qquad\lim_{x\to3^{+}}\frac{x(x-2)}{x-3}=+\infty$
Conventions used in this section

Read once. These are the rules every answer below is written to, and the ones a grader assumes without saying so.

  1. Where continuity is asked

    Only at points of the domain. If $a$ is not in the domain we do not hand in "$f$ is discontinuous at $a$" as a verdict about $f$; we describe the break of the graph there and, when the limit exists, the value that would repair it.

  2. DNE versus $\pm\infty$

    $\lim=\infty$ says the limit does not exist and supplies the reason: the values grow past every bound. For a jump we write DNE, because neither side runs away and no single symbol describes the failure.

  3. Radians, always

    $\sin$, $\cos$ and $\tan$ take radians. $\cos1$ is the cosine of one radian, about $0.5403$; in degrees it would be $0.9998$, and a sign change would disappear.

  4. Closed and open intervals

    $[a,b]$ contains its endpoints and is what the Intermediate Value Theorem demands; the point $c$ it returns lives in the open $(a,b)$.

  5. Both sides, every time

    At a vertical asymptote the two one sided limits are reported separately. They carry different signs more often than not.

Where it goes wrong
  • Writing "$1/x$ is discontinuous at $0$" as a statement about the function, when $0$ is not in its domain at all.

  • Answering $\infty$ where the two sides run in opposite directions — that case is DNE plus two separate one sided statements.

  • Evaluating $\cos1$ in degrees, which turns $0.5403$ into $0.9998$ and destroys a sign change you were relying on.

End behaviour of a quotient of polynomials

As soon as $x\to\pm\infty$ appears above a fraction — and only after any root has been dealt with.

  1. Degree below wins

    $\deg P<\deg Q$: the limit is $0$, and $y=0$ is the horizontal asymptote at both ends.

  2. Degrees equal

    The limit is the ratio of the leading coefficients, $\dfrac{a_{n}}{b_{m}}$ — the constant terms play no part — and it is the same at both ends.

  3. Degree above wins by exactly one

    No horizontal asymptote. Divide: $\dfrac{P}{Q}=mx+b+\dfrac{r(x)}{Q(x)}$ with the last part dying, so $y=mx+b$ is a slant asymptote.

  4. Degree above wins by more

    Neither a horizontal nor a slant asymptote: the values run to $\pm\infty$, and the leading terms plus the side decide which sign.

  5. If a root is involved, do not count degrees

    Pull $\vert x\vert$ out of the root first and resolve the bars using the side you are on; only then divide.

Where it goes wrong
  • Applying the degree rule to $\sqrt{x^{2}+3x}-x$, which is not a quotient until the conjugate makes it one.

  • Using constant terms instead of leading coefficients in the equal degree case.

  • Reporting the same sign at both ends when the degree gap is odd.

The asymptote map, in five moves

When the question says "find all asymptotes", "discuss the graph" or "sketch" and hands you a rational function.

  1. Factor both parts

    Nothing can be classified before the factors are visible.

  2. Cancel, and record what died

    Every completely cancelled factor $(x-a)$ becomes a hole; its height is the reduced formula evaluated at $a$.

  3. Read the reduced denominator

    Each of its zeros is a vertical asymptote. Give both one sided limits, using the sign of every surviving factor.

  4. Compare degrees for the two ends

    Use the degree box above on the reduced fraction, not on the original one.

  5. State each answer with its limit

    "$x=3$ is a vertical asymptote" earns little on its own; the marks sit in the limit statements that justify it.

Where it goes wrong
  • Cancelling and then forgetting to write the excluded point, so the hole vanishes from the answer.

  • Comparing degrees on the original fraction after a cancellation has changed both of them.

  • Listing an asymptote with no supporting limit, which is the difference between full marks and half.

The theorem applies: cos x = x has a solution in (0, 1)

An equation with no algebraic solution at all, settled in three lines.

Given
  • $f(x)=\cos x-x$ on $[0,1]$, with $x$ in radians

Find

A proof that $\cos x=x$ for some $x\in(0,1)$.

Solution
Check continuity on the closed interval
$\cos x\ \text{and}\ x\ \text{are continuous on}\ \mathbb{R}\Rightarrow f\ \text{continuous on}\ [0,1]$

a difference of two quotable families, so the hypothesis costs one line

Produce the sign change
$f(0)=\cos0-0=1>0$

the left endpoint value

$f(1)=\cos1-1\approx0.5403-1=-0.4597<0$

radians, not degrees — in degrees this number would be positive and the argument would collapse

$\Rightarrow\ \exists c\in(0,1):f(c)=0,\ \text{i.e.}\ \boxed{\cos c=c}$

$N=0$ lies between $-0.4597$ and $1$, so the theorem applies

Answer $$\cos x=x\ \text{has a solution in}\ (0,1)$$
Check

Two extra probes: $f(0.7)=0.7648-0.7=0.0648>0$ and $f(0.8)=0.6967-0.8=-0.1033<0$, so the solution is in fact between $0.7$ and $0.8$ — inside the interval the theorem claimed.

The theorem does not apply: tan x between π/4 and 3π/4

The endpoint values look perfect. The interval is the problem.

Given
  • $g(x)=\tan x$ on $\left[\tfrac{\pi}{4},\tfrac{3\pi}{4}\right]$

Find

Whether the sign change forces $\tan c=0$ somewhere in between.

Solution
Write down the tempting argument
$g\!\left(\tfrac{\pi}{4}\right)=1,\qquad g\!\left(\tfrac{3\pi}{4}\right)=-1$

a textbook sign change, and $N=0$ sits between them

$\text{"therefore"}\ \exists c:\tan c=0$

this is the line to refuse, and refusing it needs the hypothesis, not the conclusion

Test the hypothesis instead of the conclusion
$\lim_{x\to(\pi/2)^{-}}\tan x=+\infty,\qquad\lim_{x\to(\pi/2)^{+}}\tan x=-\infty$

$\tfrac{\pi}{2}$ lies inside the interval and $\tan$ is not even defined there, so there is no continuity on the closed interval

$\boxed{\text{the theorem does not apply, and its conclusion is false here}}$

the graph leaves through the top and comes back from the bottom, which is exactly the move continuity forbids

Answer $$\text{no conclusion; and in fact}\ \tan c\neq0\ \text{for all}\ c\in\left[\tfrac{\pi}{4},\tfrac{3\pi}{4}\right]$$
Check

Independently: $\tan c=0$ happens only at $c=k\pi$, and the interval is about $[0.785,2.356]$, which contains no multiple of $\pi$. So the failure is not a technicality — the promised point genuinely does not exist.

Both pairs of endpoint values change sign; only the first function is continuous on the whole closed interval, and that is the only difference that matters.

How to tell them apart

Before quoting the theorem, sweep the closed interval for a point where the formula divides by zero, changes rule or leaves its domain. Endpoint values can never reveal a break in the middle.

Scaffolding comes off
The common skeleton
  1. Name the junction: the single $x$ where the rule changes, and the two pieces that meet there.

  2. Compute the limit from the left using only the formula valid to the left of it.

  3. Compute the limit from the right the same way.

  4. Read $f$ at the junction from the piece whose inequality contains it.

  5. Force the three numbers to be equal and solve; if no choice of the unknown can do it, name the break instead.

1 · fully worked

Fitting a piece to a hole: choosing a in a two piece definition

The two pieces are fine on their own. The only question is whether they meet.

Given
  • $f(x)=\begin{cases}\dfrac{x^{2}-4}{x-2}&x<2\\[4pt] ax+1&x\ge2\end{cases}$

Find

The value of $a$ that makes $f$ continuous at $x=2$.

Solution
Name the junction and the two pieces
$x=2:\ \text{left piece}\ \frac{x^{2}-4}{x-2},\quad\text{right piece}\ ax+1$

the rule changes exactly once, so exactly one condition will be imposed

Limit from the left, using only the left formula
$\frac{x^{2}-4}{x-2}=\frac{(x-2)(x+2)}{x-2}=x+2\quad(x\neq2)$

the left formula is a $\tfrac00$ at the junction, and cancelling is legal because $x\neq2$ on that side anyway

$\lim_{x\to2^{-}}f(x)=4$

substitution in $x+2$

Limit from the right and the value, both from the right formula
$\lim_{x\to2^{+}}f(x)=2a+1,\qquad f(2)=2a+1$

the inequality $x\ge2$ includes the junction, so this piece owns both the right limit and the value — that is why only one equation appears

Force the three numbers to agree
$2a+1=4$

continuity at $2$ is exactly the statement that the left limit, the right limit and the value are one number

$\boxed{a=\tfrac32}$

one unknown, one equation, one answer

Answer $$a=\frac{3}{2}$$
Check

Test the fitted function on both sides of the junction: $f(1.99)=3.99$ from the left piece and $f(2.01)=1.5(2.01)+1=4.015$ from the right piece. Both sit next to $f(2)=4$, which is what a repaired junction looks like numerically.

Every question of this family is the same five moves; only the number of unknowns and the number of junctions changes.

2 · you write the reasoning

Same skeleton, easier numbers, and this time the reasons are yours to write. Find $b$ so that $f$ is continuous at $x=1$, where $f(x)=\begin{cases}x^{2}+1&x\le1\\ 4x-b&x>1\end{cases}$. Write the reason for each step before opening it.

  1. $\lim_{x\to1^{-}}f(x)=1^{2}+1=2$

    reasoning

    The left piece is a polynomial, so the left limit is substitution — and the piece that carries $x\le1$ is the one valid on that side.

  2. $f(1)=1^{2}+1=2$

    reasoning

    The value is read from the piece whose inequality contains the junction, which here is the same left piece. That is why this problem has one equation rather than two.

  3. $\lim_{x\to1^{+}}f(x)=4-b$

    reasoning

    The right limit uses the other formula only; $b$ is a constant, so it survives the limit untouched.

  4. $4-b=2$, so $b=2$

    reasoning

    Continuity is the statement that the three numbers coincide; two of them are already equal, so a single equation determines $b$.

3 · find the buried error

Two junctions, two unknowns, and a solution written by a student who was in a hurry. Exactly two of the five steps are wrong. Find them.

$f(x)=\begin{cases}2x+a&x<-1\\ x^{2}-2&-1\le x\le2\\ bx+1&x>2\end{cases}$, and $f$ is to be continuous on all of $\mathbb{R}$.

  1. Step 1. The rule changes at $x=-1$ and at $x=2$, so continuity has to be forced at both junctions.

  2. Step 2. $\lim_{x\to-1^{-}}f(x)=2(-1)+a=-2+a$ and $\lim_{x\to-1^{+}}f(x)=(-1)^{2}-2=-1$.

  3. Step 3. Setting them equal: $-2+a=-1$, so $a=-1$.

  4. Step 4. At $x=2$ the middle piece gives $2^{2}-2=2$ and the right piece gives $2b+1$, so $2b+1=2$ and $b=\tfrac12$.

  5. Step 5. Both junctions now match, so $f$ is continuous everywhere; in particular the two sided limit at $x=-1$ equals $f(-1)=2(-1)+a$.

the two buried errors (2)
⚠ step 3

The equation $-2+a=-1$ is solved as $a=-1$; it gives $a=1$.

Isolating the unknown by moving a term across and keeping its old sign is the single most common slip in a hurried exam script, and it hides well because the arithmetic looks like one step.

right

$-2+a=-1\Rightarrow a=-1+2=1$. Check it: $2(-1)+1=-1$, which matches the middle piece at $-1$.

⚠ step 5

$f(-1)$ is read from the piece defined for $x<-1$, which does not contain $-1$.

At a junction two formulas are in view and only the inequality says which one owns the point; the strict inequality is easy to read as if it included the endpoint.

right

The piece with $-1\le x\le2$ owns the junction, so $f(-1)=(-1)^{2}-2=-1$. With the corrected $a=1$ the left limit agrees with it, which is what makes the function continuous there.

4 · the bare problem
§02.2 — two junctions, two unknowns, no scaffolding●●●●○

The bare version of the skeleton. Two pieces meet a middle piece, and both meetings have to be forced at once.

Given
  • $f(x)=\begin{cases}\dfrac{x^{2}-1}{x-1}&x<1\\[4pt] ax+b&1\le x<3\\[2pt] x^{2}-5&x\ge3\end{cases}$

Find
  1. (a) Find the values of $a$ and $b$ that make $f$ continuous on all of $\mathbb{R}$.

  2. (b) State the resulting formula on $[1,3)$ and check it against both junctions.

Hint 1/4

You are looking for two numbers, so you need two equations. Ask which junctions can produce them, and which piece owns the value at each junction.

Hint 2/4

At a junction, continuity means left limit $=$ right limit $=$ value. The piece whose inequality contains the junction supplies the value.

Hint 3/4

At $x=1$: the left piece is $\frac{x^{2}-1}{x-1}=x+1$ for $x\neq1$, so the left limit is $2$, while the middle piece gives $a+b$. At $x=3$: the middle piece gives $3a+b$ and the right piece gives $3^{2}-5=4$.

Hint 4/4

Solving $a+b=2$ and $3a+b=4$ gives $a=1$ and $b=1$.

Show solution
First junction, x = 1
$\lim_{x\to1^{-}}\frac{x^{2}-1}{x-1}=\lim_{x\to1^{-}}(x+1)=2$

the left piece is a $\tfrac00$ at the junction; cancelling is legal on that side because $x\neq1$ there

$f(1)=a+b\ \text{and}\ \lim_{x\to1^{+}}f(x)=a+b$

the middle piece owns the junction, since its inequality is $1\le x$

$\Rightarrow\ a+b=2$

the first equation

Second junction, x = 3
$\lim_{x\to3^{-}}f(x)=3a+b,\qquad f(3)=3^{2}-5=4$

this time the right piece owns the junction, because its inequality is $x\ge3$

$\Rightarrow\ 3a+b=4$

the second equation

Solve the pair
$(3a+b)-(a+b)=4-2\Rightarrow2a=2\Rightarrow a=1$

subtracting kills $b$, which is why elimination is cheaper here than substitution

$b=2-a=1$

back into the first equation

$\boxed{a=1,\ b=1}$

and the middle piece is $x+1$, the same formula the left piece reduces to

Answer $$a=1,\ b=1$$
Check

The fitted middle piece is $x+1$, which is exactly what the left piece equals for $x\neq1$ — so the first two pieces are one straight line, and at $x=3$ it gives $4=3^{2}-5$. Both junctions check out without reusing the equations that produced them.

When the two junction equations share an unknown, eliminate rather than substitute: the system is linear and subtraction removes $b$ in one line.

Full exam-style question

Exam-style: the full picture of (2x² − x − 3)/(x² − 1), and a solution huntexam format

One function, four questions — the standard shape of the asymptote item on a midterm, with the existence part attached at the end.

Given
  • $f(x)=\dfrac{2x^{2}-x-3}{x^{2}-1}$

Find

(a) the domain and any holes, with coordinates; (b) every vertical asymptote with both one sided limits; (c) the horizontal asymptote with the limit that supports it; (d) a proof that $f(x)=0$ has a solution in $(1.1,2)$, and the reason $[1,2]$ would not have been a legal interval for that argument.

Solution
(a) Factor, reduce, and read the domain
$2x^{2}-x-3=(2x-3)(x+1),\qquad x^{2}-1=(x-1)(x+1)$

both factorisations are needed before anything can be classified

$f(x)=\frac{(2x-3)(x+1)}{(x-1)(x+1)}=\frac{2x-3}{x-1}\quad(x\neq-1)$

the factor $x+1$ cancels, so $-1$ is a hole rather than an asymptote, and the restriction has to travel with the reduced formula

$\text{domain}=\mathbb{R}\setminus\{-1,1\};\quad\lim_{x\to-1}f(x)=\frac{-5}{-2}=\frac52\Rightarrow\text{hole at}\ \left(-1,\tfrac52\right)$

the reduced formula is continuous at $-1$, so its value there is the height of the missing dot

(b) The surviving zero of the denominator
$x\to1:\ 2x-3\to-1\neq0,\quad x-1\to0$

nonzero over zero: the values have to explode

$\lim_{x\to1^{-}}f(x)=\frac{-1}{0^{-}}=+\infty,\qquad\lim_{x\to1^{+}}f(x)=\frac{-1}{0^{+}}=-\infty$

left of $1$ the denominator is a small negative and the numerator is negative, so the quotient is a large positive; on the right the denominator flips

$\Rightarrow\ x=1\ \text{is a vertical asymptote}$

stated with both sides, because they differ

(c) The two ends
$f(x)=\frac{2x-3}{x-1}=\frac{2-\tfrac{3}{x}}{1-\tfrac{1}{x}}\ \longrightarrow\ \frac{2}{1}=2\quad(x\to\pm\infty)$

equal degrees after the cancellation, so the ratio of leading coefficients is the answer; dividing by $x$ is what makes that visible

$\Rightarrow\ y=2\ \text{is a horizontal asymptote at both ends}$

one line, valid on the left and on the right

(d) Existence, and the interval that would have been illegal
$f\ \text{is continuous on}\ [1.1,2]$

on that interval the denominator $x-1$ is at least $0.1$, so nothing divides by zero and the quotient rule for continuity applies

$f(1.1)=\frac{2(1.1)-3}{0.1}=\frac{-0.8}{0.1}=-8,\qquad f(2)=\frac{1}{1}=1$

the endpoint values, which is all the theorem compares

$-8<0<1\Rightarrow\exists c\in(1.1,2):f(c)=0$

the sign change plus continuity is the whole argument

$\boxed{[1,2]\ \text{is illegal: }f\ \text{is not even defined at}\ x=1}$

the vertical asymptote sits at the left endpoint, so continuity on the closed interval fails before the theorem is reached

Answer $$\text{hole }\left(-1,\tfrac52\right);\ \text{VA }x=1;\ \text{HA }y=2;\ \text{a zero exists in }(1.1,2)$$
Check

Part (d) can be checked exactly, which is rare: $2x-3=0$ gives $x=1.5$, and $1.5$ does lie in $(1.1,2)$. The asymptote claim survives its own probe too: $f(0.9)=\frac{-1.2}{-0.1}=12$ and $f(1.1)=-8$, large and of opposite signs.

One factorisation served parts (a), (b) and (c); part (d) needed nothing but two substitutions and one sentence about the interval.

The trap in (d) is the habit of taking the interval the question suggests. Choose the interval yourself, and choose it so that the hypothesis is true.

Practice

A · concept 3 questions
1§02.1 — two finite sides are not enough●●○○○

A sentence that appears in scripts every term, and is wrong by one word.

Given
  • $f$ is defined near $a$, and both $\lim_{x\to a^{-}}f(x)$ and $\lim_{x\to a^{+}}f(x)$ exist and are finite

Find
  1. (a) True or false: it follows that $f$ is continuous at $a$.

Hint 1/4

Write down the full definition of continuity at $a$ and count how many conditions the given sentence has actually supplied.

Hint 2/4

Continuity at $a$ needs three things: $f(a)$ defined, the two sided limit existing, and the two being equal. Two finite one sided limits deliver at most part of the second.

Hint 3/4

The sentence never says the two sides are equal, and never mentions $f(a)$ at all.

Hint 4/4

False — a jump has two finite one sided limits, and so does a removable break.

Show solution
A jump satisfies the hypothesis
$f(x)=\frac{\vert x\vert}{x}\ (x\neq0),\ f(0)=0:\ \lim_{x\to0^{-}}=-1,\ \lim_{x\to0^{+}}=1$

both sides finite, and the function is visibly broken at $0$

So does a removable break
$g(x)=\frac{x^{2}-4}{x-2}\ (x\neq2):\ \text{both sides}\ \to4,\ g(2)\ \text{undefined}$

both sides finite and equal, and still no continuity, because the third condition has no value to test

$\Rightarrow\ \boxed{\text{false}}$

two independent ways to satisfy the hypothesis and fail the conclusion

Answer $$\text{false}$$
Check

The two counterexamples fail for different reasons — disagreeing sides in one, a missing value in the other — so the claim is not rescued by patching either hole.

2§02.5 — whether an asymptote can be crossed●●○○○

The word asymptote suggests a fence. The definition says something weaker.

Given
  • $f$ has the horizontal asymptote $y=L$ as $x\to\infty$

Find
  1. (a) True or false: it follows that the graph of $f$ can meet the line $y=L$ at infinitely many points.

Hint 1/4

Read what the asymptote statement actually claims. Is it a claim about every $x$, or about the behaviour of $f(x)$ as $x$ grows?

Hint 2/4

$y=L$ is a horizontal asymptote when $\lim_{x\to\infty}f(x)=L$. A limit constrains the tail, and puts no restriction on any individual value.

Hint 3/4

$\dfrac{\sin x}{x}$ is squeezed between $-\tfrac1x$ and $\tfrac1x$, so its limit at $\infty$ is $0$, and it equals $0$ at every $x=k\pi$.

Hint 4/4

True: that function crosses its asymptote infinitely often.

Show solution
Produce the limit
$-\frac{1}{x}\le\frac{\sin x}{x}\le\frac{1}{x}\quad(x>0)$

$\sin$ stays inside $[-1,1]$ however large $x$ gets, so dividing by $x$ traps the quotient

$\Rightarrow\lim_{x\to\infty}\frac{\sin x}{x}=0$

both bounds tend to $0$, so the Squeeze Theorem applies and $y=0$ is an asymptote

Produce the crossings
$\frac{\sin x}{x}=0\iff\sin x=0\iff x=k\pi$

the quotient vanishes exactly where the numerator does, and there are infinitely many such points to the right

$\boxed{\text{true}}$

one function both settles on the line and meets it endlessly

Answer $$\text{true}$$
Check

The crossings are not an artefact of the example: any function of the form $\frac{h(x)}{x}$ with $h$ oscillating through $0$ does the same, so the conclusion does not depend on $\sin$ in particular.

3§02.4 — when the theorem says nothing at all●●●○○

The hypothesis has two halves, and the second one is the half people forget to check.

Given
  • $f$ is continuous on $[-1,3]$

  • $f(-1)=4$ and $f(3)=4$

Find
  1. (a) Which conclusion does the Intermediate Value Theorem allow here?

Hint 1/4

Write out the hypothesis in full and compare it with what you were given. One of the two halves is in trouble.

Hint 2/4

The theorem needs a value $N$ strictly between $f(a)$ and $f(b)$. If the two endpoint values are equal, no such $N$ exists.

Hint 3/4

Here $f(-1)=f(3)=4$, so there is no number strictly between them, and the theorem never starts.

Hint 4/4

The theorem yields nothing here — which does not mean $f$ has no zero, only that this tool cannot produce one.

Show solution
Check the hypothesis about N
$\text{no}\ N\ \text{lies strictly between}\ 4\ \text{and}\ 4$

the interval of admissible values is empty, so the theorem produces no $c$ at all

Kill the tempting conclusions with one function
$f(x)=4\ \text{for all}\ x$

continuous, matches both given values, has no zero and takes no value other than $4$

$\boxed{\text{the theorem yields nothing}}$

any conclusion that this example contradicts cannot follow from the hypotheses

Answer $$\text{no conclusion}$$
Check

The constant function is not a special case: $f(x)=4+\sin\!\left(\frac{\pi(x+1)}{2}\right)$ also fits the data and is not constant, so neither of the two tempting readings survives.

B · computation 5 questions
1§02.1 — two candidates, two different verdicts●●●○○

The denominator vanishes twice and the numerator only helps at one of the two points.

Given
  • $\displaystyle f(x)=\frac{x^{2}-5x+6}{x^{2}-4}$

Find
  1. (a) Classify the break at $x=2$ and, if it is removable, give the value that repairs it.

  2. (b) Classify the break at $x=-2$, giving both one sided limits.

  3. (c) State the equation of every vertical asymptote of $f$.

Hint 1/4

Neither point can be classified from the denominator alone. Ask, at each of the two points, whether the numerator vanishes as well.

Hint 2/4

Factor both parts and cancel. A factor that dies on both floors leaves a hole; a surviving zero of the denominator with a nonzero numerator gives an infinite break.

Hint 3/4

$x^{2}-5x+6=(x-2)(x-3)$ and $x^{2}-4=(x-2)(x+2)$, so $f(x)=\dfrac{x-3}{x+2}$ for $x\neq2$. At $-2$ the reduced numerator is $-5$.

Hint 4/4

At $2$: removable, repaired by $f(2):=-\tfrac14$. At $-2$: infinite, with $+\infty$ from the left and $-\infty$ from the right, so $x=-2$ is the only vertical asymptote.

Show solution
Factor and reduce
$f(x)=\frac{(x-2)(x-3)}{(x-2)(x+2)}=\frac{x-3}{x+2}\quad(x\neq2)$

the common factor is the whole difference between the two candidates

The cancelled point
$\lim_{x\to2}\frac{x-3}{x+2}=\frac{-1}{4}=-\frac14$

the reduced formula is continuous at $2$, so substitution gives the limit

$\Rightarrow\ \text{removable, repaired by}\ f(2):=-\tfrac14$

finite limit, no value: the first branch of the classification

The surviving point, with its two directions
$x\to-2:\ x-3\to-5\neq0,\qquad x+2\to0$

nonzero over zero, so the values explode and only the sign is left to decide

$\lim_{x\to-2^{-}}f(x)=\frac{-5}{0^{-}}=+\infty,\qquad\lim_{x\to-2^{+}}f(x)=\frac{-5}{0^{+}}=-\infty$

left of $-2$ the denominator is a small negative, so a negative over a negative is a large positive; on the right it flips

$\boxed{x=-2\ \text{is the only vertical asymptote}}$

the other candidate cancelled, so it never had a chance

Answer $$\text{hole at}\ \left(2,-\tfrac14\right);\quad x=-2:\ \text{infinite break}$$
Check

Probe both points with numbers: $f(1.99)=\frac{-1.01}{3.99}\approx-0.2531$, next to $-\tfrac14$, while $f(-2.01)=\frac{-5.01}{-0.01}=501$ and $f(-1.99)=\frac{-4.99}{0.01}=-499$. Calm at one point, explosive at the other, with the signs as claimed.

A zero of the denominator is a candidate, never a verdict. The numerator casts the deciding vote.

2§02.1 — choosing the value that repairs a root●●●○○

A parameter question wearing a conjugate problem as a disguise.

Given
  • $f(x)=\dfrac{\sqrt{x+7}-3}{x-2}$ for $x\neq2$, and $f(2)=k$

Find
  1. (a) Find the value of $k$ that makes $f$ continuous at $x=2$.

Hint 1/4

Continuity at $2$ forces $k$ to be one specific number. Which number is it, before you compute anything?

Hint 2/4

$k$ must equal $\lim_{x\to2}f(x)$. Substitution gives $\tfrac00$ because of the root, and the standard repair for a root in a $\tfrac00$ is multiplying by the conjugate.

Hint 3/4

Multiply top and bottom by $\sqrt{x+7}+3$: the numerator becomes $(x+7)-9=x-2$, and the given data is $f(x)=\frac{\sqrt{x+7}-3}{x-2}$ with $f(2)=k$.

Hint 4/4

After cancelling $x-2$ the limit is $\dfrac{1}{\sqrt{9}+3}=\dfrac16$, so $k=\tfrac16$.

Show solution
Say what continuity demands
$k=f(2)=\lim_{x\to2}f(x)$

the three part test fixes $k$ before any computation, so the problem is really a limit problem

Bring the root to the denominator
$\frac{\sqrt{x+7}-3}{x-2}\cdot\frac{\sqrt{x+7}+3}{\sqrt{x+7}+3}=\frac{(x+7)-9}{(x-2)\left(\sqrt{x+7}+3\right)}$

the conjugate is chosen so the numerator becomes a difference of squares, which has no root left in it

$=\frac{x-2}{(x-2)\left(\sqrt{x+7}+3\right)}=\frac{1}{\sqrt{x+7}+3}\quad(x\neq2)$

the factor $x-2$ that blocked substitution is now visible on both floors

Substitute in the repaired formula
$\lim_{x\to2}\frac{1}{\sqrt{x+7}+3}=\frac{1}{3+3}=\boxed{\frac16}$

the reduced formula is continuous at $2$, so substitution is finally legal

Answer $$k=\frac16$$
Check

Numerical probe with the original formula at $x=2.001$: $\sqrt{9.001}\approx3.000167$, so the quotient is $\frac{0.000167}{0.001}\approx0.1667$, which is $\tfrac16$ to four digits.

Whenever a $\tfrac00$ contains a square root, the conjugate is the first move, not the last resort.

3§02.3 — a limit inside a cosine●●●○○

The outer function is continuous everywhere, which is exactly why the whole problem is about the inner one.

Given
  • $\displaystyle F(x)=\cos\!\left(\frac{\pi\left(x^{2}-9\right)}{x^{2}-3x}\right)$, with the argument in radians

Find
  1. (a) Find $\displaystyle\lim_{x\to3}F(x)$.

Hint 1/4

Split the problem in two: what does the inside approach, and is the outside continuous there? Do not touch the cosine yet.

Hint 2/4

If $\lim_{x\to a}g(x)=L$ and $f$ is continuous at $L$, then $\lim f(g(x))=f(L)$. Cosine is continuous at every real number, so the hypothesis is free.

Hint 3/4

Inside: $\dfrac{\pi(x^{2}-9)}{x^{2}-3x}=\dfrac{\pi(x-3)(x+3)}{x(x-3)}=\dfrac{\pi(x+3)}{x}$ for $x\neq3$, which tends to $\dfrac{6\pi}{3}=2\pi$.

Hint 4/4

So the limit is $\cos(2\pi)=1$.

Show solution
Reduce the inner function
$\frac{\pi\left(x^{2}-9\right)}{x^{2}-3x}=\frac{\pi(x-3)(x+3)}{x(x-3)}=\frac{\pi(x+3)}{x}\quad(x\neq3)$

the same factor $x-3$ sits on both floors, which is what makes the $\tfrac00$ removable

$\lim_{x\to3}\frac{\pi(x+3)}{x}=\frac{6\pi}{3}=2\pi$

substitution is legal now: the denominator is $3$, not $0$

Check the outer function at that number and carry the limit through
$\cos\ \text{is continuous at}\ 2\pi$

cosine is continuous on all of $\mathbb{R}$, so the check costs one line and never fails

$\lim_{x\to3}F(x)=\cos(2\pi)=\boxed{1}$

the limit moves inside the cosine because the outer function is continuous at the inner limit

Answer $$1$$
Check

Probe at $x=3.01$: the inside is $\frac{\pi(6.01)}{3.01}\approx6.27275$ radians against $2\pi\approx6.2832$, and $\cos(6.27275)\approx0.99995$ — as close to $1$ as the probe is to the limit.

The inner function was undefined at the very point being approached, and it never mattered; only its limit was ever used.

4§02.5 — two ends, two different techniques●●●○○

Part (a) is the degree rule. Part (b) looks like the degree rule and is not, because of the root and the side.

Given
  • $\displaystyle u(x)=\frac{2x^{2}-1}{3x^{2}+x}$

  • $\displaystyle v(x)=\frac{\sqrt{x^{2}+3x}}{2x-1}$

Find
  1. (a) Find $\displaystyle\lim_{x\to\infty}u(x)$.

  2. (b) Find $\displaystyle\lim_{x\to-\infty}v(x)$.

Hint 1/4

For each part, decide what the dominant power is and what you will divide by. For (b), decide the sign of that power on the side you are approaching from.

Hint 2/4

Divide by the dominant power. Inside a root, $\sqrt{x^{2}}=\vert x\vert$, and $\vert x\vert=-x$ when $x<0$.

Hint 3/4

(a) Divide by $x^{2}$: $\dfrac{2-1/x^{2}}{3+1/x}$. (b) $\sqrt{x^{2}+3x}=\vert x\vert\sqrt{1+3/x}=-x\sqrt{1+3/x}$ for $x<0$; divide top and bottom by $x$.

Hint 4/4

(a) $\dfrac{2}{3}$. (b) $\dfrac{-\sqrt{1}}{2}=-\dfrac12$.

Show solution
(a) Equal degrees
$\frac{2x^{2}-1}{3x^{2}+x}=\frac{2-\tfrac{1}{x^{2}}}{3+\tfrac{1}{x}}$

dividing by $x^{2}$, the dominant power on both floors

$\longrightarrow\frac{2-0}{3+0}=\frac23$

the leading coefficients are what survive; the constants play no part

(b) Pull the bars out first
$\sqrt{x^{2}+3x}=\vert x\vert\sqrt{1+\tfrac{3}{x}}=-x\sqrt{1+\tfrac{3}{x}}\quad(x<0)$

on the far left $x$ is negative, so $\vert x\vert$ resolves to $-x$; this is the only place a sign enters

$\frac{-x\sqrt{1+\tfrac{3}{x}}}{x\left(2-\tfrac{1}{x}\right)}=\frac{-\sqrt{1+\tfrac{3}{x}}}{2-\tfrac{1}{x}}$

dividing by $x$, not by $x^{2}$: both floors are of size $x$

$\longrightarrow\frac{-1}{2}=\boxed{-\frac12}$

the small terms die and the minus stays

Answer $$\text{(a)}\ \frac23,\qquad\text{(b)}\ -\frac12$$
Check

Numerical probes: $u(1000)=\frac{1\,999\,999}{3\,001\,000}\approx0.6664$ against $\tfrac23$, and $v(-1000)=\frac{\sqrt{997\,000}}{-2001}\approx\frac{998.4989}{-2001}\approx-0.499$ against $-\tfrac12$. Both signs and both sizes agree.

Only one of these two problems has a sign in it, and it is the one with the root. That is the pattern worth memorising.

5§02.5 — a difference of two things that both run away●●●○○

$\infty-\infty$ is not a number and not an answer; it is a request to rewrite.

Given
  • $\displaystyle w(x)=\sqrt{4x^{2}+5x}-2x$, with $x\to\infty$

Find
  1. (a) Find $\displaystyle\lim_{x\to\infty}w(x)$.

  2. (b) State the horizontal asymptote of $w$ as $x\to\infty$.

Hint 1/4

Two competing giants cancel and leave something small. To see what is left, turn the difference into a single fraction.

Hint 2/4

Multiply by $\dfrac{\sqrt{4x^{2}+5x}+2x}{\sqrt{4x^{2}+5x}+2x}$; the numerator collapses to $\left(4x^{2}+5x\right)-4x^{2}=5x$. Then divide by the dominant power.

Hint 3/4

$w(x)=\dfrac{5x}{\sqrt{4x^{2}+5x}+2x}$, and here $x>0$, so $\sqrt{4x^{2}+5x}=x\sqrt{4+5/x}$.

Hint 4/4

Dividing by $x$: $\dfrac{5}{\sqrt{4+5/x}+2}\to\dfrac{5}{2+2}=\dfrac54$.

Show solution
Turn the difference into a quotient
$w(x)=\frac{\left(4x^{2}+5x\right)-4x^{2}}{\sqrt{4x^{2}+5x}+2x}=\frac{5x}{\sqrt{4x^{2}+5x}+2x}$

the conjugate is what makes the two giants cancel on paper instead of in your head

Divide by the dominant power
$\sqrt{4x^{2}+5x}=x\sqrt{4+\tfrac{5}{x}}\quad(x>0)$

on the far right the bars cost nothing, since $\vert x\vert=x$ there

$w(x)=\frac{5}{\sqrt{4+\tfrac{5}{x}}+2}\longrightarrow\frac{5}{2+2}=\boxed{\frac54}$

the root contributes a $2$ of its own, which is why the answer is not $\tfrac52$

Answer $$\frac54,\quad y=\frac54$$
Check

Probe at $x=1000$: $\sqrt{4\,005\,000}\approx2001.2496$, minus $2000$ leaves $1.2496$ — next to $\tfrac54$, and far from the $0$ that dropping the $5x$ would have predicted.

The rule of thumb: when the two leading terms cancel, the answer is decided by the terms everybody wanted to throw away.

C · exam level 3 questions
1§02.4 — reading a table of values for guaranteed roots●●●○○

A continuous function is sampled at six points and you are asked what is forced, not what is likely. Exam wording for this is "must contain a solution".

Given
  • $f$ is continuous on $[0,5]$

  • $f(0)=-3$, $f(1)=2$, $f(2)=1$, $f(3)=-4$, $f(4)=-1$, $f(5)=6$

Find
  1. (a) Which intervals between consecutive sample points are guaranteed to contain a zero of $f$?

Hint 1/4

You are not looking for large changes in value; you are looking for one specific event between consecutive samples.

Hint 2/4

A zero is forced between two consecutive samples exactly when their signs differ, because then $N=0$ lies strictly between the two values and the theorem applies on that subinterval.

Hint 3/4

The signs of $-3,2,1,-4,-1,6$ are $-,+,+,-,-,+$, so the sign changes between the 1st and 2nd samples, the 3rd and 4th, and the 5th and 6th.

Hint 4/4

The guaranteed intervals are $(0,1)$, $(2,3)$ and $(4,5)$.

Show solution
Turn values into signs
$-,\ +,\ +,\ -,\ -,\ +$

only the sign matters: the theorem compares $N=0$ with the two endpoint values, and size is irrelevant

Apply the theorem on each flip
$f(0)=-3<0<2=f(1)\Rightarrow\exists c\in(0,1)$

$f$ is continuous on the subinterval because it is continuous on all of $[0,5]$

$f(2)=1>0>-4=f(3),\qquad f(4)=-1<0<6=f(5)$

the same argument twice more

$\boxed{(0,1),\ (2,3),\ (4,5)}$

three flips, three guarantees, and no guarantee anywhere else

Answer $$(0,1),\ (2,3),\ (4,5)$$
Check

Check that the missing intervals really are not forced: a continuous function may pass from $2$ to $1$ on $[1,2]$ without ever reaching $0$ — the straight segment between them does exactly that.

Between two samples of the same sign, anything can happen; the theorem is silent, and silence is not a denial.

2§02.6 — the complete asymptote statement●●●●○

The exam version of this question always asks for all of them at once, and marks the answer as a package.

Given
  • $\displaystyle g(x)=\frac{3x^{2}+5}{x^{2}-x-6}$

Find
  1. (a) Give all vertical asymptotes and all horizontal asymptotes of $g$.

Hint 1/4

Two separate questions in one: what happens at the points where the formula refuses, and what happens at the two ends. Do them in that order.

Hint 2/4

Factor the denominator and check whether the numerator shares any zero. Then compare degrees: equal degrees give the ratio of leading coefficients.

Hint 3/4

$x^{2}-x-6=(x-3)(x+2)$, and $3x^{2}+5$ is $32$ at $x=3$ and $17$ at $x=-2$, so nothing cancels. Degrees are $2$ and $2$, leading coefficients $3$ and $1$.

Hint 4/4

So the vertical asymptotes are $x=3$ and $x=-2$, and the horizontal asymptote is $y=3$.

Show solution
Factor and check for cancellation
$x^{2}-x-6=(x-3)(x+2)$

the two candidates for vertical asymptotes

$3(3)^{2}+5=32\neq0,\qquad3(-2)^{2}+5=17\neq0$

the numerator vanishes nowhere at all, so nothing cancels and both candidates survive

Vertical asymptotes
$\lim_{x\to3^{-}}g(x)=-\infty,\quad\lim_{x\to3^{+}}g(x)=+\infty$

near $3$ the factor $x+2$ is about $5>0$, so the sign of the quotient follows $x-3$

$\lim_{x\to-2^{-}}g(x)=+\infty,\quad\lim_{x\to-2^{+}}g(x)=-\infty$

near $-2$ the factor $x-3$ is about $-5<0$, which flips both directions

The two ends
$g(x)=\frac{3+\tfrac{5}{x^{2}}}{1-\tfrac{1}{x}-\tfrac{6}{x^{2}}}\longrightarrow\frac{3}{1}=3$

equal degrees, so the leading coefficients decide, and the same computation serves both ends

$\boxed{x=3,\ x=-2\ \text{vertical};\ y=3\ \text{horizontal}}$

three lines, each earned by a limit

Answer $$x=3,\ x=-2,\ y=3$$
Check

Probe the claims: $g(100)=\frac{30\,005}{9894}\approx3.03$, close to $3$ from above, and $g(2.99)=\frac{31.82}{-0.0499}\approx-638$, large and negative just left of $3$ — both as predicted.

When the numerator has no real zeros, no cancellation is possible and every zero of the denominator is an asymptote. Checking that costs ten seconds.

3§02.1 — a parameter that decides the type of the break●●●●○

A midterm favourite, because one parameter controls whether the break is repairable at all, and the second parameter then has exactly one possible value.

Given
  • $f(x)=\dfrac{x^{2}+ax-6}{x-2}$ for $x\neq2$, and $f(2)=b$

Find
  1. (a) Find the value of $a$ for which $\lim_{x\to2}f(x)$ exists.

  2. (b) With that $a$, find the value of $b$ that makes $f$ continuous at $x=2$.

  3. (c) For every other value of $a$, what kind of break does $f$ have at $x=2$?

Hint 1/4

Ask what has to be true of the numerator at $x=2$ for the quotient to have a finite limit there, given that the denominator vanishes.

Hint 2/4

A finite limit at a zero of the denominator requires the numerator to vanish there too; otherwise the quotient is nonzero over zero, which is an infinite break. Then continuity forces $b=\lim_{x\to2}f(x)$.

Hint 3/4

Numerator at $2$: $4+2a-6=2a-2$. Setting it to $0$ gives $a=1$, and then $x^{2}+x-6=(x-2)(x+3)$.

Hint 4/4

So $a=1$, the limit is $2+3=5$, hence $b=5$; and for $a\neq1$ the break is infinite.

Show solution
(a) Force the numerator to vanish at 2
$2^{2}+2a-6=2a-2$

if this is not $0$, the quotient is a nonzero number over something shrinking to zero, and no finite limit is possible

$2a-2=0\Rightarrow a=1$

one condition, one unknown

(b) Compute the limit and match the value
$x^{2}+x-6=(x-2)(x+3)\Rightarrow f(x)=x+3\quad(x\neq2)$

with $a=1$ the numerator carries the factor $x-2$, which is what makes the break removable

$\lim_{x\to2}f(x)=5\Rightarrow\boxed{b=5}$

continuity means the value equals the limit, and the limit is now a substitution

(c) Classify the other cases
$a\neq1:\ \text{numerator}\to2a-2\neq0,\ \text{denominator}\to0$

the signature of an infinite break

$\Rightarrow\ \text{infinite break};\ x=2\ \text{is a vertical asymptote}$

and no choice of $b$ can repair it, since no finite value competes with an explosion

Answer $$a=1,\quad b=5,\quad\text{otherwise an infinite break}$$
Check

Test the borderline with a value $a$ that is close but wrong: for $a=1.1$, $f(2.001)=\frac{4.004+2.2011-6}{0.001}=\frac{0.2051}{0.001}\approx205$, already exploding — while $a=1$ gives $f(2.001)=5.001$.

Two different questions live in this problem: which $a$ makes a repair possible, and which $b$ performs it. Answering the second without the first is the usual way to lose the marks.

D · interleaved 3 questions
1§02 — mixed practice, technique not named●●●○○

No hints about which tool this needs; deciding that is the exercise.

Given
  • $\displaystyle m(x)=\frac{\sqrt{1+x}-\sqrt{1-x}}{x}$, with $x\to0$

Find
  1. (a) Find $\displaystyle\lim_{x\to0}m(x)$.

Hint 1/4

Try substitution first and name precisely what goes wrong. That name decides which repair is available.

Hint 2/4

Substitution gives $\tfrac00$, and the numerator is a difference of two roots, so multiply by the conjugate $\sqrt{1+x}+\sqrt{1-x}$.

Hint 3/4

The numerator becomes $(1+x)-(1-x)=2x$, so $m(x)=\dfrac{2x}{x\left(\sqrt{1+x}+\sqrt{1-x}\right)}$ for $x\neq0$.

Hint 4/4

Cancelling $x$ and substituting: $\dfrac{2}{1+1}=1$.

Show solution
Name the obstruction
$m(0)=\frac{\sqrt1-\sqrt1}{0}=\frac{0}{0}$

$\tfrac00$ with roots in it: the conjugate is the standard repair

Rationalise and cancel
$m(x)=\frac{(1+x)-(1-x)}{x\left(\sqrt{1+x}+\sqrt{1-x}\right)}=\frac{2x}{x\left(\sqrt{1+x}+\sqrt{1-x}\right)}$

the roots disappear from the numerator, which is the whole purpose of the move

$=\frac{2}{\sqrt{1+x}+\sqrt{1-x}}\quad(x\neq0)$

the factor $x$ that blocked substitution is gone

Substitute in the repaired formula
$\lim_{x\to0}\frac{2}{\sqrt{1+x}+\sqrt{1-x}}=\frac{2}{2}=\boxed{1}$

the reduced formula is continuous at $0$, so this last step is legal

Answer $$1$$
Check

Probe at $x=0.01$: $\sqrt{1.01}-\sqrt{0.99}=1.004987562-0.994987437=0.010000125$, divided by $0.01$ gives $1.0000125$.

A difference of two roots always calls the same move, whether the point is finite or at infinity.

2§02 — mixed practice, two blocks at once●●●○○

Two different techniques from two different blocks meet here, and the question does not say which is which.

Given
  • $\displaystyle n(x)=\frac{\sin x}{x}$, for $x>0$

Find
  1. (a) Find $\displaystyle\lim_{x\to\infty}n(x)$, with a justification.

  2. (b) State the horizontal asymptote, if there is one.

  3. (c) How many times does the graph meet that asymptote for $x>0$?

Hint 1/4

The numerator does not settle anywhere, so no limit law applies to it. Ask instead what the numerator can never do.

Hint 2/4

$-1\le\sin x\le1$ for every $x$; dividing by a positive $x$ preserves the inequalities, and the Squeeze Theorem finishes the job.

Hint 3/4

For $x>0$: $-\dfrac{1}{x}\le\dfrac{\sin x}{x}\le\dfrac{1}{x}$, and both bounds tend to $0$ as $x\to\infty$.

Hint 4/4

So the limit is $0$, the asymptote is $y=0$, and the graph meets it at every $x=k\pi$ with $k=1,2,3,\dots$ — infinitely often.

Show solution
Trap the quotient
$-1\le\sin x\le1\Rightarrow-\frac1x\le\frac{\sin x}{x}\le\frac1x\quad(x>0)$

dividing an inequality by a positive number keeps its direction; this is why the restriction $x>0$ was stated

$\lim_{x\to\infty}\left(-\frac1x\right)=\lim_{x\to\infty}\frac1x=0$

a constant over a growing power dies at both ends

$\Rightarrow\lim_{x\to\infty}\frac{\sin x}{x}=0$

the Squeeze Theorem, with the two bounds agreeing on $0$

Read off the asymptote and the crossings
$y=0\ \text{is a horizontal asymptote}$

that is precisely what a limit of $0$ at $\infty$ means

$\frac{\sin x}{x}=0\iff\sin x=0\iff x=k\pi$

a fraction is zero exactly when its numerator is, and the denominator never vanishes for $x>0$

$\boxed{\text{infinitely many crossings}}$

there are infinitely many positive multiples of $\pi$

Answer $$0,\quad y=0,\quad\text{infinitely many crossings}$$
Check

Two probes far out: $n(100\pi)=0$ exactly, and $n\!\left(100\pi+\tfrac{\pi}{2}\right)=\dfrac{1}{100\pi+\pi/2}\approx0.0032$ — the values alternate around $0$ and shrink, which is what settling on an asymptote while crossing it looks like.

Bounded divided by growing is one of the few patterns where no algebra is needed at all; recognising it is the entire solution.

3§02 — mixed practice, a junction and a limit●●●●○

The two pieces are ordinary. What happens where they meet is not.

Given
  • $f(x)=\begin{cases}x^{2}&x\le1\\[2pt]\dfrac{1}{x-1}&x>1\end{cases}$

Find
  1. (a) Compute $\lim_{x\to1^{-}}f(x)$ and $\lim_{x\to1^{+}}f(x)$.

  2. (b) Classify the break at $x=1$ and say whether $x=1$ is a vertical asymptote.

  3. (c) Is $f$ continuous from the left at $x=1$? Justify in one line.

Hint 1/4

Each side of the junction has its own formula; decide which one is valid on which side before computing anything.

Hint 2/4

The left limit uses the piece valid for $x<1$, the right limit the piece valid for $x>1$, and the value $f(1)$ comes from the piece whose inequality contains $1$. An infinite one sided limit makes the break infinite and the line $x=1$ an asymptote.

Hint 3/4

Left of $1$ the formula is $x^{2}$, so the left limit is $1$, and $f(1)=1^{2}=1$ because $x\le1$ owns the junction. Right of $1$ the formula is $\frac{1}{x-1}$, whose denominator shrinks to $0$ through positive values.

Hint 4/4

So the left limit is $1$, the right limit is $+\infty$, the break is infinite with $x=1$ a vertical asymptote, and $f$ is continuous from the left because $f(1)=1$ equals the left limit.

Show solution
(a) One side at a time
$\lim_{x\to1^{-}}f(x)=\lim_{x\to1^{-}}x^{2}=1$

left of the junction the function is a polynomial, so substitution is legal

$\lim_{x\to1^{+}}\frac{1}{x-1}=+\infty$

just right of $1$ the denominator is a small positive number, so the quotient grows past every bound

(b) Classify
$\text{one side is}\ +\infty\Rightarrow\text{infinite break}$

the third branch of the classification needs only one runaway side

$\Rightarrow\ x=1\ \text{is a vertical asymptote}$

the definition asks for at least one infinite one sided limit, not for both

(c) One sided continuity
$f(1)=1^{2}=1$

the piece with $x\le1$ contains the junction, so it supplies the value

$f(1)=1=\lim_{x\to1^{-}}f(x)\Rightarrow\boxed{\text{continuous from the left}}$

the left hand test passes even though the two sided one cannot

Answer $$1,\ +\infty;\ \text{infinite break};\ \text{continuous from the left}$$
Check

Numerical probe on both sides: $f(0.999)=0.998001$, sitting next to $f(1)=1$, while $f(1.001)=1000$. One side is calm and one is not, exactly as the classification claims.

Continuity from one side is a real and separate property; on a closed interval ending at $1$, this function would count as continuous.

Mistake ledger (14 entries)
⚠ Repairing a jump with the midpoint

both sides are finite, so it feels as though a number is merely missing and the average is the fair choice

wrong$f(2):=\tfrac{(-1)+1}{2}=0\ \Rightarrow\ f\ \text{continuous at }2$
right$\lim_{x\to2^{-}}f\neq\lim_{x\to2^{+}}f\ \Rightarrow\ \text{no value of }f(2)\ \text{works}$
⚠ Losing the restriction that comes with a cancellation

the cancelled factor leaves the page and then leaves memory

wrong$\frac{x-1}{x^{2}-1}=\frac{1}{x+1}\ \text{for all }x$
right$\frac{x-1}{x^{2}-1}=\frac{1}{x+1}\quad(x\neq1)$
⚠ Demanding a two sided limit at an endpoint

the definition at an interior point is the one everybody memorised, and it gets applied everywhere

wrong$\text{on}\ [a,b]:\ \lim_{x\to a}f(x)=f(a)$
right$\text{on}\ [a,b]:\ \lim_{x\to a^{+}}f(x)=f(a)$
⚠ Calling a function discontinuous at a point outside its domain

the graph is missing there, and missing looks like broken

wrong$\sqrt{x-2}\ \text{is discontinuous at}\ x=1$
right$\sqrt{x-2}\ \text{is undefined at}\ x=1;\ \text{continuity is not asked there}$
⚠ Evaluating the outer function at $a$

$x\to a$ is written on the limit sign, so $a$ is the number in front of your eyes when the outer function asks for input

wrong$\lim_{x\to a}f\bigl(g(x)\bigr)=f(a)$
right$\lim_{x\to a}f\bigl(g(x)\bigr)=f\Bigl(\lim_{x\to a}g(x)\Bigr)$
⚠ Demanding continuity of the inner function too

the rule for a composition of continuous functions is the one people memorise, and it is stronger than what limits need

wrong$\text{need }g\ \text{continuous at }a\ \text{and}\ f\ \text{continuous at }L$
right$\text{need only}\ \lim_{x\to a}g=L\ \text{and}\ f\ \text{continuous at }L$
⚠ Reading "at least one" as "exactly one"

the picture people draw has a single crossing, and the picture gets remembered instead of the statement

wrong$f(a) < 0 < f(b)\Rightarrow f\ \text{has exactly one zero in}\ (a,b)$
right$f(a) < 0 < f(b)\Rightarrow f\ \text{has at least one zero in}\ (a,b)$
⚠ Using the theorem across a vertical asymptote

the endpoint values look perfect, and the break sits in the middle where nobody evaluates anything

wrong$\tan\tfrac{\pi}{4}=1,\ \tan\tfrac{3\pi}{4}=-1\Rightarrow\exists c:\tan c=0$
right$\tan\ \text{is not continuous on}\ \left[\tfrac{\pi}{4},\tfrac{3\pi}{4}\right]\ \text{(break at}\ \tfrac{\pi}{2})$
⚠ Turning $\sqrt{x^{2}}$ into $x$ on the far left

the identity $\sqrt{x^{2}}=x$ is true for every $x$ anybody ever practises with, because those are all positive

wrong$\sqrt{4x^{2}+1}\approx2x\quad(x\to-\infty)$
right$\sqrt{4x^{2}+1}=\vert x\vert\sqrt{4+\tfrac{1}{x^{2}}}\approx-2x\quad(x\to-\infty)$
⚠ Dropping the small term under a root

$6x$ really is negligible next to $x^{2}$, so the approximation feels safe — but the two huge terms then cancel and the neglected part is all that is left

wrong$\sqrt{x^{2}+6x}-x\approx x-x=0$
right$\sqrt{x^{2}+6x}-x=\frac{6x}{\sqrt{x^{2}+6x}+x}\to3$
⚠ Listing every zero of the original denominator

the denominator is where asymptotes come from, and reducing takes an extra minute that nobody has in an exam

wrong$\frac{x^{2}-9}{x-3}:\ x=3\ \text{is a vertical asymptote}$
right$\frac{x^{2}-9}{x-3}=x+3\ (x\neq3):\ \text{hole at}\ (3,6)$
⚠ Giving one infinity for both sides

the graph explodes, and one symbol feels like enough to say so

wrong$\lim_{x\to3}\frac{x(x-2)}{x-3}=+\infty$
right$\lim_{x\to3^{-}}\frac{x(x-2)}{x-3}=-\infty,\qquad\lim_{x\to3^{+}}\frac{x(x-2)}{x-3}=+\infty$
⚠ Moving a term across an equals sign without changing its sign

in a hurry the two operations — move and negate — collapse into one, and the line still looks like ordinary algebra

wrong$-2+a=-1\Rightarrow a=-1$
right$-2+a=-1\Rightarrow a=-1+2=1$
⚠ Reading $f$ at a junction from the piece with the strict inequality

two formulas are in view at a junction and only the inequality signs say which one owns the point

wrong$f(-1)=2(-1)+a\ \text{from the piece}\ x<-1$
right$f(-1)=(-1)^{2}-2\ \text{from the piece}\ -1\le x\le2$
Formula card
Classification of a break
$\text{removable / jump / infinite, decided by the two one sided limits}$

$f$ defined on both sides of $a$

Repair of a removable break
$f(a):=\lim_{x\to a}f(x)$

the two sided limit exists and is finite

Algebra of continuous functions
$f\pm g,\ fg,\ cf,\ f/g\ (g(a)\neq0)\ \text{continuous at}\ a$

$f,g$ continuous at $a$

Continuity on a closed interval
$\text{interior points}+\lim_{x\to a^{+}}f=f(a)+\lim_{x\to b^{-}}f=f(b)$

the interval is $[a,b]$

Limit through a continuous outer function
$\lim_{x\to a}f\bigl(g(x)\bigr)=f\Bigl(\lim_{x\to a}g(x)\Bigr)$

$\lim_{x\to a}g=L$ exists and $f$ is continuous at $L$

Intermediate Value Theorem
$f\in C[a,b],\ N\ \text{between}\ f(a),f(b)\Rightarrow\exists c\in(a,b):f(c)=N$

continuity on the closed interval; $N$ strictly between the endpoint values

Powers die at infinity
$\lim_{x\to\pm\infty}\frac{1}{x^{n}}=0\quad(n>0)$

$n>0$

Degree rule for a quotient of polynomials
$\lim_{x\to\pm\infty}\frac{P}{Q}=\begin{cases}0&\deg P<\deg Q\\[2pt]\dfrac{a_{n}}{b_{m}}&\deg P=\deg Q\\[2pt]\pm\infty&\deg P>\deg Q\end{cases}$

$P,Q$ polynomials, fraction reduced first

Root at infinity
$\sqrt{ax^{2}+bx}=\vert x\vert\sqrt{a+\tfrac{b}{x}},\quad\vert x\vert=-x\ (x<0)$

the inside is nonnegative on the side you are approaching from

Hole versus vertical asymptote
$\tilde Q(a)=0,\ \tilde P(a)\neq0\Rightarrow\text{asymptote};\quad(x-a)\ \text{cancelled}\Rightarrow\text{hole}$

the fraction has been reduced to lowest terms

Slant asymptote by division
$\deg P=\deg Q+1:\ \frac{P}{Q}=mx+b+\frac{r(x)}{Q(x)},\ \frac{r}{Q}\to0$

the degree gap is exactly one

Check yourself

Close the page and write, from memory: the three kinds of break and how the two one sided limits distinguish them; the two hypotheses of the Intermediate Value Theorem; and the first move for a limit at $-\infty$ containing a square root. Then reopen and compare.

  • Classify a break as removable, jump or infinite and give the repairing value when there is one?

    c-classify

  • Say on which set a formula built from the standard families is continuous, endpoints included, without writing a single limit?

    c-interval

  • State at which number the continuity of the outer function has to be checked in $f(g(x))$, and why the inner one need not be continuous at all?

    c-composite

  • Write the two line proof that an equation has a solution in an interval, including the sentence about continuity that graders look for?

    c-ivt

  • Compute a limit at $-\infty$ with a square root in it and get the sign right?

    c-infinity

  • Produce holes, vertical asymptotes with both directions, and the end behaviour of a rational function, each with its limit?

    c-asymptote-map

Glossary (12 terms)
continuity on an intervalaralıkta süreklilik

Continuity at every point of the interval; at an endpoint of a closed interval only the side reaching into the interval is tested.

tek yönlü süreklilik

$\lim_{x\to a^{+}}f(x)=f(a)$ (from the right) or $\lim_{x\to a^{-}}f(x)=f(a)$ (from the left); a function can have one without the other.

sonsuz süreksizlik

A break at which at least one one sided limit is $\pm\infty$; the line $x=a$ is then a vertical asymptote.

holeboşluk

A single missing point of a graph, at height $\lim_{x\to a}f(x)$; it is what a completely cancelled factor leaves behind.

Intermediate Value TheoremAra Değer Teoremi

A function continuous on $[a,b]$ takes every value strictly between $f(a)$ and $f(b)$ at some point of $(a,b)$.

closed intervalkapalı aralık

$[a,b]$, endpoints included; the setting the Intermediate Value Theorem requires.

ikiye bölme

Repeated halving of an interval carrying a sign change, each halving doubling the accuracy of the location of a root.

limit at infinitysonsuzda limit

The number $f(x)$ settles on as $x$ grows past every bound, or as $x$ decreases past every bound.

horizontal asymptoteyatay asimptot

A line $y=L$ with $\lim_{x\to\infty}f(x)=L$ or $\lim_{x\to-\infty}f(x)=L$; the graph may cross it any number of times.

slant asymptoteeğik asimptot

A line $y=mx+b$ with $m\neq0$ that the graph approaches at $\pm\infty$; for a rational function it appears when the degree above is exactly one more than the degree below.

baskın terim

The term that grows fastest far from the origin; dividing by it is what turns a limit at infinity into an ordinary computation.

end behaviour

What a graph does as $x\to\infty$ and as $x\to-\infty$, reported separately because the two ends can differ.

What comes next
§03 · Derivatives: definition and basic differentiation rules

Continuity says a graph has no gaps. The next section asks a sharper question about the same graph — does it have a direction at each point — and the answer turns out to be a limit of exactly the kind computed here, taken on the difference quotient.

Sources
  • James Stewart, Calculus, Metric Version, Ninth Edition — sections 1.6 and 1.8 The section numbers are the ones this week's plan names; the classification of breaks, the algebra of continuous functions and the Intermediate Value Theorem are in 1.8, the limit laws behind them in 1.6.
  • Course syllabus: weekly plan and grade weights Week 2 reads "Limits 1.6, 1.8". The weights quoted on the card and the rule about the two midterms come from the same document.
  • Conventions for reporting continuity, infinite limits and asymptotes Collected in the conventions box, so that every answer in this section is written the same way a grader expects to read it.

Spotted something missing or wrong? tell us · share your own notes or an old exam.