\chapter{Continuous Functions} \label{lim:chapter}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

\section{Limits of functions}
\label{sec:limoffunc}

%mbxINTROSUBSECTION

\sectionnotes{2--3 lectures}

Before we define continuity of functions, we visit a somewhat
more general notion of a limit than that of a sequence.
Given a function $f \colon S \to
\R$, we want to see how $f(x)$ behaves as $x$ tends to a certain point.

\subsection{Cluster points}

First,
we return to a concept we have seen previously in an exercise.
When moving within the set $S$, we can only approach points
that have elements of $S$ arbitrarily near.

\begin{defn}
Let $S \subset \R$ be a set.  A number $x \in \R$ is called
a \emph{\myindex{cluster point}} of $S$
if for every $\epsilon > 0$, the set $(x-\epsilon,x+\epsilon) \cap S
\setminus \{ x \}$ is not empty.
\end{defn}

That is, $x$ is a cluster point of $S$ if there are points of $S$
arbitrarily close to $x$.  Another way to phrase the definition is to say
that $x$ is a cluster point of $S$ if for every $\epsilon > 0$, there
exists a $y \in S$ such that $y \neq x$ and $\sabs{x - y} < \epsilon$.
Note that a cluster point of $S$ need not lie in $S$.

Let us see some examples.
\begin{enumerate}[(i)]
\item The set
$\{ \nicefrac{1}{n} : n \in \N \}$ has zero as its unique cluster point.
\item The cluster points of the open interval $(0,1)$ are
all points in the closed interval $[0,1]$.
\item
The set of cluster points of $\Q$ is the whole real line $\R$.
\item 
The set of cluster points of $[0,1) \cup \{ 2 \}$ is the interval $[0,1]$.
\item The set $\N$ has no cluster points in $\R$.
\end{enumerate}

\begin{prop}
Let $S \subset \R$.  Then $x \in \R$ is a cluster point of $S$
if and only if
there exists a convergent sequence of numbers $\{ x_n \}_{n=1}^\infty$ such that
$x_n \neq x$ and $x_n \in S$ for all $n$, and $\lim\limits_{n\to\infty} x_n = x$.
\end{prop}

\begin{proof}
First suppose $x$ is a cluster point of $S$.
For every $n \in \N$, pick $x_n$ to be an arbitrary point of
$(x-\nicefrac{1}{n},x+\nicefrac{1}{n}) \cap S \setminus \{x\}$, which
is nonempty because $x$ is a cluster point of $S$.
Then
$x_n$ is within $\nicefrac{1}{n}$ of $x$, that is,
\begin{equation*}
\sabs{x-x_n} < \nicefrac{1}{n} .
\avoidbreak
\end{equation*}
As $\{ \nicefrac{1}{n} \}_{n=1}^\infty$ converges to zero,
$\{ x_n \}_{n=1}^\infty$ converges to $x$.

On the other hand, if we start with a sequence of numbers
$\{ x_n \}_{n=1}^\infty$ in $S$
converging to $x$ such that $x_n \neq x$ for all $n$, then for every
$\epsilon > 0$ there is an $M$ such that, in particular, $\sabs{x_M - x} <
\epsilon$.  That is, $x_M \in (x-\epsilon,x+\epsilon) \cap S \setminus \{x\}$.
\end{proof}

\subsection{Limits of functions}

If a function $f$ is defined on a set $S$ and $c$ is a cluster point of $S$,
then we define the limit of $f(x)$ as $x$ approaches $c$.  
It is irrelevant for the definition whether $f$ is defined at $c$ or not.
Even if the function is defined at $c$, the limit of the
function as $x$ goes to $c$ can very well be different
from $f(c)$.

\begin{defn}
\index{limit!of a function}%
Let $f \colon S \to \R$ be a function and $c$ a cluster point of
$S \subset \R$.
Suppose there exists an $L \in \R$ and for every $\epsilon > 0$,
there exists a $\delta > 0$ such that whenever $x \in S \setminus \{ c \}$
and $\sabs{x - c} < \delta$, we have
\begin{equation*}
\babs{f(x) - L} < \epsilon .
\end{equation*}
We then say $f(x)$ \emph{converges}\index{converges!function} to $L$ as $x$ goes
to $c$, and we write
\glsadd{not:limitasarrows}%
\begin{equation*}
f(x) \to L \quad\text{as}\quad x \to c .
\end{equation*}
We say $L$ is a \emph{limit} of $f(x)$ as $x$
goes to $c$, and if $L$ is unique (it is), we write
\glsadd{not:limfunc}%
\begin{equation*}
\lim_{x \to c} f(x) \coloneqq L .
\end{equation*}
If no such $L$ exists, then we say that the limit does not exist or
that $f$ \emph{\myindex{diverges}} at $c$.
\end{defn}

Again, the notation and language we are using above assumes the limit $L$, if
it exists, is unique, which needs to be proved.  Note that the fact that $c$
is a cluster point is important to prove uniqueness.

\begin{prop}
Let $c$ be a cluster point of $S \subset \R$ and let $f \colon S \to \R$
be a function such that $f(x)$ converges as $x$ goes to $c$.  Then
the limit of $f(x)$ as $x$ goes to $c$ is unique.
\end{prop}

\begin{proof}
Let $L_1$ and $L_2$ be two numbers that both satisfy the definition.
Take an $\epsilon > 0$ and find a $\delta_1 > 0$ such that
$\babs{f(x)-L_1} < \nicefrac{\epsilon}{2}$ 
for all $x \in S \setminus \{c\}$ with $\sabs{x-c} < \delta_1$.
Also find $\delta_2 > 0$ such that
$\babs{f(x)-L_2} < \nicefrac{\epsilon}{2}$
for all $x \in S \setminus \{c\}$ with $\sabs{x-c} < \delta_2$.
Put $\delta \coloneqq \min \{ \delta_1, \delta_2 \}$.  Suppose $x \in S$,
$\sabs{x-c} < \delta$, and $x \neq c$.  As $\delta > 0$ and $c$ is a cluster
point, such an $x$ exists.  Then
\begin{equation*}
\sabs{L_1 - L_2} =
\babs{L_1 - f(x) + f(x) - L_2} \leq
\babs{L_1 - f(x)} + \babs{f(x) - L_2} < \frac{\epsilon}{2} + \frac{\epsilon}{2}
= \epsilon.
\end{equation*}
As $\sabs{L_1-L_2} < \epsilon$ for arbitrary $\epsilon > 0$, we have
$L_1 = L_2$.
\end{proof}

\begin{example}
Consider $f \colon \R \to \R$ defined by $f(x) \coloneqq x^2$.  Then
for any $c \in \R$,
\begin{equation*}
\lim_{x\to c} f(x) = \lim_{x\to c} x^2 = c^2 .
\end{equation*}

Proof: Let $c \in \R$ be fixed, and suppose $\epsilon > 0$ is given.  Write
\begin{equation*}
\delta \coloneqq \min \left\{ 1 , \, \frac{\epsilon}{2\sabs{c}+1} \right\} .
\end{equation*}
Take $x \neq c$ such that $\sabs{x-c} < \delta$.  In particular,
$\sabs{x-c} < 1$.  By the reverse triangle inequality,
\begin{equation*}
\sabs{x}-\sabs{c} \leq \sabs{x-c} < 1 .
\end{equation*}
Adding $2\sabs{c}$ to both sides, we obtain
$\sabs{x} + \sabs{c} < 2\sabs{c} + 1$.  Estimate
\begin{equation*}
\begin{split}
\babs{f(x) - c^2} &= \sabs{x^2-c^2} \\
&= \babs{(x+c)(x-c)} \\
&= \sabs{x+c}\sabs{x-c} \\
&\leq \bigl(\sabs{x}+\sabs{c}\bigr)\sabs{x-c} \\
&< \bigl(2\sabs{c}+1\bigr)\sabs{x-c} \\
&< \bigl(2\sabs{c}+1\bigr)\frac{\epsilon}{2\sabs{c}+1} = \epsilon .
\end{split}
\end{equation*}
\end{example}

\begin{example}
Define $f \colon [0,1) \to \R$ by
\begin{equation*}
f(x) \coloneqq 
\begin{cases}
x & \text{if } x > 0 , \\
1 & \text{if } x = 0 .
\end{cases}
\end{equation*}
Then
$\lim\limits_{x\to 0} f(x) = 0$,
even though $f(0) = 1$.  See \figureref{fig:limvaldiff}.
\begin{myfigureht}
\myincludegraphics{limvaldiff}{%
A graph of a function that is 1 when the independent
variable x is 0, then when the independent variable
x is any bigger than 0, the function is equal to x.
It is indicated that the function is not defined at
x equal to 1 or any x greater, nor at any x less than 0.}
\caption{Function with a different limit and value at $0$.\label{fig:limvaldiff}}
\end{myfigureht}

Proof:  Let $\epsilon > 0$ be given.  Let $\delta \coloneqq \epsilon$.
For $x \in [0,1)$, $x \neq 0$, and $\sabs{x-0} < \delta$, we get
\begin{equation*}
\babs{f(x) - 0} = \sabs{x} < \delta = \epsilon .
\end{equation*}
\end{example}

\subsection{Sequential limits} \label{subseq:sequentiallimits}

Let us connect the limit as defined above with limits of sequences.

\begin{lemma}\label{seqflimit:lemma}
Let $S \subset \R$, let $c$ be a cluster point of $S$, let $f \colon S \to
\R$ be a function, and let $L \in \R$.

Then
$f(x) \to L$ as $x \to c$ if and only if for every sequence
$\{ x_n \}_{n=1}^\infty$
such that $x_n \in S \setminus \{c\}$ for all $n$,
and such that $\lim_{n\to\infty} x_n = c$,
we have that the sequence $\bigl\{ f(x_n) \bigr\}_{n=1}^\infty$ converges to $L$.
\end{lemma}

\begin{proof}
Suppose 
$f(x) \to L$ as $x \to c$, and $\{ x_n \}_{n=1}^\infty$ is a sequence
such that
$x_n \in S \setminus \{c\}$ and
$\lim_{n\to\infty} x_n = c$.
We wish to show that $\bigl\{ f(x_n) \bigr\}_{n=1}^\infty$ converges to $L$.
Let $\epsilon > 0$ be given.  Find a $\delta > 0$ such that
if $x \in S \setminus \{c\}$ and $\sabs{x-c} < \delta$, then
$\babs{f(x) - L} < \epsilon$.  As
$\{ x_n \}_{n=1}^\infty$  converges to $c$, find an $M$ such that for $n \geq M$,
we have that $\sabs{x_n - c} < \delta$.  Therefore, for $n \geq M$,
\begin{equation*}
\babs{f(x_n) - L} < \epsilon .
\end{equation*}
Thus $\bigl\{ f(x_n) \bigr\}_{n=1}^\infty$ converges to $L$.

For the other direction, we use proof by contrapositive.  Suppose 
it is not true that $f(x) \to L$ as $x \to c$.  The negation of the
definition is that there exists an $\epsilon > 0$ such that for every
$\delta > 0$ there exists an $x \in S \setminus \{c\}$, where
$\sabs{x-c} < \delta$
and $\babs{f(x)-L} \geq \epsilon$.

Let us use $\nicefrac{1}{n}$ for $\delta$ in the statement above to
construct a sequence $\{ x_n \}_{n=1}^\infty$.  We have
that there exists an $\epsilon > 0$ such that for every $n$,
there exists a point $x_n \in S \setminus \{c\}$, where
$\sabs{x_n-c} < \nicefrac{1}{n}$
and $\babs{f(x_n)-L} \geq \epsilon$.
The sequence $\{ x_n \}_{n=1}^\infty$ just constructed converges to $c$, but
the sequence $\bigl\{ f(x_n) \bigr\}_{n=1}^\infty$ does not converge to $L$.
And we are done.
\end{proof}

It is possible to strengthen the reverse direction of
the lemma by simply stating that
\myquote{$\bigl\{ f(x_n) \bigr\}_{n=1}^\infty$ converges,} without requiring a specific limit.
See \exerciseref{exercise:seqflimitalt}.

\begin{example}
$\displaystyle \lim_{x \to 0} \sin( \nicefrac{1}{x} )$
does not exist, but 
$\displaystyle \lim_{x \to 0} x\sin( \nicefrac{1}{x} ) = 0$.
See \figureref{figsin1x}.

\begin{myfigureht}
%left guy also used in 10.4
\myincludepdft{sin1x_xsin1x}{%
The graphs of functions on an interval around the origin.
On the left the function oscillates
between minus 1 and 1
faster and faster as x approaches 0, so much so that near the origin
the graph appears as solid black.
On the right, the function again oscillates faster and faster
as x approaches 0, but now the amplitude also goes down to 0
and so what sort of seems like a solid black region is really like the
tip of a triangle.}
\caption{Graphs of $\sin(\nicefrac{1}{x})$ and $x \sin(\nicefrac{1}{x})$.
Note that the computer cannot properly graph $\sin(\nicefrac{1}{x})$
near zero as it oscillates too fast.\label{figsin1x}}
\end{myfigureht}

Proof:
We start with $\sin(\nicefrac{1}{x})$.  Define a sequence
by
$x_n \coloneqq \frac{1}{\pi n + \nicefrac{\pi}{2}}$.  It is not hard to see
that $\lim_{n\to\infty} x_n = 0$.  Furthermore,
\begin{equation*}
\sin ( \nicefrac{1}{x_n} )
=
\sin (\pi n + \nicefrac{\pi}{2})
= {(-1)}^n .
\end{equation*}
Therefore, $\bigl\{ \sin ( \nicefrac{1}{x_n} ) \bigr\}_{n=1}^\infty$ does not converge.
By
\lemmaref{seqflimit:lemma}, 
$\displaystyle \lim_{x \to 0} \sin( \nicefrac{1}{x} )$ does not exist.

Now consider $x\sin(\nicefrac{1}{x})$.  Let $\{ x_n \}_{n=1}^\infty$ be a sequence
such that $x_n \neq 0$ for all $n$, and such that $\lim_{n\to\infty} x_n = 0$.
Notice that $\babs{\sin(t)} \leq 1$ for all $t \in \R$.  Therefore,
\begin{equation*}
\babs{x_n\sin(\nicefrac{1}{x_n})-0}
=
\sabs{x_n}\babs{\sin(\nicefrac{1}{x_n})}
\leq
\sabs{x_n} .
\end{equation*}
As $x_n$ goes to zero, $\sabs{x_n}$ goes to zero, and hence
$\bigl\{ x_n\sin(\nicefrac{1}{x_n}) \bigr\}_{n=1}^\infty$ converges to zero.  By
\lemmaref{seqflimit:lemma}, 
$\displaystyle \lim_{x \to 0} x\sin( \nicefrac{1}{x} ) = 0$.
\end{example}

Keep in mind the phrase \myquote{for every sequence} in the lemma.
For example, take $\sin(\nicefrac{1}{x})$ and the sequence given by
$x_n \coloneqq \nicefrac{1}{\pi n}$.
Then $\bigl\{ \sin (\nicefrac{1}{x_n}) \bigr\}_{n=1}^\infty$
is the constant zero sequence, and
therefore converges to zero, but the limit of 
$\sin(\nicefrac{1}{x})$ as $x \to 0$ does not exist.

Using \lemmaref{seqflimit:lemma}, 
we can start applying everything we know about
sequential limits to limits of functions.  Let us give a few important
examples.

\begin{cor}
Let $S \subset \R$ and let $c$ be a cluster point of $S$.  
Suppose $f \colon S \to
\R$ and $g \colon S \to \R$ are functions
such that
the limits of $f(x)$ and $g(x)$ as $x$ goes to $c$ both exist,
and
\begin{equation*}
f(x) \leq g(x) \qquad \text{for all } x \in S \setminus \{ c \}.
\end{equation*}
Then
\begin{equation*}
\lim_{x\to c} f(x) \leq \lim_{x\to c} g(x) .
\end{equation*}
\end{cor}

\begin{proof}
Take $\{ x_n \}_{n=1}^\infty$ to be a sequence of numbers in $S \setminus \{ c \}$
that converges to $c$.  Let
\begin{equation*}
L_1 \coloneqq \lim_{x\to c} f(x), \qquad \text{and} \qquad L_2 \coloneqq \lim_{x\to c} g(x) .
\end{equation*}
\lemmaref{seqflimit:lemma} says that $\bigl\{ f(x_n) \bigr\}_{n=1}^\infty$ converges to
$L_1$ and $\bigl\{ g(x_n) \bigr\}_{n=1}^\infty$ converges to $L_2$.  We also
have $f(x_n) \leq g(x_n)$ for all $n$.
We obtain $L_1 \leq L_2$ using
\lemmaref{limandineq:lemma}.
\end{proof}

By applying constant functions, we get the following corollary.  The
proof is left as an exercise.

\begin{cor} \label{fconstineq:cor}
Let $S \subset \R$ and let $c$ be a cluster point of $S$.  Suppose $f \colon S \to
\R$ is a function such that the limit of $f(x)$ as $x$ goes to $c$
exists.
Suppose there are two real numbers $a$ and $b$ such that
\begin{equation*}
a \leq f(x) \leq b \qquad \text{for all } x \in S \setminus \{ c \}.
\end{equation*}
Then
\begin{equation*}
a \leq \lim_{x\to c} f(x) \leq b .
\end{equation*}
\end{cor}

Using \lemmaref{seqflimit:lemma} in the same way as above, we also get
the following corollaries, whose proofs are again left as exercises.

\begin{cor} \label{fsqueeze:cor}
Let $S \subset \R$ and let $c$ be a cluster point of $S$.
Suppose $f \colon S \to \R$,
$g \colon S \to \R$, and $h \colon S \to \R$ are functions such that
\begin{equation*}
f(x) \leq g(x) \leq h(x) \qquad \text{for all } x \in S \setminus \{ c \}.
\end{equation*}
Suppose the limits of $f(x)$ and $h(x)$ as $x$ goes to $c$ both exist, and
\begin{equation*}
\lim_{x\to c} f(x) = \lim_{x\to c} h(x) .
\end{equation*}
Then the limit of $g(x)$ as $x$ goes to $c$ exists and
\begin{equation*}
\lim_{x\to c} g(x) =
\lim_{x\to c} f(x) = \lim_{x\to c} h(x) .
\end{equation*}
\end{cor}

\begin{cor} \label{falg:cor}
Let $S \subset \R$ and let $c$ be a cluster point of $S$.  
Suppose $f \colon S \to \R$ and
$g \colon S \to \R$ are functions
such that 
the limits of $f(x)$ and $g(x)$ as $x$ goes to $c$ both exist.
Then
\begin{enumerate}[(i)]
\item
$\displaystyle
\lim_{x\to c} \bigl(f(x)+g(x)\bigr) = \left(\lim_{x\to c} f(x)\right) + 
\left(\lim_{x\to c} g(x)\right) .
$
\item
$\displaystyle
\lim_{x\to c} \bigl(f(x)-g(x)\bigr) = \left(\lim_{x\to c} f(x)\right) -
\left(\lim_{x\to c} g(x)\right) .
$
\item
$\displaystyle
\lim_{x\to c} \bigl(f(x)g(x)\bigr) = \left(\lim_{x\to c} f(x)\right)
\left(\lim_{x\to c} g(x)\right) .
$
\item \label{falg:cor:iv} If
$\displaystyle \lim_{x\to c} g(x) \neq 0$
and $g(x) \neq 0$ for all $x \in S \setminus \{ c \}$, then
\begin{equation*}
\lim_{x\to c} \frac{f(x)}{g(x)} =
\frac{\lim_{x\to c} f(x)}{\lim_{x\to c} g(x)} .
\end{equation*}
\end{enumerate}
\end{cor}

\begin{cor} \label{fabs:cor}
Let $S \subset \R$ and let $c$ be a cluster point of $S$.
Suppose $f \colon S \to \R$ is a function such that the limit of $f(x)$ as $x$ goes to $c$
exists.
Then
\begin{equation*}
\lim_{x\to c} \babs{f(x)} =
\abs{\lim_{x\to c} f(x)}.
\end{equation*}
\end{cor}

\subsection{Limits of restrictions and one-sided limits}

Sometimes we work with the function defined on a subset.

\begin{defn}
Let $f \colon S \to \R$ be a function and $A \subset S$.  Define the
function $f|_A \colon A \to \R$ by
\begin{equation*}
f|_A (x) \coloneqq f(x)  \qquad \text{for } x \in A.
\end{equation*}
We call $f|_A$ the \emph{\myindex{restriction}} of $f$ to $A$.
\end{defn}

The function $f|_A$ is simply the function $f$ taken on a smaller domain.
The following proposition is the analogue of taking a tail of a sequence.
It says that the limit is \myquote{local}: The limit only depends on points
arbitrarily near $c$.

\begin{prop} \label{prop:limrest}
Let $S \subset \R$, $c \in \R$, and
let $f \colon S
\to \R$ be a function.
Suppose
$A \subset S$ is such that for some $\alpha > 0$, we have
$(A \setminus \{ c \}) \cap (c-\alpha,c+\alpha) = (S \setminus \{ c \}) \cap (c-\alpha,c+\alpha)$.
\begin{enumerate}[(i)]
\item The point $c$ is a cluster point of $A$ if and only if $c$ is a cluster point
of $S$.
\item Supposing $c$ is a cluster point of $S$, then $f(x) \to L$ as $x \to c$ if and only if
$f|_A(x) \to L$ as $x \to c$.
\end{enumerate}
\end{prop}

\begin{proof}
First, let $c$ be a cluster point of $A$.
Since $A \subset S$, if $( A \setminus \{ c\} ) \cap
(c-\epsilon,c+\epsilon)$ is nonempty for every $\epsilon > 0$,
then $( S \setminus \{ c\} ) \cap
(c-\epsilon,c+\epsilon)$ is nonempty for every $\epsilon > 0$.
Thus $c$ is a cluster point of $S$.
Second, suppose $c$ is a cluster
point of $S$.  Then for $\epsilon > 0$ such that $\epsilon < \alpha$
we get that $( A \setminus \{ c\} ) \cap (c-\epsilon,c+\epsilon) =
( S \setminus \{ c\} ) \cap (c-\epsilon,c+\epsilon)$, which is nonempty.  This is true for all
$\epsilon < \alpha$ and hence 
$( A \setminus \{ c\} ) \cap (c-\epsilon,c+\epsilon)$ must be nonempty for all
$\epsilon > 0$.  Thus $c$ is a cluster point of $A$.

Now suppose $c$ is a cluster point of $S$ and $f(x) \to L$ as $x \to c$.  That is, for every $\epsilon > 0$
there is a $\delta > 0$ such that if $x \in S \setminus \{ c \}$
and $\sabs{x-c} < \delta$, then $\babs{f(x)-L} < \epsilon$.  Because $A \subset S$,
if $x \in A \setminus \{ c \}$, then $x \in S \setminus \{ c \}$,
and hence $f|_A(x) \to L$ as $x \to c$.

Finally, suppose $f|_A(x) \to L$ as $x \to c$ and let $\epsilon > 0$ be
given.
There is a $\delta' > 0$ such that if $x \in A \setminus \{ c \}$
and $\sabs{x-c} < \delta'$, then $\babs{ f|_A(x)-L } < \epsilon$.
Take $\delta \coloneqq \min \{ \delta', \alpha \}$.
Now suppose $x \in S \setminus \{ c \}$ and
$\sabs{x-c} < \delta$.  As $\sabs{x-c} < \alpha$, we find $x \in A \setminus \{ c \}$,
and as $\sabs{x-c} < \delta'$, 
we get $\babs{f(x)-L} = \babs{ f|_A(x)-L } < \epsilon$.
\end{proof}

The hypothesis on $A$ in the proposition is necessary.  For an arbitrary
restriction we generally get an implication in only one direction
(see \exerciseref{exercise:restrictionlimitexercise}).
The usual notation for the limit is
\begin{equation*}
\lim_{\substack{x \to c\\x \in A}} f(x) \coloneqq \lim_{x \to c} f|_A(x) .
\end{equation*}
A common use of restriction with respect to limits, which does not satisfy
the hypothesis in the proposition, is the
\emph{\myindex{one-sided limit}}%
\footnote{%
One sees a plethora of one-sided limit notations.
E.g.,
$\lim\limits_{\substack{x \to c\\x < c}} f(x)$,
$\lim\limits_{x \uparrow c} f(x)$, or
$\lim\limits_{x \nearrow c} f(x)$ for
$\lim\limits_{x \to c^-} f(x)$.}.

\begin{defn} \label{defn:onesidedlimits}
Let $f \colon S \to \R$ be a function and let $c \in \R$.  If
$c$ is a cluster point of
$S \cap (c,\infty)$
and the limit
of the restriction of $f$ to $S \cap (c,\infty)$ 
 as $x \to c$ exists, define
\glsadd{not:onesidedlim}
\begin{equation*}
\lim_{x \to c^+} f(x) \coloneqq \lim_{x\to c} f|_{S \cap (c,\infty)}(x) .
\end{equation*}
Similarly, if $c$ is a cluster point of 
$S \cap (-\infty,c)$ and the limit of the restriction as $x \to c$
exists, define
\begin{equation*}
\lim_{x \to c^-} f(x) \coloneqq \lim_{x\to c} f|_{S \cap (-\infty,c)}(x) .
\end{equation*}
\end{defn}

\propref{prop:limrest} does not apply to one-sided limits.
It is possible to have one-sided limits, but no limit at a point.  For
example, define $f \colon \R \to \R$ by $f(x) \coloneqq 1$ for $x < 0$ and
$f(x) \coloneqq 0$ for $x \geq 0$.
We leave it to the reader to verify that
\begin{equation*}
\lim_{x \to 0^-} f(x) = 1, \qquad
\lim_{x \to 0^+} f(x) = 0, \qquad
\lim_{x \to 0} f(x) \quad \text{does not exist.}
\end{equation*}
All is not lost, however, for we have the following replacement.

\begin{prop} \label{prop:onesidedlimits}
Let $S \subset \R$ be such that $c$ is a cluster point
of both $S \cap (-\infty,c)$ and $S \cap (c,\infty)$, let
$f \colon S \to \R$ be a function, and let $L \in \R$.  Then $c$ is a cluster point of $S$ and
\begin{equation*}
\lim_{x \to c} f(x) = L
\qquad \text{if and only if} \qquad
\lim_{x \to c^-} f(x) =
\lim_{x \to c^+} f(x) =
L .
\end{equation*}
\end{prop}

That is, a limit at $c$ exists if and only if both one-sided limits exist and are equal.  The
proof is a straightforward application of the definition of limit
and is left as an exercise.  The key point is that
$\bigl( S \cap (-\infty,c) \bigr) \cup \bigl( S \cap (c,\infty) \bigr)
= S \setminus \{ c \}$.

\subsection{Exercises}

\begin{exercise}
Find the limit (and prove it of course) or prove that the limit does not
exist:

\medskip

\noindent
\begin{tabular}{lllll}
a)
$\displaystyle
\lim_{x\to c} \sqrt{x}
$, for $c \geq 0$
& &
b)
$\displaystyle
\lim_{x\to c} x^2+x+1
$, for $c \in \R$
& &
c)
$\displaystyle
\lim_{x\to 0} x^2 \cos (\nicefrac{1}{x})
$
\\
d)
$\displaystyle
\lim_{x\to 0} \sin(\nicefrac{1}{x}) \cos (\nicefrac{1}{x})
$
& &
e)
$\displaystyle
\lim_{x\to 0} \sin(x) \cos (\nicefrac{1}{x})
$ & 
\end{tabular}
\end{exercise}

\begin{exercise}
Prove \corref{fconstineq:cor}.
\end{exercise}

\begin{exercise}
Prove \corref{fsqueeze:cor}.
\end{exercise}

\begin{exercise}
Prove \corref{falg:cor}.
\end{exercise}

\begin{exercise}
Let $A \subset S$.  Show that if $c$ is a cluster point of $A$, then $c$
is a cluster point of $S$.  Note the difference from
\propref{prop:limrest}.
\end{exercise}

\begin{exercise} \label{exercise:restrictionlimitexercise}
Let $A \subset S$.  Suppose $c$ is a cluster point of $A$ and
it is also a cluster point of $S$.
Let $f \colon S \to \R$ be a function.  Show that if
$f(x) \to L$ as $x \to c$, then
$f|_A(x) \to L$ as $x \to c$.
Note the difference from
\propref{prop:limrest}.
\end{exercise}

\begin{exercise}
Find an example of a function $f \colon [-1,1] \to \R$, where for
$A\coloneqq [0,1]$, we have
$f|_A(x) \to 0$ as $x \to 0$, but the limit of $f(x)$ as $x \to 0$
does not exist.  Note why you cannot apply
\propref{prop:limrest}.
\end{exercise}

\begin{exercise}
Find example functions $f$ and $g$ such that the limit of neither $f(x)$
nor $g(x)$ exists as $x \to 0$, but such that the limit of $f(x)+g(x)$ exists
as $x \to 0$.
\end{exercise}

\begin{exercise} \label{exercise:contlimitcomposition}
Let $c_1$ be a cluster point of $A \subset \R$ and $c_2$ be
a cluster point of $B \subset \R$.  Suppose 
$f \colon A \to B$ and $g \colon B \to \R$ are functions
such that
$f(x) \to c_2$ as $x \to c_1$ and
$g(y) \to L$ as $y \to c_2$.  If $c_2 \in B$, also suppose that $g(c_2) = L$.
Let $h(x) \coloneqq g\bigl(f(x)\bigr)$ and show
$h(x) \to L$ as $x \to c_1$.
Hint: Note that $f(x)$ could equal $c_2$ for many $x \in A$
(see also
\exerciseref{exercise:contlimitbadcomposition}).
\end{exercise}

\begin{exercise}
Suppose $f \colon \R \to \R$ is a function such that for every 
sequence $\{x_n\}_{n=1}^\infty$ in $\R$, the sequence
$\bigl\{ f(x_n) \bigr\}_{n=1}^\infty$ converges.  Prove that
$f$ is constant, that is,
$f(x) = f(y)$ for all $x,y \in \R$.
\end{exercise}

\begin{exercise} \label{exercise:seqflimitalt}
Prove the following stronger version of one direction of
\lemmaref{seqflimit:lemma}:
Let $S \subset \R$, $c$ be a cluster point of $S$, and $f \colon S \to
\R$ be a function.
Suppose that for every sequence $\{x_n\}_{n=1}^\infty$ in $S \setminus \{c\}$ such that
$\lim_{n\to\infty} x_n = c$ the sequence $\bigl\{ f(x_n) \bigr\}_{n=1}^\infty$ is convergent.
Then show that the limit of $f(x)$ as $x \to c$ exists.
\end{exercise}

\begin{exercise}
Prove \propref{prop:onesidedlimits}.
\end{exercise}

\begin{exercise}
Suppose $S \subset \R$ and $c$ is a cluster point of $S$.  Suppose $f \colon
S \to \R$ is bounded.  Show that there exists a sequence
$\{ x_n \}_{n=1}^\infty$
with $x_n \in S \setminus \{ c \}$ and $\lim_{n\to\infty} x_n = c$ such that
$\bigl\{ f(x_n) \bigr\}_{n=1}^\infty$ converges.
\end{exercise}

\begin{exercise}[Challenging] \label{exercise:contlimitbadcomposition}
Show that the hypothesis that $g(c_2) = L$ in
\exerciseref{exercise:contlimitcomposition} is necessary.  That is, find $f$
and $g$ such that $f(x) \to c_2$ as $x \to c_1$ and
$g(y) \to L$ as $y \to c_2$, but $g\bigl(f(x)\bigr)$ does not go to $L$
as $x \to c_1$.
\end{exercise}

\begin{exercise}
Show that the condition of being a cluster point is necessary to have a
reasonable definition of a limit.  That is, suppose $c$ is not a cluster
point of $S \subset \R$, and $f \colon S \to \R$ is a function.  Show that
every $L$ would satisfy the definition of limit at $c$ without the condition
on $c$ being a cluster point.
\end{exercise}

\begin{exercise}
\leavevmode
\begin{enumerate}[a)]
\item
Prove \corref{fabs:cor}.
\item
Find an example showing that the converse of
the corollary does not hold.
\end{enumerate}
\end{exercise}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

\sectionnewpage
\section{Continuous functions}
\label{sec:cont}

%mbxINTROSUBSECTION

\sectionnotes{2--2.5 lectures}

A high-school criterion for the concept of continuity
is that a function is continuous if
we can draw its graph without lifting the pen from the paper.  While that
intuitive concept may be useful in simple situations, we require
rigor.  The following definition took three great mathematicians
(Bolzano, Cauchy, and finally Weierstrass) to formulate correctly and its final
form dates only to the late 1800s.

\subsection{Definition and basic properties}

\begin{defn}
Suppose $S \subset \R$ and $c \in S$.
We say $f \colon S \to \R$
is \emph{continuous at $c$}\index{continuous at $c$}
if for every $\epsilon > 0$
there is a $\delta > 0$ such that whenever $x \in S$ and $\sabs{x-c} <
\delta$, we have
$\babs{f(x)-f(c)} < \epsilon$.

%\medskip

When $f \colon S \to \R$ is continuous at all $c \in S$, then we simply say
$f$ is a \emph{\myindex{continuous function}}\index{function!continuous}.
\end{defn}
\begin{myfigureht}
\myincludepdft{contigr}{%
A diagram of the graph of a function y equals f of x,
near a point x equals c.  The graph is shown and the
strip within delta of c is shown and a strip within
epsilon of f of c is shown.  The box that fits within
both strips is shaded, and within this range the graph
fits within the vertical range.}
\caption{For $\sabs{x-c} < \delta$, the graph of $f(x)$ should be within the gray region.\label{fig:contigr}}
\end{myfigureht}

If $f$ is continuous for all $c \in A$, we say
$f$ is \emph{continuous on $A \subset S$}.  A straightforward
exercise (\exerciseref{exercise:restrictioncontinuous})
shows that this implies that $f|_A$ is continuous, although
the converse does not hold (as we will see in \exampleref{example:removablediscont}).

Continuity may be the most important definition to understand in analysis,
and it is not an easy one.  See \figureref{fig:contigr}.
Note that $\delta$ depends not only
on $\epsilon$, but also on $c$;  we need not pick
one $\delta$ for all $c \in S$.
It is no accident 
that the definition of continuity is similar to the definition of a
limit of a function.  The main feature of continuous functions
is that these are precisely the functions that behave nicely with limits.

\begin{prop} \label{contbasic:prop}
Consider a function $f \colon S \to \R$ defined on a set  $S \subset \R$
and let $c \in S$.
Then:
%\begin{enumerate}[(i),itemsep=0.5\itemsep,parsep=0.5\parsep,topsep=0.5\topsep,partopsep=0.5\partopsep]
\begin{enumerate}[(i)]
\item \label{contbasic:prop:i}
If $c$ is not a cluster point of $S$, then $f$ is continuous at $c$.
\item \label{contbasic:prop:ii}
If $c$ is a cluster point of $S$, then $f$ is continuous at $c$
if and only if the limit of $f(x)$ as $x \to c$ exists and
\begin{equation*}
\lim_{x\to c} f(x) = f(c) .
\end{equation*}
\item \label{contbasic:prop:iii}
The function $f$ is continuous at $c$ if and only if for every sequence
$\{ x_n \}_{n=1}^\infty$
where $x_n \in S$ and $\lim\limits_{n\to\infty} x_n = c$, the sequence
$\bigl\{ f(x_n) \bigr\}_{n=1}^\infty$ converges
to $f(c)$.
\end{enumerate}
\end{prop}

\begin{proof}
\pagebreak[2]
We start with \ref{contbasic:prop:i}.  Suppose $c$ is not a cluster point of
$S$.  Then there exists a $\delta > 0$
such that $S \cap (c-\delta,c+\delta) = \{ c \}$.
For any $\epsilon > 0$, simply pick this given $\delta$.
The only $x \in S$ such that $\sabs{x-c} < \delta$ is $x=c$.  Then
$\babs{f(x)-f(c)} = \babs{f(c)-f(c)} = 0 < \epsilon$.

Let us move to \ref{contbasic:prop:ii}.
Suppose $c$ is a cluster point of $S$.  Let us first suppose
that $\lim_{x\to c} f(x) = f(c)$.  Then for every $\epsilon > 0$,
there is a $\delta > 0$ such that if $x \in S \setminus \{ c \}$
and $\sabs{x-c} < \delta$, then $\babs{f(x)-f(c)} < \epsilon$.
Also $\babs{f(c)-f(c)} = 0 < \epsilon$, so the definition of continuity at
$c$ is satisfied.  On the other hand, suppose $f$ is continuous
at $c$.  For every $\epsilon > 0$, there exists a $\delta > 0$
such that for $x \in S$ where $\sabs{x-c} < \delta$, we have
$\babs{f(x)-f(c)} < \epsilon$.  Then the statement is, of course, still true if
$x \in S \setminus \{ c \} \subset S$.  Therefore, $\lim_{x\to c} f(x) =
f(c)$.

For \ref{contbasic:prop:iii}, first suppose $f$ is continuous at $c$.
Let $\{ x_n \}_{n=1}^\infty$
be a sequence such that $x_n \in S$ and $\lim_{n\to\infty} x_n = c$.  Let $\epsilon > 0$
be given.  Find a $\delta > 0$ such that $\babs{f(x)-f(c)} < \epsilon$
for all $x \in S$ where $\sabs{x-c} < \delta$.  Find an $M \in \N$
such that for $n \geq M$, we have $\sabs{x_n-c} < \delta$.  Then for
$n \geq M$, we have that $\babs{f(x_n)-f(c)} < \epsilon$,
so $\bigl\{ f(x_n) \bigr\}_{n=1}^\infty$
converges to $f(c)$.

We prove the other direction of \ref{contbasic:prop:iii} by contrapositive.
Suppose $f$ is not
continuous at $c$.  Then there exists an $\epsilon > 0$
such that for every $\delta > 0$, there exists an $x \in S$
such that $\sabs{x-c} < \delta$ and $\babs{f(x)-f(c)} \geq \epsilon$.
Define a sequence $\{ x_n \}_{n=1}^\infty$ as follows.
Let $x_n \in S$ be such that $\sabs{x_n-c} < \nicefrac{1}{n}$
and $\babs{f(x_n)-f(c)} \geq \epsilon$.
Now $\{ x_n \}_{n=1}^\infty$ is
a sequence in $S$ such that
$\lim_{n\to\infty} x_n = c$ and such that
$\babs{f(x_n)-f(c)} \geq \epsilon$ for all $n \in \N$.
Thus $\bigl\{ f(x_n) \bigr\}_{n=1}^\infty$
does not converge to $f(c)$.  It may or may not converge, but it definitely
does not converge to $f(c)$.  
\end{proof}

The last item in the proposition is particularly powerful.  It allows us to
quickly apply what we know about limits of sequences to continuous functions
and even to prove that certain functions are continuous.
It can also be strengthened (see \exerciseref{exercise:contseqalt}).

\begin{example}
The function $f \colon (0,\infty) \to \R$
defined by $f(x) \coloneqq \nicefrac{1}{x}$ is continuous.

Proof: Fix $c \in (0,\infty)$.  
Let $\{ x_n \}_{n=1}^\infty$ be a sequence in $(0,\infty)$ such that
$\lim_{n\to\infty} x_n = c$.  Then
\begin{equation*}
f(c) = \frac{1}{c}
=
\frac{1}{\lim_{n\to\infty} x_n}
=
\lim_{n \to \infty} \frac{1}{x_n}
=
\lim_{n \to \infty} f(x_n) .
\end{equation*}
Thus $f$ is continuous at $c$.  As $f$ is continuous at all $c \in
(0,\infty)$, $f$ is continuous.
\end{example}

We have previously shown $\lim_{x \to c} x^2 = c^2$ directly.  Therefore,
the function $x^2$ is continuous.  The last item of \propref{contbasic:prop} and the continuity of
algebraic operations with respect to limits of sequences,
\propref{prop:contalg}, give a quick proof of a much more general result.

\begin{prop}
Let $f \colon \R \to \R$ be a \emph{\myindex{polynomial}}.  That is,
\begin{equation*}
f(x) = a_d x^d + a_{d-1} x^{d-1} + \cdots + a_1 x + a_0 ,
\end{equation*}
for some constants $a_0, a_1, \ldots, a_d$.
Then $f$ is continuous.
\end{prop}

\begin{proof}
Fix $c \in \R$.  
Let $\{ x_n \}_{n=1}^\infty$ be a sequence such that
$\lim_{n\to\infty} x_n = c$.  Then
\begin{equation*}
\begin{split}
f(c) &=
a_d c^d + a_{d-1} c^{d-1} + \cdots + a_1 c + a_0 
\\
&= 
a_d {\left(\lim_{n\to\infty} x_n\right)}^d
+ a_{d-1} {\left(\lim_{n\to\infty} x_n\right)}^{d-1}
+ \cdots
+ a_1 \left(\lim_{n\to\infty} x_n\right) + a_0 
\\
& =
\lim_{n \to \infty}
\left(
a_d x_n^d + a_{d-1} x_n^{d-1} + \cdots + a_1 x_n + a_0 
\right)
=
\lim_{n \to \infty}
f(x_n) .
\avoidbreak
\end{split}
\end{equation*}
Thus $f$ is continuous at $c$.  As $f$ is continuous at all $c \in \R$,
$f$ is continuous.
\end{proof}

By similar reasoning, or by appealing to \corref{falg:cor},
we can prove the following proposition.  The proof is left as an
exercise.

\begin{prop} \label{contalg:prop}
Let $f \colon S \to \R$ and $g \colon S \to \R$ be functions
continuous at $c \in S$.
\begin{enumerate}[(i)]
\item The function $h \colon S \to \R$ defined by
$h(x) \coloneqq f(x)+g(x)$ is continuous at $c$.
\item The function $h \colon S \to \R$ defined by
$h(x) \coloneqq f(x)-g(x)$ is continuous at $c$.
\item The function $h \colon S \to \R$ defined by
$h(x) \coloneqq f(x)g(x)$ is continuous at $c$.
\item If $g(x) \neq 0$ for all $x \in S$, the function $h \colon S \to \R$
given by $h(x) \coloneqq \frac{f(x)}{g(x)}$ is continuous at $c$.
\end{enumerate}
\end{prop}

\begin{example} \label{sincos:example}
The functions $\sin(x)$ and $\cos(x)$ are continuous.
In the following computations we use the sum-to-product
trigonometric identities.  We also use the simple facts that
$\babs{\sin(x)} \leq \sabs{x}$, $\babs{\cos(x)} \leq 1$,
and $\babs{\sin(x)} \leq 1$.
\begin{equation*}
\begin{split}
\babs{\sin(x)-\sin(c)} & =
\abs{
2 \sin \left( \frac{x-c}{2} \right) \cos \left( \frac{x+c}{2} \right)
}
\\
& =
2
\abs{ \sin \left( \frac{x-c}{2} \right) }
\abs{ \cos \left( \frac{x+c}{2} \right) }
\\
& \leq
2
\abs{ \sin \left( \frac{x-c}{2} \right) }
\\
& \leq
2
\abs{ \frac{x-c}{2} }
= \sabs{x-c}
\end{split}
\end{equation*}
\begin{equation*}
\begin{split}
\babs{\cos(x)-\cos(c)} & =
\abs{
-2 \sin \left( \frac{x-c}{2} \right) \sin \left( \frac{x+c}{2} \right)
}
\\
& =
2
\abs{ \sin \left( \frac{x-c}{2} \right) }
\abs{ \sin \left( \frac{x+c}{2} \right) }
\\
& \leq
2
\abs{ \sin \left( \frac{x-c}{2} \right) }
\\
& \leq
2
\abs{ \frac{x-c}{2} }
= \sabs{x-c}
\end{split}
\end{equation*}

The claim that $\sin$ and $\cos$ are continuous follows by taking an
arbitrary sequence $\{ x_n \}_{n=1}^\infty$ converging to $c$, or by applying the
definition of continuity directly.  Details are left to the
reader.
\end{example}

\subsection{Composition of continuous functions}

You probably already realized that one of the basic tools in
constructing complicated functions out of simple ones is composition.
Recall that for two functions $f$ and $g$,
the composition $f \circ g$ is defined by
$(f \circ g)(x) \coloneqq f\bigl(g(x)\bigr)$.
A composition of
continuous functions is again
continuous.

\begin{prop} \label{prop:compositioncont}
Let $A, B \subset \R$ and $f \colon B \to \R$ and $g \colon A \to B$ be
functions.  If $g$ is continuous at $c \in A$ and
$f$ is continuous at $g(c)$, then $f \circ g \colon A \to \R$ is continuous
at $c$.
\end{prop}

\begin{proof}
Let $\{ x_n \}_{n=1}^\infty$ be a sequence in $A$ such that
$\lim_{n\to\infty} x_n = c$.
As $g$ is continuous at $c$, we have  that$\bigl\{ g(x_n) \bigr\}_{n=1}^\infty$ converges to $g(c)$.
As $f$ is continuous at $g(c)$, we have that $\bigl\{ f\bigl(g(x_n)\bigr)
\bigr\}_{n=1}^\infty$ converges
to $f\bigl(g(c)\bigr)$.
Thus $f \circ g$ is continuous at $c$.
\end{proof}

\begin{example}
Claim: \emph{${\bigl(\sin(\nicefrac{1}{x})\bigr)}^2$ is a continuous
function on $(0,\infty)$.}

Proof: The function $\nicefrac{1}{x}$ is continuous on
$(0,\infty)$ and $\sin(x)$ is continuous on $(0,\infty)$ (actually
on $\R$, but $(0,\infty)$ is the range for $\nicefrac{1}{x}$).
Hence, the composition $\sin(\nicefrac{1}{x})$ is continuous.  Also,
$x^2$ is continuous on the interval $[-1,1]$ (the range of $\sin$).
Thus, the composition
${\bigl(\sin(\nicefrac{1}{x})\bigr)}^2$ is continuous on $(0,\infty)$.
\end{example}

\subsection{Discontinuous functions}

When $f$ is not continuous at $c$, we
say $f$ is \emph{\myindex{discontinuous}} at $c$, or that it has a
\emph{\myindex{discontinuity}} at~$c$.
The following proposition is a useful test and follows immediately
from the third item of \propref{contbasic:prop}.

\begin{prop}
Let $f \colon S \to \R$ be a function and $c \in S$.  Suppose 
there exists a sequence $\{ x_n \}_{n=1}^\infty$, $x_n \in S$ for all $n$,
and $\lim_{n\to\infty} x_n = c$
such that $\bigl\{ f(x_n) \bigr\}_{n=1}^\infty$
does not converge to $f(c)$.
Then $f$ is discontinuous at $c$.
\end{prop}

Again, 
saying that 
$\bigl\{ f(x_n) \bigr\}_{n=1}^\infty$ does not converge to $f(c)$
means that it
either does not converge
at all, or it converges to something
other than $f(c)$.

\begin{example} \label{example:jumpdiscont}
The function $f \colon \R \to \R$ defined by
\begin{equation*}
f(x) \coloneqq 
\begin{cases}
-1 & \text{if } x < 0, \\
1 & \text{if } x \geq 0
\end{cases}
\end{equation*}
is not continuous at 0.

Proof: Consider $\{ \nicefrac{-1}{n} \}_{n=1}^\infty$, which converges to 0.  Then
$f(\nicefrac{-1}{n}) = -1$ for every $n$,
and so
$\lim_{n\to\infty} f(\nicefrac{-1}{n}) = -1$, but $f(0) = 1$.  Thus the function is
not continuous at 0.  See
\figureref{fig:jumpdiscont}.

\begin{myfigureht}
\myincludegraphics{jumpdiscont}{%
A graph of a function that is minus 1 for negative x
and 1 for nonnegative x.  The points minus 1,
minus a half, minus a third, etc. are marked on
the x-axis and the corresponding values on
the graph are marked with dots.}
\caption{Jump discontinuity.  The values of
$f(\nicefrac{-1}{n})$ and $f(0)$ are marked as black dots.\label{fig:jumpdiscont}}
\end{myfigureht}

Notice that $f(\nicefrac{1}{n}) = 1$ for all $n \in \N$.
Hence, $\lim_{n\to\infty} f(\nicefrac{1}{n}) = f(0) = 1$.  So
$\bigl\{ f(x_n) \bigr\}_{n=1}^\infty$ may converge to $f(0)$
for some specific
sequence $\{ x_n \}_{n=1}^\infty$ going to 0,
despite the function being discontinuous at 0.

Finally, consider $f\Bigl(\frac{{(-1)}^n}{n}\Bigr) = {(-1)}^n$.
This sequence diverges.
\end{example}

\begin{example}
For an extreme example, consider the
\emph{\myindex{Dirichlet function}}\footnote{Named after the German
mathematician
\href{https://en.wikipedia.org/wiki/Peter_Gustav_Lejeune_Dirichlet}{Johann Peter Gustav Lejeune
Dirichlet}
(1805--1859).}.
\begin{equation*}
f(x) \coloneqq
\begin{cases}
1 & \text{if } x \text{ is rational,} \\
0 & \text{if } x \text{ is irrational.}
\end{cases}
\avoidbreak
\end{equation*}
The function $f$ is discontinuous at all $c \in \R$.

Proof:
If $c$ is rational, take a sequence $\{ x_n \}_{n=1}^\infty$
of irrational numbers in $(0,1)$ such that $\lim_{n\to\infty} x_n = c$ (why can we?).  Then $f(x_n) = 0$
and so $\lim_{n\to\infty} f(x_n) = 0$, but $f(c) = 1$.
If $c$ is irrational, take a sequence
$\{ x_n \}_{n=1}^\infty$
of rational numbers in $(0,1)$
that converges to $c$ (why can we?).  Then $\lim_{n\to\infty} f(x_n) = 1$, but $f(c) = 0$.
\end{example}

Let us test the limits of our intuition.  Can
there exist a function continuous at all irrational numbers, but
discontinuous at all rational numbers?  There are rational numbers
arbitrarily close to any irrational number.  Perhaps strangely, the
answer is yes, such a function exists.  The following example is called the
\emph{\myindex{Thomae function}}\footnote{Named after the German
mathematician
\href{https://en.wikipedia.org/wiki/Carl_Johannes_Thomae}{Carl Johannes Thomae}
(1840--1921).} or the
\emph{\myindex{popcorn function}}.

\begin{example} \label{popcornfunction:example}
Define $f \colon (0,1) \to \R$ as
\begin{equation*}
f(x) \coloneqq 
\begin{cases}
\nicefrac{1}{k} & \text{if } x=\nicefrac{m}{k}, \text{ where } m,k \in \N
\text{ and have no common divisors (lowest terms),} \\
0 & \text{if } x \text{ is irrational.}
\end{cases}
\end{equation*}
See the graph of $f$
in \figureref{popcornfig}.
We claim that
$f$ is continuous at all irrational $c$ and 
discontinuous at all rational $c$.
\begin{myfigureht}
\myincludegraphics{popcornfig}{%
A graph of a function that is composed of dots.
These dots seem arranged in what seems like
various triangles, but it is clear this
function is discontinuous at all those points,
and there are more and more of such points
the closer we get to the x-axis.}
\caption{Graph of the \myquote{popcorn function.}\label{popcornfig}}
\end{myfigureht}

Proof:
Let $c = \nicefrac{m}{k}$ be rational and in lowest terms.  Take a sequence of
irrational numbers $\{ x_n \}_{n=1}^\infty$ such that $\lim_{n\to\infty} x_n = c$.  Then
$\lim_{n\to\infty} f(x_n) = \lim_{n\to\infty} 0 = 0$, but $f(c) = \nicefrac{1}{k} \neq 0$.  So $f$
is discontinuous at $c$.

Now let $c$ be irrational, so $f(c) = 0$.  Take a sequence 
$\{ x_n \}_{n=1}^\infty$ in $(0,1)$ such that $\lim_{n\to\infty} x_n = c$.
Given $\epsilon > 0$, find $K \in \N$ such
that $\nicefrac{1}{K} < \epsilon$
by the \hyperref[thm:arch:i]{Archimedean property}.
If $\nicefrac{m}{k} \in (0,1)$ and $m,k \in \N$, then $0 < m < k$.
So there are only finitely many rational numbers in $(0,1)$
whose denominator $k$ in lowest terms is less than $K$.
As $\lim_{n\to\infty} x_n = c$, every number not equal to $c$ can appear at most
finitely many times in $\{ x_n \}_{n=1}^\infty$.
Hence,
there is an $M$ such that for $n \geq M$, all the numbers $x_n$
that are rational
have a denominator larger than or equal to $K$.  Thus for $n \geq M$,
\begin{equation*}
\babs{f(x_n) - 0} = f(x_n) \leq \nicefrac{1}{K} < \epsilon .
\end{equation*}
Therefore, $f$ is continuous at irrational $c$.
\end{example}

Let us end on an easier example.

\begin{example} \label{example:removablediscont}
Define
$g \colon \R \to \R$ by $g(x) \coloneqq 0$ if $x \neq 0$ and
$g(0) \coloneqq 1$.  Then $g$ is not continuous at zero, but continuous everywhere else (why?).
The point $x=0$ is called a \emph{\myindex{removable discontinuity}}.  That
is because if we would change the definition of $g$, by insisting that
$g(0)$ be $0$, we would obtain a continuous function.  On the other hand,
let $f$ be the function of \exampleref{example:jumpdiscont}.
Then $f$ does not have a
removable discontinuity at $0$.  No matter how we would define $f(0)$, the function
would still fail to be continuous.  The difference is that 
$\lim_{x\to 0} g(x)$ exists while
$\lim_{x\to 0} f(x)$ does not.

We stay with this example to show another phenomenon.
Let $A \coloneqq \{ 0 \}$, then $g|_A$ is continuous (why?), while $g$ is not continuous on $A$.
Similarly, if $B \coloneqq \R \setminus \{0 \}$, then $g|_B$ is also continuous,
and $g$ is in fact continuous on $B$.
\end{example}

\subsection{Exercises}

\begin{exercise}
Using the definition of continuity directly, prove that
$f \colon \R \to \R$ defined by
$f(x) \coloneqq x^2$ is continuous.
\end{exercise}

\begin{exercise}
Using the definition of continuity directly, prove that
$f \colon (0,\infty) \to \R$ defined by
$f(x) \coloneqq \nicefrac{1}{x}$ is continuous.
\end{exercise}

\begin{exercise}
Define $f \colon \R \to \R$ by
\begin{equation*}
f(x) \coloneqq
\begin{cases}
x & \text{if } x \text{ is rational,} \\
x^2 & \text{if } x \text{ is irrational.}
\end{cases}
\avoidbreak
\end{equation*}
Using the definition of continuity directly, prove that
$f$ is continuous at $1$ and discontinuous at $2$.
\end{exercise}

\begin{exercise}
Define $f \colon \R \to \R$ by
\begin{equation*}
f(x) \coloneqq
\begin{cases}
\sin(\nicefrac{1}{x}) & \text{if } x \neq 0, \\
0 & \text{if } x=0.
\end{cases}
\avoidbreak
\end{equation*}
Is $f$ continuous?  Prove your assertion.
\end{exercise}

\begin{exercise}
Define $f \colon \R \to \R$ by
\begin{equation*}
f(x) \coloneqq
\begin{cases}
x \sin(\nicefrac{1}{x}) & \text{if } x \neq 0, \\
0 & \text{if } x=0.
\end{cases}
\avoidbreak
\end{equation*}
Is $f$ continuous?  Prove your assertion.
\end{exercise}

\begin{exercise}
Prove \propref{contalg:prop}.
\end{exercise}

\begin{exercise} \label{exercise:restrictioncontinuous}
Let $S \subset \R$ and $A \subset S$.  Let $f \colon S \to \R$
be a continuous function.
Prove that the restriction $f|_A$ is continuous.
\end{exercise}

\begin{exercise}
Suppose $S \subset \R$ such that $(c-\alpha,c+\alpha) \subset S$ for some $c \in \R$
and $\alpha > 0$.
Let $f \colon S \to \R$ be a function and $A \coloneqq (c-\alpha,c+\alpha)$.  Prove that
if $f|_A$ is continuous at $c$, then $f$ is continuous at $c$.
\end{exercise}

\begin{exercise}
Give an example of functions $f \colon \R \to \R$ and $g \colon \R \to \R$
such that the function $h$, defined by $h(x) \coloneqq f(x) + g(x)$, is continuous,
but $f$ and $g$ are not continuous.  Can you find $f$ and $g$ that are nowhere
continuous, but $h$ is a continuous function?
\end{exercise}

\begin{exercise}
Let $f \colon \R \to \R$ and 
$g \colon \R \to \R$ be continuous functions.  Suppose that
$f(r) = g(r)$ for all $r \in \Q$.
Show that $f(x) = g(x)$ for all $x \in \R$.
\end{exercise}

\begin{exercise} \label{exercise:positivecontneigh}
Let $f \colon \R \to \R$ be continuous.  Suppose $f(c) > 0$.  Show that
there exists an $\alpha > 0$ such that for all $x \in (c-\alpha,c+\alpha)$,
we have $f(x) > 0$.
\end{exercise}

\begin{exercise}
Let $f \colon \Z \to \R$ be a function.  Show that $f$ is continuous.
\end{exercise}

\begin{exercise} \label{exercise:contseqalt}
Let $f \colon S \to \R$ be a function and $c \in S$ such that for every
sequence $\{ x_n \}_{n=1}^\infty$ in $S$ with $\lim_{n\to\infty} x_n = c$, the sequence
$\bigl\{ f(x_n) \bigr\}_{n=1}^\infty$ converges.  Show that $f$ is continuous at $c$.
\end{exercise}

\begin{exercise}
Suppose $f \colon [-1,0] \to \R$ and $g \colon [0,1] \to \R$ are continuous
and $f(0) = g(0)$.  Define $h \colon [-1,1] \to \R$ by 
$h(x) \coloneqq f(x)$ if $x \leq 0$ and $h(x) \coloneqq g(x)$ if $x > 0$.  Show that
$h$ is continuous.
\end{exercise}

\begin{exercise}
Suppose $g \colon \R \to \R$ is a continuous function such that $g(0) = 0$,
and suppose $f \colon \R \to \R$ is such that
$\babs{f(x)-f(y)} \leq g(x-y)$ for all $x$ and $y$.  Show that $f$ is
continuous.
\end{exercise}

\begin{exercise}[Challenging]
Suppose $f \colon \R \to \R$ is continuous at $0$ and
such that $f(x+y) = f(x) + f(y)$ for every $x$ and $y$.
Show that $f(x) = ax$ for some $a \in \R$.
Hint: Show that $f(nx) = nf(x)$, then show $f$ is continuous on $\R$.
Then show that $\nicefrac{f(x)}{x} = f(1)$ for all rational $x \neq 0$.
\end{exercise}

\begin{exercise} \label{exercise:minmaxcont}
Suppose $S \subset \R$ and
let $f \colon S \to \R$ and
$g \colon S \to \R$ be continuous functions.
Define $p \colon S \to \R$ by
$p(x) \coloneqq \max \bigl\{ f(x) , g(x) \bigr\}$ and
$q \colon S \to \R$ by
$q(x) \coloneqq \min \bigl\{ f(x) , g(x) \bigr\}$.  Prove that $p$ and $q$ are
continuous.
\end{exercise}

\begin{exercise}
Suppose $f \colon [-1,1] \to \R$ is a function continuous at all $x \in
[-1,1] \setminus \{ 0 \}$.  Show that for every $\epsilon$ such
that $0 < \epsilon < 1$, there exists
a function $g \colon [-1,1] \to \R$ continuous on all of $[-1,1]$ such that
$f(x) = g(x)$ for all $x \in [-1,-\epsilon] \cup [\epsilon,1]$, and 
$\babs{g(x)} \leq \babs{f(x)}$ for all $x \in [-1,1]$.
\end{exercise}

\begin{exercise}[Challenging]
A function $f \colon I \to \R$ is \emph{\myindex{convex}} if
whenever $a \leq x \leq b$ for $a,x,b$ in $I$, we have
$f(x) \leq f(a) \frac{b-x}{b-a} + f(b) \frac{x-a}{b-a}$.  In other words,
the line drawn between $\bigl(a,f(a)\bigr)$ and $\bigl(b,f(b)\bigr)$ 
is above the graph of $f$.
\begin{enumerate}[a)]
\item
Prove that
if $I = (\alpha,\beta)$ an open interval and $f \colon I \to \R$ is convex,
then $f$ is continuous.
\item
Find an example of a convex $f \colon [0,1] \to \R$ that is
not continuous.
\end{enumerate}
\end{exercise}


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

\sectionnewpage
\section{Extreme and intermediate value theorems}
\label{sec:minmaxint}

%mbxINTROSUBSECTION

\sectionnotes{1.5 lectures}

Continuous functions on closed and bounded intervals
are quite well behaved.

\subsection{Min-max or extreme value theorem}

Recall that $f \colon [a,b] \to \R$ is
\emph{bounded}\index{bounded function}\index{function!bounded}
if there exists a $B \in \R$ such that
$\babs{f(x)} \leq B$ for all $x \in [a,b]$.
For a continuous function on a closed and bounded interval, we have the following lemma.

\begin{lemma}
A continuous function $f \colon [a,b] \to \R$ is bounded.
\end{lemma}

\begin{proof}
We prove the claim by contrapositive.  Suppose $f$ is not bounded.
Then for each
$n \in \N$, there is an $x_n \in [a,b]$ such that
\begin{equation*}
\babs{f(x_n)} \geq n .
\end{equation*}
The sequence $\{ x_n \}_{n=1}^\infty$ is bounded as $a \leq x_n \leq b$.
By the \hyperref[thm:bwseq]{Bolzano--Weierstrass theorem},
there is a convergent subsequence $\{ x_{n_i} \}_{i=1}^\infty$.
Let $x \coloneqq \lim_{i\to\infty} x_{n_i}$.
Since $a \leq x_{n_i} \leq b$ for all $i$, we have $a \leq x \leq b$.
The sequence $\bigl\{ f(x_{n_i}) \bigr\}_{i=1}^\infty$ is not bounded 
as 
$\babs{f(x_{n_i})} \geq n_i \geq i$.
Thus $f$ is not continuous at $x$ as
\begin{equation*}
f(x)
=
f\Bigl( \lim_{i\to\infty} x_{n_i} \Bigr) ,
\qquad \text{but} \qquad
\lim_{i\to\infty} f(x_{n_i}) \enspace \text{does not exist.} \qedhere
\end{equation*}
\end{proof}

Notice a key point of the proof.
Boundedness of $[a,b]$ allows us to use Bolzano--Weierstrass,
while the fact that it is closed gives us that the limit is back in $[a,b]$.
The technique is a common one: Find a sequence with a certain property,
then use Bolzano--Weierstrass to make such a sequence that also converges.

Recall from calculus that $f \colon S \to \R$ achieves an
\emph{\myindex{absolute minimum}}\index{minimum!absolute}%
\index{achieves absolute minimum}
at $c \in S$ if
\begin{equation*}
f(x) \geq f(c) \qquad \text{for all } x \in S.
\end{equation*}
On the other hand, $f$ achieves an 
\emph{\myindex{absolute maximum}}\index{maximum!absolute}%
\index{achieves absolute maximum}
at $c \in S$ if
\begin{equation*}
f(x) \leq f(c) \qquad \text{for all } x \in S.
\end{equation*}
If such a $c \in S$ exists, then we say
$f$ \emph{achieves an absolute minimum (resp.\ absolute maximum) on
$S$}, and we call
$f(c)$ the \emph{absolute minimum (resp.\ absolute maximum)}.
%See \figureref{fig:minmax}.
\begin{myfigureht}
\myincludegraphics{minmax}{%
A graph  of a function on the interval from a to b.
The graph is between two horizontal lines,
one marked as the absolute maximum of f which equals
f of c, and
one marked as the absolute minimum of f which equals
f of d.}
\caption{$f \colon [a,b] \to \R$ achieves an absolute maximum $f(c)$ at
$c$, and an absolute minimum $f(d)$ at $d$.\label{fig:minmax}}
\end{myfigureht}

If $S$ is a closed
and bounded interval, then a continuous $f$
is not just bounded, it must achieve an absolute minimum and an absolute
maximum on $S$.

\begin{thm}[Minimum-maximum theorem / Extreme value theorem]
\index{minimum-maximum theorem}%
\index{maximum-minimum theorem}%
\index{extreme value theorem}%
A continuous function $f \colon [a,b] \to \R$
achieves both an absolute minimum and an absolute maximum on $[a,b]$.
\end{thm}

Again, we remark that it is important that the domain of $f$ is a closed and bounded interval $[a,b]$.

\begin{proof}
The lemma says that $f$ is bounded, so
the set $f\bigl([a,b]\bigr) = \bigl\{ f(x) : x \in [a,b] \bigr\}$ has a supremum and an infimum.
There exist sequences
in the set $f\bigl([a,b]\bigr)$ that approach its supremum and its infimum.
That is, there are sequences
$\bigl\{ f(x_n) \bigr\}_{n=1}^\infty$ and $\bigl\{ f(y_n)
\bigr\}_{n=1}^\infty$, where $x_n$ and $y_n$ are in $[a,b]$,
such that
\begin{equation*}
\lim_{n\to\infty} f(x_n) = \inf f\bigl([a,b]\bigr) \qquad \text{and} \qquad
\lim_{n\to\infty} f(y_n) = \sup f\bigl([a,b]\bigr).
\end{equation*}
We are not done yet; we need to find where the minima and the maxima are.
The problem is that the sequences $\{ x_n \}_{n=1}^\infty$ and
$\{ y_n \}_{n=1}^\infty$ need not converge.
We know $\{ x_n \}_{n=1}^\infty$ and $\{ y_n \}_{n=1}^\infty$ are bounded
(their elements belong to a bounded interval $[a,b]$).
Apply the 
\hyperref[thm:bwseq]{Bolzano--Weierstrass theorem}
to find
convergent subsequences
$\{ x_{n_i} \}_{i=1}^\infty$ and 
$\{ y_{m_i} \}_{i=1}^\infty$.  Let
\begin{equation*}
x \coloneqq \lim_{i\to\infty} x_{n_i}
\qquad \text{and} \qquad
y \coloneqq \lim_{i\to\infty} y_{m_i}.
\end{equation*}
As $a \leq x_{n_i} \leq b$ for all $i$, we have $a \leq x \leq b$.
Similarly, $a \leq y \leq b$.  So $x$ and $y$ are in $[a,b]$.
For convergence sequences,
a limit of a subsequence is the same as the limit of the
sequence.  Also, we can take a limit past the continuous function $f$.  So,
\begin{equation*}
\inf f\bigl([a,b]\bigr) = \lim_{n\to\infty} f(x_n)
= \lim_{i\to\infty} f(x_{n_i}) = 
f \Bigl( \lim_{i\to\infty} x_{n_i} \Bigr) = f(x) .
\end{equation*}
Similarly,
\begin{equation*}
\sup f\bigl([a,b]\bigr) = \lim_{n\to\infty} f(y_n)
= \lim_{i\to\infty} f(y_{m_i}) = 
f \Bigl( \lim_{i\to\infty} y_{m_i} \Bigr) = f(y) .
\end{equation*}
Hence, $f$ achieves an absolute minimum at $x$ and
an absolute maximum at $y$.
\end{proof}

\begin{example}
The function $f(x) \coloneqq x^2+1$ defined on the interval $[-1,2]$ achieves a minimum
at $x=0$ when $f(0) = 1$.  It achieves a maximum at $x=2$ where $f(2) = 5$.
Do note that the domain of definition matters.  If we instead took the domain
to be $[-10,10]$, then $f$ would no longer have a maximum at $x=2$.
Instead,
the maximum would be achieved at $x=10$ and also at $x=-10$.
\end{example}

We show by examples that the different hypotheses of the theorem are
truly needed.

\begin{example}
The function $f \colon \R \to \R$ defined by $f(x) \coloneqq x$
achieves neither a minimum nor a maximum.  So it is important that
we are looking at a bounded interval.
\end{example}

\begin{example}
The function $f \colon (0,1) \to \R$ defined by $f(x) \coloneqq \nicefrac{1}{x}$
achieves neither a minimum, nor a maximum.
It is continuous, but $(0,1)$ is not closed.
The values of the function are
unbounded as we approach $0$.  Also as we approach $x=1$, the values of the
function approach $1$, but $f(x) > 1$ for all $x \in (0,1)$.  There is
no $x \in (0,1)$ such that $f(x) = 1$.  So it is important that
we are looking at a closed interval.
\end{example}

\begin{example}
Continuity is important.
Define $f \colon [0,1] \to \R$ by 
$f(x) \coloneqq \nicefrac{1}{x}$ for $x > 0$ and let $f(0) \coloneqq 0$.
The function does not achieve a maximum.  The domain $[0,1]$ is closed and
bounded, but the problem is that
the function is not continuous at 0.
\end{example}

\subsection{Bolzano's intermediate value theorem}

Bolzano's intermediate value theorem is one of the cornerstones of analysis.
It is sometimes only called the intermediate value theorem or just
Bolzano's theorem.  To prove Bolzano's theorem we prove the
following simpler lemma.

\begin{lemma} \label{IVT:lemma}
Let $f \colon [a,b] \to \R$ be a continuous function.
Suppose $f(a) < 0$ and $f(b) > 0$. 
Then there exists a number $c \in (a,b)$
such that $f(c) = 0$.
\end{lemma}

\begin{proof}
We define two sequences $\{ a_n \}_{n=1}^\infty$
and $\{ b_n \}_{n=1}^\infty$ inductively:
\begin{enumerate}[(i)]
\item Let $a_1 \coloneqq a$ and $b_1 \coloneqq b$.
\item If $f\left(\frac{a_n+b_n}{2}\right) \geq 0$, let $a_{n+1} \coloneqq a_n$ and
$b_{n+1} \coloneqq \frac{a_n+b_n}{2}$.
\item If $f\left(\frac{a_n+b_n}{2}\right) < 0$, let $a_{n+1} \coloneqq \frac{a_n+b_n}{2}$ and
$b_{n+1} \coloneqq b_n$.
\end{enumerate}
\begin{myfigureht}
\myincludegraphics{bisect}{%
Graph of a function that crosses the
x-axis at c going upwards.  The interval
a sub 1 to b sub 1 is marked and f is negative
at a sub 1 and positive at b sub 1.
Next interval is a sub 2 which is equal to a sub 1
and b sub 2 is in the middle of the previous interval.
Again the function is negative at a sub 2 and positive
at b sub 2.  We continue with the intervals
with f being negative on the left and positive
on the right.
The next interval is from a sub 3 which equals
a sub 2 and b sub 3 which
is in the middle of the previous interval.  
For the next interval, a sub 4 is in the middle and
b sub 4 is equal to b sub 3.
The next a sub 5 is in the middle again and b sub 5 
is equal to b sub 4.  The number c is
inside all of these intervals.}
\caption{Finding roots (bisection method).\label{bisectfig}}
\end{myfigureht}
See \figureref{bisectfig} for an example of the first five steps.
If $a_n < b_n$, then $a_n < \frac{a_n+b_n}{2} < b_n$.  So
$a_{n+1} < b_{n+1}$.
As $a_1 = a < b = b_1$,
\hyperref[induction:thm]{induction} gives that
$a_n < b_n$ for all $n$.
Furthermore, $a_n \leq a_{n+1}$ and 
$b_n \geq b_{n+1}$ for all $n$, that is, the sequences are monotone.
As $a_n < b_n \leq b_1 = b$ and 
$b_n > a_n \geq a_1 = a$ for all $n$,
the sequences are also bounded.  Therefore, the
sequences converge.
Let $c \coloneqq \lim_{n\to\infty} a_n$ and $d \coloneqq \lim_{n\to\infty} b_n$,
where also $a \leq c \leq d \leq b$.  We need
to show that $c=d$.
Notice
\begin{equation*}
b_{n+1} - a_{n+1} = \frac{b_n-a_n}{2}.
\end{equation*}
By \hyperref[induction:thm]{induction},
\begin{equation*}
b_n - a_n = \frac{b_1-a_1}{2^{n-1}} = 2^{1-n} (b-a) .
\end{equation*}
As $2^{1-n}(b-a)$ converges to zero, we take the limit as $n$ goes to
infinity to get
\begin{equation*}
d-c = \lim_{n\to\infty} (b_n - a_n) =
\lim_{n\to\infty} 2^{1-n} (b-a) = 0.
\avoidbreak
\end{equation*}
In other words, $c=d$.

By construction, for all $n$,
\begin{equation*}
f(a_n) < 0
\qquad \text{and} \qquad
f(b_n) \geq 0 .
\end{equation*}
Since
$\lim_{n\to\infty} a_n = \lim_{n\to\infty} b_n = c$
and $f$ is continuous at $c$, we may take 
limits in those inequalities:
\begin{equation*}
f(c) = \lim_{n\to\infty} f(a_n) \leq 0
\qquad \text{and} \qquad
f(c) = \lim_{n\to\infty} f(b_n) \geq 0 .
\end{equation*}
As $f(c) \geq 0$ and 
$f(c) \leq 0$, we conclude $f(c) = 0$.
Thus also $c \neq a$ and $c \neq b$, so
$a < c < b$.
\end{proof}

\begin{thm}[Bolzano's intermediate value theorem] \label{IVT:thm}
\index{Bolzano's theorem}
\index{Bolzano's intermediate value theorem}
\index{intermediate value theorem}
Let $f \colon [a,b] \to \R$ be a continuous function.
Suppose $y \in \R$ is such that $f(a) < y < f(b)$
or $f(a) > y > f(b)$.  Then there exists a $c \in (a,b)$
such that $f(c) = y$.
\end{thm}

The theorem says that a continuous function on a closed interval
achieves all the values between the values at the endpoints.

\begin{proof}
If $f(a) < y < f(b)$, then define $g(x) \coloneqq f(x)-y$.  Then 
$g(a) < 0$ and $g(b) > 0$, and we apply \lemmaref{IVT:lemma}
to $g$ to find $c$.  If $g(c) = 0$, then $f(c) = y$.

Similarly, if $f(a) > y > f(b)$, then define $g(x) \coloneqq y-f(x)$.
Again, $g(a) < 0$ and $g(b) > 0$, and we apply \lemmaref{IVT:lemma} to
find $c$.
As before, if $g(c) = 0$, then $f(c) = y$.
\end{proof}

If a function is continuous, then the restriction
to a subset is continuous; if $f \colon S \to \R$ is continuous and
$[a,b] \subset S$, then $f|_{[a,b]}$ is also continuous.  We generally
apply the theorem to a function continuous on some large set $S$,
but we restrict our attention to an interval.


The proof of the lemma tells us how to find the root $c$.  The
proof is not only useful for us pure mathematicians,
it is a useful idea in applied mathematics,
where it is called the \emph{\myindex{bisection method}}.


\begin{example}[Bisection method] %\index{bisection method}
The polynomial $f(x) \coloneqq x^3-2x^2+x-1$ has a real root in $(1,2)$.  We simply
notice that $f(1) = -1$ and $f(2) = 1$.  Hence, there must exist a point $c
\in (1,2)$ such that $f(c) = 0$.  To find a better approximation of
the root we follow the proof of \lemmaref{IVT:lemma}.  
We look at 1.5 and find that $f(1.5) = -0.625$.  Therefore,
there is a root of the polynomial in $(1.5,2)$.  Next, we look at 1.75
and note that $f(1.75) \approx -0.016$.  Hence, there is a root of $f$ in
$(1.75,2)$.  Next, we look at 1.875 and find that $f(1.875) \approx 0.44$,
thus there is a root in $(1.75,1.875)$.  We follow this procedure until we gain
sufficient precision.  In fact, the root is at $c \approx 1.7549$.
\end{example}

The technique is the simplest method of finding roots of polynomials,
a common problem in applied mathematics.
In general, finding roots is hard to do quickly, precisely,
and automatically.
There are other, faster methods of finding roots of polynomials, such
as Newton's method.  One advantage of the method above is its
simplicity.  The
moment we find an interval where the intermediate value theorem
applies, we are guaranteed to find a root up to a desired
precision in finitely many steps.  Furthermore, the bisection
method finds roots of any
continuous function, not just a polynomial.

The theorem guarantees one $c$ such that $f(c) = y$, but there
may be other roots of the equation $f(c) = y$.  If we follow
the procedure of the proof, we are guaranteed to find approximations to
one such root.  We need to work harder to find any other roots.

\medskip

Polynomials of even degree may not have any real roots.
There is no real number $x$ such that $x^2+1 = 0$.  Odd polynomials, on the
other hand, always have at least one real root.

\begin{prop}
Let $f(x)$ be a polynomial of odd degree.  Then $f$ has a real root.
\end{prop}

\begin{proof}
Suppose $f$ is a polynomial of odd degree $d$.  We write
\begin{equation*}
f(x) = a_d x^d + a_{d-1} x^{d-1} + \cdots + a_1 x + a_0 ,
\end{equation*}
where $a_d \neq 0$.  We divide by $a_d$ to obtain a 
\emph{\myindex{monic polynomial}}\footnote{The word \emph{monic} means that
the coefficient of $x^d$ is 1.}
\begin{equation*}
g(x) \coloneqq x^d + b_{d-1} x^{d-1} + \cdots + b_1 x + b_0 ,
\end{equation*}
where $b_k = \nicefrac{a_k}{a_d}$.
Let us show that $g(n)$ is
positive for some large $n \in \N$.
We first compare the highest order term with the rest:
\begin{equation*}
\begin{split}
\abs{\frac{b_{d-1} n^{d-1} + \cdots + b_1 n + b_0}{n^d}}
& =
\frac{\sabs{b_{d-1} n^{d-1} + \cdots + b_1 n + b_0}}{n^d}
\\
& \leq
\frac{\sabs{b_{d-1}} n^{d-1} + \cdots + \sabs{b_1} n + \sabs{b_0}}{n^d}
\\
& \leq
\frac{\sabs{b_{d-1}} n^{d-1} + \cdots + \sabs{b_1} n^{d-1} + \sabs{b_0} n^{d-1}}{n^d}
\\
& =
\frac{n^{d-1}\bigl(\sabs{b_{d-1}} + \cdots + \sabs{b_1} + \sabs{b_0}\bigr)}{n^d}
\\
& =
\frac{1}{n}
\bigl(\sabs{b_{d-1}} + \cdots + \sabs{b_1} + \sabs{b_0}\bigr) .
\end{split}
\end{equation*}
Therefore,
\begin{equation*}
\lim_{n\to\infty} \frac{b_{d-1} n^{d-1} + \cdots + b_1 n + b_0}{n^d}
= 0 .
\end{equation*}
Thus there exists an $M \in \N$ such that 
\begin{equation*}
\abs{\frac{b_{d-1} M^{d-1} + \cdots + b_1 M + b_0}{M^d}} < 1 ,
\end{equation*}
which implies
\begin{equation*}
-(b_{d-1} M^{d-1} + \cdots + b_1 M + b_0) < M^d .
\end{equation*}
Therefore, $g(M) > 0$.

Next, consider $g(-n)$ for $n \in \N$.  By a similar argument,
there exists a $K \in \N$ such that
$b_{d-1} {(-K)}^{d-1} + \cdots + b_1 (-K) + b_0 < K^d$
and therefore $g(-K) < 0$ (see \exerciseref{exercise:odddegnegativeK}).
In the proof,
 make sure you use the fact that $d$ is odd.
In particular, if $d$ is odd, then ${(-n)}^d = -(n^d)$.

We appeal to the intermediate value theorem to find a
$c \in (-K,M)$ such that $g(c) = 0$.  As $g(x) = \frac{f(x)}{a_d}$,
then $f(c) = 0$, and the proof is done.
\end{proof}

\begin{example}
You may recall how hard we worked
in \exampleref{example:sqrt2}
to show that $\sqrt{2}$ exists.
With Bolzano's
theorem, we can prove the existence of a $k$th root of any positive number
$y > 0$ without any effort.
We claim that for any $k \in \N$ and any $y > 0$,
there exists a number $x > 0$ such that $x^k = y$.

Proof: If $y=1$, then it is clear, so assume $y\neq 1$.
Similarly, assume $k \geq 2$.
Let $f(x) \coloneqq x^k - y$.  We notice $f(0) = -y < 0$.
If $y < 1$, then $f(1) = 1^k -y > 0$.  If $y > 1$,
then $f(y) = y^k-y = y(y^{k-1}-1) > 0$.
In either case, apply Bolzano's theorem to find an $x > 0$ such that $f(x) = 0$,
or in other words $x^k = y$.
\end{example}

\begin{example}
Interestingly,
there exist discontinuous functions with
the intermediate value property.
The function
\begin{equation*}
f(x) \coloneqq
\begin{cases}
\sin(\nicefrac{1}{x}) & \text{if } x \neq 0, \\
0 & \text{if } x=0,
\end{cases}
\end{equation*}
is not continuous at $0$; however, $f$ has the intermediate value property:
Whenever $a < b$ and $y$ is such that $f(a) < y < f(b)$
or $f(a) > y > f(b)$,
there exists a $c \in (a,b)$ such that $f(c) = y$.
See \figureref{figsin1x} for a graph of $\sin(\nicefrac{1}{x})$.
The proof is left as \exerciseref{exercise:meanvaluepropsin1x}.
\end{example}

The intermediate value theorem says that if $f \colon [a,b] \to \R$ is
continuous, then $f\bigl([a,b]\bigr)$ contains all the values between $f(a)$ and
$f(b)$.  In fact, more is true.  Combining all the results of this section
one can prove the following useful corollary whose proof is left as an exercise.
Hint: See \figureref{fig:imageinterval} and notice what the endpoints of the
image interval are.

\begin{cor} \label{cor:imageofinterval}
If $f \colon [a,b] \to \R$ is continuous, then the direct image
$f\bigl([a,b]\bigr)$
is a closed and bounded interval or a single number.
\end{cor}

\begin{myfigureht}
\myincludepdft{figimageinterval}{%
A graph of a function y equals f of x on the interval
from a to b marked in a bold line.  On the vertical
axis, the set of all values achieved by the function
is marked as f of the closed interval from a to b.}
\caption{The image of a continuous $f \colon [a,b] \to \R$.\label{fig:imageinterval}}
\end{myfigureht}

\subsection{Exercises}

\begin{exercise}
Find an example of a discontinuous function $f \colon [0,1] \to \R$
where the conclusion of the intermediate value theorem fails.
\end{exercise}

\begin{exercise}
Find an example of a \emph{bounded} discontinuous function $f \colon [0,1]
\to \R$ that has neither an absolute minimum nor an absolute maximum.
\end{exercise}

\begin{exercise}
Let $f \colon (0,1) \to \R$ be a continuous function such that
$\displaystyle \lim_{x\to 0} f(x) =
\displaystyle \lim_{x\to 1} f(x) = 0$.  Show that
$f$ achieves either an absolute minimum or an absolute maximum on $(0,1)$
(but perhaps not both).
\end{exercise}

\begin{exercise} \label{exercise:meanvaluepropsin1x}
Let
\begin{equation*}
f(x) \coloneqq
\begin{cases}
\sin(\nicefrac{1}{x}) & \text{if } x \neq 0, \\
0 & \text{if } x=0.
\end{cases}
\end{equation*}
Show that $f$ has the intermediate value property.
That is, whenever $a < b$, if there exists a $y$ such that $f(a) < y < f(b)$
or $f(a) > y > f(b)$, then
there exists a $c \in (a,b)$ such that $f(c) = y$.
\end{exercise}

\begin{exercise} \label{exercise:odddegnegativeK}
Suppose $g(x)$ is a monic polynomial of odd degree $d$, that is,
\begin{equation*}
g(x) = x^d + b_{d-1} x^{d-1} + \cdots + b_1 x + b_0 ,
\end{equation*}
for some real numbers $b_{0}, b_1, \ldots, b_{d-1}$.  Show that there exists
a $K \in \N$ such that $g(-K) < 0$.  Hint: Make sure to use the fact that
$d$ is odd.  You will have to use that ${(-n)}^d = -(n^d)$.
\end{exercise}

\begin{exercise}
Suppose $g(x)$ is a monic polynomial of positive even degree $d$, that is,
\begin{equation*}
g(x) = x^d + b_{d-1} x^{d-1} + \cdots + b_1 x + b_0 ,
\end{equation*}
for some real numbers $b_{0}, b_1, \ldots, b_{d-1}$.  Suppose 
$g(0) < 0$.  Show that $g$ has at least two distinct real roots.
\end{exercise}

\begin{exercise}
Prove \corref{cor:imageofinterval}:
Suppose $f \colon [a,b] \to \R$ is a continuous function.  Prove
that the direct image $f\bigl([a,b]\bigr)$ is a closed and bounded interval or
a single number.
\end{exercise}

\begin{exercise}
Suppose $f \colon \R \to \R$ is continuous and periodic with period
$P > 0$.  That is, $f(x+P) = f(x)$ for all $x \in \R$.  Show that $f$
achieves an absolute minimum and an absolute maximum.
\end{exercise}

\begin{exercise}[Challenging]
Suppose $f(x)$ is a bounded polynomial,
in other words, there is an $M$ such that $\babs{f(x)} \leq M$
for all $x \in \R$.  Prove that $f$ must be a constant.
\end{exercise}

\begin{exercise}
Suppose $f \colon [0,1] \to [0,1]$ is continuous.  Show that $f$
has a fixed point, in other words, show that there exists an $x \in [0,1]$ such that
$f(x) = x$.
\end{exercise}

\begin{exercise}
Find an example of a continuous bounded function $f \colon \R \to \R$ that
achieves neither an absolute minimum nor an absolute maximum on $\R$.
\end{exercise}

\begin{exercise}
Suppose $f \colon \R \to \R$ is continuous such that
$x \leq f(x) \leq x+1$ for all $x \in \R$.  Find $f(\R)$.
\end{exercise}

\begin{exercise}
True/False, prove or find a counterexample.  If $f \colon \R \to
\R$ is a continuous function such that $f|_{\Z}$ is bounded, then $f$
is bounded.
\end{exercise}

\begin{exercise}
Suppose $f \colon [0,1] \to (0,1)$ is a bijection.  Prove that $f$ is not
continuous.
\end{exercise}

\begin{exercise}
Suppose $f \colon \R \to \R$ is continuous.
\begin{enumerate}[a)]
\item
Prove that if there is a $c$ such that $f(c)f(-c) < 0$,
then there is a $d \in \R$ such that $f(d) = 0$.
\item
Find a continuous function $f$ such that
$f(\R) = \R$, but $f(x)f(-x) \geq 0$ for all $x \in \R$.
\end{enumerate}
\end{exercise}

\begin{exercise}
Suppose $g(x)$ is a monic polynomial of even degree $d$, that is,
\begin{equation*}
g(x) = x^d + b_{d-1} x^{d-1} + \cdots + b_1 x + b_0 ,
\end{equation*}
for some real numbers $b_{0}, b_1, \ldots, b_{d-1}$.
Show that $g$ achieves an absolute minimum on $\R$.
\end{exercise}

\begin{exercise}
Suppose $f(x)$ is a polynomial of degree $d$ and 
$f(\R) = \R$.  Show that $d$ is odd.
\end{exercise}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

\sectionnewpage
\section{Uniform continuity}
\label{sec:unifcont}

%mbxINTROSUBSECTION

\sectionnotes{1.5--2 lectures (continuous extension can be optional)}

\subsection{Uniform continuity}

We made a fuss of saying that the $\delta$ in the definition of
continuity depended on the point $c$.  There are situations when it is
advantageous to be able to pick a $\delta$ independent of any point, and so we
give a name to this concept.

\begin{defn}
Let $S \subset \R$, and let $f \colon S \to \R$ be a function.
Suppose for every $\epsilon > 0$ there exists a $\delta > 0$
such that whenever $x, c \in S$ and
$\sabs{x-c} < \delta$, then $\babs{f(x)-f(c)} < \epsilon$.
Then we say $f$ is \emph{\myindex{uniformly continuous}}.
\end{defn}

A uniformly continuous function must be continuous.
The only difference in the definitions
is that in uniform continuity,
for a given $\epsilon > 0$ we pick a $\delta > 0$ that
works for all $c \in S$.  That is, $\delta$ can no longer depend on $c$;
it only depends on $\epsilon$.  The domain of definition
of the function makes a difference now.  A function that is not uniformly
continuous on a larger set may be uniformly continuous when restricted to a
smaller set.
Note that $x$ and $c$ are not treated any differently
in this definition.

\begin{example}
$f \colon [0,1] \to \R$ defined by $f(x) \coloneqq x^2$ is uniformly continuous.

Proof: Note that $0 \leq x,c \leq 1$.  Then
\begin{equation*}
\sabs{x^2-c^2} = \sabs{x+c}\sabs{x-c}
\leq \bigl(\sabs{x}+\sabs{c}\bigr) \sabs{x-c}
\leq (1+1)\sabs{x-c} .
\end{equation*}
Therefore, given $\epsilon > 0$, let $\delta \coloneqq \nicefrac{\epsilon}{2}$.
If $\sabs{x-c} < \delta$, then $\sabs{x^2-c^2} \leq 2 \sabs{x-c} < \epsilon$.

\medskip

On the other hand, $g \colon \R \to \R$ defined by $g(x) \coloneqq x^2$ is not uniformly
continuous.

Proof: Suppose it is uniformly continuous.
Then for every $\epsilon > 0$,
there would exist a $\delta > 0$ such that
if $\sabs{x-c} < \delta$, then $\sabs{x^2 -c^2} < \epsilon$.
Take $x > 0$ and let
$c \coloneqq x+\nicefrac{\delta}{2}$.  Write
\begin{equation*}
\epsilon >
\sabs{x^2-c^2} = \sabs{x+c}\sabs{x-c}
=
(2x+\nicefrac{\delta}{2})\nicefrac{\delta}{2} 
\geq 
\delta x .
\end{equation*}
Therefore, $x < \nicefrac{\epsilon}{\delta}$ for all $x > 0$, which is a
contradiction.
\end{example}


\begin{example}
The function $f \colon (0,1) \to \R$ defined by $f(x) \coloneqq \nicefrac{1}{x}$ is not
uniformly continuous.

Proof: Given $\epsilon > 0$, the inequality $\epsilon >
\abs{\nicefrac{1}{x}-\nicefrac{1}{y}}$ holds if and only if
\begin{equation*}
\epsilon >
\abs{\nicefrac{1}{x}-\nicefrac{1}{y}}
=
\frac{\sabs{y-x}}{\sabs{xy}} 
=
\frac{\sabs{y-x}}{xy} ,
\end{equation*}
or
\begin{equation*}
\sabs{x-y} < xy \epsilon .
\end{equation*}
Suppose $\epsilon < 1$.
We wish to see if a 
small $\delta > 0$ would work.
If $x \in (0,1)$ and 
$y = x+\nicefrac{\delta}{2} \in (0,1)$,
then $\sabs{x-y} = \nicefrac{\delta}{2} < \delta$.
We plug $y$ into the inequality above to get $\nicefrac{\delta}{2} <
x \bigl( x+\nicefrac{\delta}{2} \bigr) \epsilon < x$.
If the definition of uniform continuity is satisfied, then
the inequality $\nicefrac{\delta}{2} < x$ would hold for all small $x > 0$.
But that implies $\delta \leq 0$.
Therefore, no single $\delta > 0$ works for all points.
\end{example}

The examples show that if $f$ is defined on an interval that is either not closed
or not bounded, then $f$ can be continuous, but not uniformly continuous.
For a closed and bounded interval $[a,b]$, we can, however,
make the following statement.

\begin{thm} \label{unifcont:thm}
Let $f \colon [a,b] \to \R$ be a continuous function.  Then $f$
is uniformly continuous.
\end{thm}

\begin{proof}
We prove the statement by contrapositive.
Suppose $f$ is not uniformly continuous.  We will prove
that there is some
$c \in [a,b]$ where $f$ is not continuous.  Let us negate
the definition of uniform continuouity:
There exists an $\epsilon > 0$
such that for every $\delta > 0$, there exist points $x, y$ in $[a,b]$ with
$\sabs{x-y} < \delta$ and $\babs{f(x)-f(y)} \geq \epsilon$.

So for the $\epsilon > 0$ above,
we find sequences $\{ x_n \}_{n=1}^\infty$ and $\{ y_n \}_{n=1}^\infty$ such that
$\sabs{x_n-y_n} < \nicefrac{1}{n}$ and such that $\babs{f(x_n)-f(y_n)} \geq
\epsilon$.  By
\hyperref[thm:bwseq]{Bolzano--Weierstrass},
there exists a convergent subsequence
$\{ x_{n_k} \}_{k=1}^\infty$.  Let $c \coloneqq \lim_{k\to\infty} x_{n_k}$.
As $a \leq x_{n_k} \leq b$ for all $k$, we have $a \leq c \leq b$.  Estimate
\begin{equation*}
\sabs{y_{n_k} - c} =
\sabs{y_{n_k} - x_{n_k} + x_{n_k} - c} \leq
\sabs{y_{n_k} - x_{n_k}}
+
\sabs{x_{n_k}-c}
<
\nicefrac{1}{n_k} 
+
\sabs{x_{n_k}-c} .
\end{equation*}
As $\nicefrac{1}{n_k}$ and $\sabs{x_{n_k}-c}$ both go to zero when
$k$ goes to infinity, $\{ y_{n_k} \}_{k=1}^\infty$ converges and the limit
is $c$.  We now show that $f$ is not continuous at $c$.
Estimate
\begin{equation*}
\begin{split}
\babs{f(x_{n_k}) - f(c)} & =
\babs{f(x_{n_k}) - f(y_{n_k}) + f(y_{n_k}) - f(c)} \\
& \geq
\babs{f(x_{n_k}) - f(y_{n_k})} - \babs{f(y_{n_k}) - f(c)} \\
& \geq
\epsilon - \babs{f(y_{n_k})-f(c)} .
\end{split}
\end{equation*}
Or in other words,
\begin{equation*}
\babs{f(x_{n_k})-f(c)} 
+
\babs{f(y_{n_k})-f(c)}  \geq
\epsilon .
\end{equation*}
At least one of the sequences $\bigl\{ f(x_{n_k}) \bigr\}_{k=1}^\infty$  or
$\bigl\{ f(y_{n_k}) \bigr\}_{k=1}^\infty$ cannot converge to $f(c)$.
Otherwise the
left-hand side of the inequality would go to zero while the right-hand side is positive.
Thus $f$ cannot be continuous at $c$.
\end{proof}

As before, note what is key in the proof: We can apply Bolzano--Weierstrass
because the interval $[a,b]$ is bounded, and the limit of the subsequence is
back in $[a,b]$ because the interval is closed.

\subsection{Continuous extension}

Uniformly continuous functions on open intervals extend continuously
to the endpoints.
The key is the following lemma, which also has many other uses.
It says that uniformly continuous functions preserve Cauchy sequences.
The new issue here is that for a Cauchy sequence,
the limit may not end up in the domain of the function.

\begin{lemma} \label{unifcauchycauchy:lemma}
Let $S \subset \R$ and let $f \colon S \to \R$ be a uniformly continuous function.  Let
$\{ x_n \}_{n=1}^\infty$ be a Cauchy sequence in $S$.
Then $\bigl\{ f(x_n) \bigr\}_{n=1}^\infty$ is Cauchy.
\end{lemma}

\begin{proof}
Let $\epsilon > 0$ be given.  There is a $\delta > 0$ such that
$\babs{f(x)-f(y)} < \epsilon$ whenever $x,y \in S$ and $\sabs{x-y} < \delta$.  Find an $M
\in \N$ such that for all $n, k \geq M$, we have $\sabs{x_n-x_k} < \delta$.
Then for all $n, k \geq M$, we have $\babs{f(x_n)-f(x_k)} < \epsilon$.
\end{proof}

An application of the lemma above is the following extension result.  It says that
a function on an open interval is uniformly continuous if and only if
it can be extended to a continuous function on the closed interval.

\begin{prop} \label{context:prop}
A function $f \colon (a,b) \to \R$ is uniformly continuous if and only if
the limits 
\begin{equation*}
L_a \coloneqq \lim_{x \to a} f(x) \qquad \text{and} \qquad
L_b \coloneqq \lim_{x \to b} f(x)
\end{equation*}
exist and the function $\widetilde{f} \colon [a,b] \to \R$
defined by
\begin{equation*}
\widetilde{f}(x) \coloneqq
\begin{cases}
f(x) & \text{if } x \in (a,b), \\
L_a & \text{if } x = a, \\
L_b & \text{if } x = b
\end{cases}
\end{equation*}
is continuous.
\end{prop}

\begin{proof}
One direction is quick.  If $\widetilde{f}$ is continuous, then
it is uniformly continuous by \thmref{unifcont:thm}.  As $f$ is the
restriction of $\widetilde{f}$ to $(a,b)$, $f$ is also uniformly continuous
(exercise).

Now suppose $f$ is uniformly continuous.  We must first show
that the limits $L_a$ and $L_b$ exist.  Let us concentrate on $L_a$.
Take $\{ x_n \}_{n=1}^\infty$ in $(a,b)$ such that
$\lim_{n\to\infty} x_n = a$.
The sequence $\{ x_n \}_{n=1}^\infty$ is Cauchy, so by
\lemmaref{unifcauchycauchy:lemma}
the sequence $\bigl\{ f(x_n) \bigr\}_{n=1}^\infty$ is Cauchy and thus convergent.
Let $L_1 \coloneqq \lim_{n\to\infty} f(x_n)$.  Take another sequence
$\{ y_n \}_{n=1}^\infty$ in $(a,b)$ such that $\lim_{n\to\infty} y_n = a$.
By the same reasoning,
we get $L_2 \coloneqq \lim_{n\to\infty} f(y_n)$.  If we show that $L_1 = L_2$, then
the limit $L_a = \lim_{x\to a} f(x)$ exists.  Let $\epsilon > 0$ be given.
Find $\delta > 0$ such that $\sabs{x-y} < \delta$ implies $\babs{f(x)-f(y)} <
\nicefrac{\epsilon}{3}$.  Find $M \in \N$ such that for
$n \geq M$, we have $\sabs{a-x_n} < \nicefrac{\delta}{2}$,
$\sabs{a-y_n} < \nicefrac{\delta}{2}$,
$\babs{f(x_n)-L_1} < \nicefrac{\epsilon}{3}$, and
$\babs{f(y_n)-L_2} < \nicefrac{\epsilon}{3}$.  Then for $n \geq M$,
\begin{equation*}
\sabs{x_n-y_n} = 
\sabs{x_n-a+a-y_n} \leq
\sabs{x_n-a}+\sabs{a-y_n} < \nicefrac{\delta}{2} + \nicefrac{\delta}{2} =
\delta.
\end{equation*}
So
\begin{equation*}
\begin{split}
\sabs{L_1-L_2} &=
\babs{L_1-f(x_n)+f(x_n)-f(y_n)+f(y_n)-L_2} \\
& \leq 
\babs{L_1-f(x_n)}+\babs{f(x_n)-f(y_n)}+\babs{f(y_n)-L_2} \\
& \leq
\nicefrac{\epsilon}{3} + \nicefrac{\epsilon}{3} + \nicefrac{\epsilon}{3}
=
\epsilon .
\end{split}
\end{equation*}
Therefore, $L_1 = L_2$.
Thus $L_a$ exists.  To show that $L_b$ exists is left as an exercise.

If $L_a = \lim_{x\to a} f(x)$
exists, then $\lim_{x\to a} \widetilde{f}(x)$ exists
and equals $L_a$
(see \propref{prop:limrest}).  Similarly for $L_b$.
Hence, $\widetilde{f}$ is continuous at $a$ and $b$.  
And since $f$ is continuous at $c \in (a,b)$,
we find that
$\widetilde{f}$ is continuous at $c \in (a,b)$ (\propref{prop:limrest} again).
\end{proof}

A typical application of this proposition (together with
\propref{prop:onesidedlimits})
is the following.
If $f \colon (-1,0) \cup (0,1) \to \R$ is uniformly continuous,
then $\lim_{x\to 0} f(x)$ exists and the function
has a \emph{\myindex{removable singularity}}.
That is,
we can extend the function to a continuous function on $(-1,1)$.


\subsection{Lipschitz continuous functions}

\begin{defn}
A function $f \colon S \to \R$
is \emph{\myindex{Lipschitz continuous}}\index{function!Lipschitz}%
\footnote{Named after the German mathematician
\href{https://en.wikipedia.org/wiki/Rudolf_Lipschitz}{Rudolf Otto Sigismund Lipschitz}
(1832--1903).} if there exists a $K \in \R$ such that
\begin{equation*}
\babs{f(x)-f(y)} \leq K \sabs{x-y} 
\qquad \text{for all } x \text{ and } y \text{ in } S.
\end{equation*}
\end{defn}

A large class of functions is Lipschitz continuous.  Be careful. Just as
for uniformly continuous functions, the
domain of definition of the function is important.  See the examples below
and the exercises.  First, we justify the use of the word \emph{continuous}.

\begin{prop}
A Lipschitz continuous function is uniformly continuous.
\end{prop}

\begin{proof}
Let $f \colon S \to \R$ be a function and let $K$ be a constant such that
$\babs{f(x)-f(y)} \leq K \sabs{x-y}$
for all $x, y$ in $S$.
Let $\epsilon > 0$ be given.  Take $\delta \coloneqq
\nicefrac{\epsilon}{K}$.
For all $x$ and $y$ in $S$ such that
$\sabs{x-y} < \delta$,
\begin{equation*}
\babs{f(x)-f(y)} \leq K \sabs{x-y} < K \delta = K \frac{\epsilon}{K} =
\epsilon .
\end{equation*}
Therefore, $f$ is uniformly continuous.
\end{proof}

We interpret Lipschitz continuity geometrically.  Let $f$ be a Lipschitz
continuous function with some constant $K$.  We rewrite the inequality 
to say that for $x \neq y$, we have
\begin{equation*}
\abs{\frac{f(x)-f(y)}{x-y}} \leq K .
\end{equation*}
The quantity $\frac{f(x)-f(y)}{x-y}$ is the slope of the line
between the points $\bigl(x,f(x)\bigr)$
and $\bigl(y,f(y)\bigr)$, that is, a \emph{\myindex{secant line}}.  Therefore, $f$ is Lipschitz
continuous if and only if every line that intersects the graph of $f$ in at least two
distinct
points has slope in absolute value less than or equal to $K$.  See \figureref{fig:lipschitz}.
\begin{myfigureht}
\myincludegraphics{lipschitzfig}{%
A graph of a function with some corners.  Two
points on the horizontal axis are marked with x
and y.  A line through the corresponding points
on the graph is drawn and its slope is labeled
as f of x minus f of y all divided by x minus y.}
\caption{The slope of a secant line.
A function is Lipschitz if %$\abs{\text{slope}} =
$\abs{\frac{f(x)-f(y)}{x-y}} \leq K$ for all $x$ and $y$.\label{fig:lipschitz}}
\end{myfigureht}

\begin{example}
The functions $\sin(x)$ and $\cos(x)$ are Lipschitz continuous.
In \exampleref{sincos:example} we have seen the following two inequalities.
\begin{equation*}
\babs{\sin(x)-\sin(y)} 
\leq \sabs{x-y}
\qquad \text{and} \qquad
\babs{\cos(x)-\cos(y)}
\leq \sabs{x-y} .
\end{equation*}

Hence, sine and cosine are Lipschitz continuous with $K=1$.
\end{example}

\begin{example}
The function $f \colon [1,\infty) \to \R$ defined by $f(x) \coloneqq \sqrt{x}$
is Lipschitz continuous. Proof:
\begin{equation*}
\abs{\sqrt{x}-\sqrt{y}} = 
\abs{\frac{x-y}{\sqrt{x}+\sqrt{y}}}
=
\frac{\sabs{x-y}}{\sqrt{x}+\sqrt{y}} .
\end{equation*}
As $x \geq 1$ and $y \geq 1$, we see that $\frac{1}{\sqrt{x}+\sqrt{y}}
\leq \frac{1}{2}$.  Therefore,
\begin{equation*}
\abs{\sqrt{x}-\sqrt{y}} = 
\abs{\frac{x-y}{\sqrt{x}+\sqrt{y}}}
\leq
\frac{1}{2}
\sabs{x-y}.
\end{equation*}

On the other hand, $g \colon [0,\infty) \to \R$ defined by
$g(x) \coloneqq \sqrt{x}$ is not Lipschitz continuous.  Proof:
Suppose there is a $K$ so that for all $x,y \in [0,\infty)$,
\begin{equation*}
\abs{\sqrt{x}-\sqrt{y}} 
\leq
K \sabs{x-y} ,
\end{equation*}
We find that $K > 0$, as $K < 0$ is impossible,
and $K=0$ would mean that $f$ is constant.
Set $y=0$ to obtain
$\sqrt{x} \leq K x$.
For $x > 0$, we get
$\nicefrac{1}{K} \leq \sqrt{x}$ or $\nicefrac{1}{K^2} \leq x$.
This cannot possibly be true for all
$x > 0$.  Thus no $K$ exists and $g$ is not
Lipschitz continuous.  See \figureref{fig:sqrtgraph} and note how secant
lines would be more and more vertical as we get closer to $x=0$.
\begin{myfigureht}
\myincludegraphics{sqrtgraph}{%
A graph of a function that starts rising vertically
at the left endpoint and quickly curves to become
much more horizontal at the right endpoint.
Secant lines through the left endpoint and a
sequence of points on the graph getting closer
and closer to the left endpoint are drawn.
We notice that the secant lines are more
and more vertical as the second
point on the secant line is closer
to the left endpoint.}
\caption{Graph of $\sqrt{x}$ and some secant lines through $(0,0)$ and
$(x,\sqrt{x})$.\label{fig:sqrtgraph}}
\end{myfigureht}

The last example $g$ is a function that is uniformly
continuous but not Lipschitz continuous.
To see that $\sqrt{x}$ is
uniformly continuous as a function on $[0,\infty)$,
note that it is uniformly continuous when restricted to $[0,1]$ by \thmref{unifcont:thm}.
It is also Lipschitz (and so uniformly continuous) when restricted to $[1,\infty)$.
It is not hard (exercise) to show that this means that $\sqrt{x}$ is a
uniformly continuous function on $[0,\infty)$.
\end{example}

\subsection{Exercises}

\begin{exercise}
Let $f \colon S \to \R$ be uniformly continuous.  Let $A \subset S$.
Then the restriction $f|_A$ is uniformly continuous.
\end{exercise}

\begin{exercise}
Let $f \colon (a,b) \to \R$ be a uniformly continuous function.
Finish the proof of \propref{context:prop} by showing that
the limit
$\lim\limits_{x \to b} f(x)$
exists.
\end{exercise}

\begin{exercise}
Show that $f \colon (c,\infty) \to \R$ for some $c > 0$
and defined by $f(x) \coloneqq \nicefrac{1}{x}$ is Lipschitz continuous.
\end{exercise}

\begin{exercise}
Show that $f \colon (0,\infty) \to \R$
defined by $f(x) \coloneqq \nicefrac{1}{x}$ is not Lipschitz continuous.
\end{exercise}

\begin{samepage}
\begin{exercise}
Let $A, B$ be intervals.
Let $f \colon A \to \R$ and $g \colon B \to \R$ be uniformly continuous
functions such that $f(x) = g(x)$ for $x \in A \cap B$.  Define
the function $h \colon A \cup B \to \R$ by $h(x) \coloneqq f(x)$ if
$x \in A$ and $h(x) \coloneqq g(x)$ if $x \in B \setminus A$.
\begin{enumerate}[a)]
\item
Prove that if $A \cap B \neq \emptyset$, then $h$ is uniformly continuous.
\item
Find an example where $A \cap B = \emptyset$ and $h$ is not even
continuous.
\end{enumerate}
\end{exercise}
\end{samepage}

\begin{exercise}[Challenging]
Let $f \colon \R \to \R$ be a polynomial of degree 
$d \geq 2$.  Show that $f$ is not Lipschitz
continuous.
\end{exercise}

\begin{exercise}
Let $f \colon (0,1) \to \R$ be a bounded continuous function.  Show that
the function
$g(x) \coloneqq x(1-x)f(x)$ is uniformly continuous.
\end{exercise}

\begin{exercise}
Show that $f \colon (0,\infty) \to \R$ defined by $f(x) \coloneqq \sin
(\nicefrac{1}{x})$ is not uniformly continuous.
\end{exercise}

\begin{exercise}[Challenging]
Let $f \colon \Q \to \R$ be a uniformly continuous function.  Show that
there exists a uniformly continuous function $\widetilde{f} \colon \R \to \R$
such that $f(x) = \widetilde{f}(x)$ for all $x \in \Q$.
\end{exercise}

\begin{samepage}
\begin{exercise}
\leavevmode
\begin{enumerate}[a)]
\item
Find a continuous $f \colon (0,1) \to \R$ and a sequence
$\{ x_n \}_{n=1}^\infty$ in
$(0,1)$ that is Cauchy, but such that $\bigl\{ f(x_n) \bigr\}_{n=1}^\infty$ is not Cauchy.
\item
Prove that if $f \colon \R \to \R$ is continuous,
and $\{ x_n \}_{n=1}^\infty$ is
Cauchy, then $\bigl\{ f(x_n) \bigr\}_{n=1}^\infty$ is Cauchy.
\end{enumerate}
\end{exercise}
\end{samepage}

\begin{samepage}
\begin{exercise}
Prove:
\begin{enumerate}[a)]
\item
If $f \colon S \to \R$ and $g \colon S \to \R$ are uniformly continuous,
then $h \colon S \to \R$ given by $h(x) \coloneqq f(x) + g(x)$
is uniformly continuous.
\item
If $f \colon S \to \R$ is uniformly continuous and $a \in \R$,
then $h \colon S \to \R$ given by $h(x) \coloneqq a f(x)$
is uniformly continuous.
\end{enumerate}
\end{exercise}
\end{samepage}

\begin{exercise}
Prove:
\begin{enumerate}[a)]
\item
If $f \colon S \to \R$ and $g \colon S \to \R$ are Lipschitz,
then $h \colon S \to \R$ given by $h(x) \coloneqq f(x) + g(x)$
is Lipschitz.
\item
If $f \colon S \to \R$ is Lipschitz and $a \in \R$,
then $h \colon S \to \R$ given by $h(x) \coloneqq a f(x)$
is Lipschitz.
\end{enumerate}
\end{exercise}

\begin{exercise}
\pagebreak[2]
\leavevmode
\begin{enumerate}[a)]
\item
If $f \colon [0,1] \to \R$ is given by $f(x) \coloneqq x^m$ for an integer
$m \geq 0$,
show $f$ is Lipschitz and find the best (the smallest) Lipschitz constant
$K$ (depending on $m$ of course).
Hint:
$(x-y)(x^{m-1} + x^{m-2}y + x^{m-3}y^2 + \cdots + x y^{m-2} + y^{m-1}) = x^m
- y^m$ for positive $m$.
\item
Using the previous exercise, show that if $f \colon [0,1] \to \R$
is a polynomial, that is, $f(x) \coloneqq a_m x^m + a_{m-1} x^{m-1} + \cdots + a_0$,
then $f$ is Lipschitz.
\end{enumerate}
\end{exercise}

\begin{exercise}
\pagebreak[2]
Suppose for $f \colon [0,1] \to \R$, we have $\babs{f(x)-f(y)} \leq K
\sabs{x-y}$ for all $x,y$ in $[0,1]$,
and $f(0) = f(1) = 0$.
Prove that $\babs{f(x)} \leq \nicefrac{K}{2}$ for all $x \in [0,1]$.  Further show by example that
$\nicefrac{K}{2}$ is the best possible.
That is, fund such a continuous function
for which $\babs{f(x)} = \nicefrac{K}{2}$ for some $x \in [0,1]$.
\end{exercise}

\begin{exercise}
Suppose $f \colon \R \to \R$ is continuous and periodic with period
$P > 0$.  That is, $f(x+P) = f(x)$ for all $x \in \R$.  Show that $f$
is uniformly continuous.
\end{exercise}

\begin{exercise}
Suppose $f \colon S \to \R$ and $g \colon [0,\infty) \to [0,\infty)$
are functions, $g$ is continuous at $0$, $g(0) = 0$, and
whenever $x$ and $y$ are in $S$, we have $\babs{f(x)-f(y)} \leq
g\bigl(\sabs{x-y}\bigr)$.
Prove that $f$ is uniformly continuous.
\end{exercise}

\begin{exercise}
Suppose $f \colon [a,b] \to \R$ is a function such that for every $c \in
[a,b]$ there is a $K_c > 0$ and an $\epsilon_c > 0$ for which
$\babs{f(x)-f(y)} \leq K_c \sabs{x-y}$ for all $x$ and $y$ in
$(c-\epsilon_c,c+\epsilon_c) \cap [a,b]$.  In other words, $f$ is
\myquote{locally Lipschitz.}
\begin{enumerate}[a)]
\item
Prove that there exists a single $K > 0$ such that
$\babs{f(x)-f(y)} \leq K \sabs{x-y}$ for all $x,y$ in $[a,b]$.
\item
Find a counterexample to the above if the interval is open, that is,
find an $f \colon (a,b) \to \R$ that is locally Lipschitz, but not
Lipschitz.
\end{enumerate}
\end{exercise}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

\sectionnewpage
\section{Limits at infinity}
\label{sec:limitatinf}

%mbxINTROSUBSECTION

\sectionnotes{less than 1 lecture (optional, can safely be omitted unless
\sectionref{sec:monotonefunc} or
\sectionref{sec:impropriemann} is also covered)}

\subsection{Limits at infinity}

Just as with sequences, a continuous variable can also approach infinity.

\begin{defn}
We say $\infty$ is a \emph{\myindex{cluster point}} of $S \subset \R$ if for every
$M \in \R$, there exists an $x \in S$ such that $x \geq M$.  Similarly,
$- \infty$ is a \emph{cluster point} of $S \subset \R$ if for every
$M \in \R$, there exists an $x \in S$ such that $x \leq M$.

\index{limit!of a function at infinity}%
Let $f \colon S \to \R$ be a function, where 
$\infty$ is a cluster point of $S$.
If there exists an $L \in \R$
with the property
that for every $\epsilon > 0$, there is an $M \in \R$ such that
\begin{equation*}
\babs{f(x) - L} < \epsilon 
\end{equation*}
whenever $x \in S$ and $x \geq M$, then we say $f(x)$
\emph{converges}\index{converges!function} to $L$
as $x$ goes to $\infty$.  We call $L$ a \emph{limit}
and, if unique, write
\glsadd{not:limfuncinf}
\begin{equation*}
\lim_{x \to \infty} f(x) \coloneqq L .
\end{equation*}
Alternatively, we write $f(x) \to L$ as $x \to \infty$.

Similarly, if $-\infty$ is a cluster point of $S$
and
there exists an $L \in \R$
such that for every $\epsilon > 0$, there is an $M \in \R$ such that
\begin{equation*}
\babs{f(x) - L} < \epsilon 
\end{equation*}
whenever $x \in S$ and $x \leq M$, then we say $f(x)$ \emph{converges} to $L$
as $x$ goes to $-\infty$.
Alternatively, we write $f(x) \to L$ as $x \to -\infty$.
We call $L$ a \emph{limit} and, if unique, write
\begin{equation*}
\lim_{x \to -\infty} f(x) \coloneqq L .
\end{equation*}
\end{defn}

The first thing to do, as usual, is to prove that the limit, if it exists,
is unique.
We leave it as an exercise for the reader.

\begin{prop} \label{liminfty:unique}
The limit at $\infty$ or $-\infty$ as defined above is unique if it exists.
\end{prop}

\begin{example}
Let $f(x) \coloneqq \frac{1}{\sabs{x}+1}$.  Then
\begin{equation*}
\lim_{x\to \infty} f(x) = 0 \qquad \text{and} \qquad
\lim_{x\to -\infty} f(x) = 0 .
\end{equation*}

Proof:
Let $\epsilon > 0$ be given.  Find $M > 0$ large enough
so that $\frac{1}{M+1} < \epsilon$.  If
$x \geq M$, then $0 < \frac{1}{\sabs{x}+1} = \frac{1}{x+1} \leq \frac{1}{M+1} < \epsilon$.
The first limit follows.
The proof for $-\infty$ is left to the reader.
\end{example}

\begin{example}
Let $f(x) \coloneqq \sin(\pi x)$.  Then $\lim_{x\to\infty} f(x)$ does not exist.
To prove this fact, note that if $x = 2n+\nicefrac{1}{2}$ for some $n \in
\N$, then $f(x)=1$,
while if $x = 2n+\nicefrac{3}{2}$, then $f(x)=-1$.
Both of these values, $1$ and $-1$, cannot be
within a small $\epsilon$ of a single real number.

Be careful not to confuse continuous limits with limits of sequences.
We could say
\begin{equation*}
\lim_{n \to \infty} \sin(\pi n) = 0, \qquad \text{but} \qquad
\lim_{x \to \infty} \sin(\pi x) \enspace \text{does not exist}.
\end{equation*}
Of course, the notation is ambiguous:  Are we thinking of the
sequence $\bigl\{ \sin (\pi n) \bigr\}_{n=1}^\infty$ or the function $\sin(\pi x)$
of a real variable?  We are simply using the convention
that $n \in \N$, while $x \in \R$.  When the notation is not clear,
it is good to explicitly mention where the variable lives, or what kind
of limit you are using.  If there is a possibility of confusion, one can
write, for example,
\begin{equation*}
\lim_{\substack{n \to \infty\\n \in \N}} \sin(\pi n) .
\end{equation*}
\end{example}

There is a connection between continuous limits and limits of sequences, but we must take all
sequences going to infinity, just as before in \lemmaref{seqflimit:lemma}.

\begin{lemma} \label{seqflimitinf:lemma}
Suppose $f \colon S \to \R$ is a function, $\infty$ is a cluster
point of $S \subset \R$, and $L \in \R$.  Then
\begin{equation*}
\lim_{x\to\infty} f(x) = L
\qquad \text{if and only if} \qquad
\lim_{n\to\infty} f(x_n) = L
\end{equation*}
for all sequences $\{ x_n \}_{n=1}^\infty$ in $S$ such that $\lim\limits_{n\to\infty} x_n = \infty$.
\end{lemma}

The lemma also holds for the limit as $x \to -\infty$.
Its proof is almost identical and
is left as an exercise.

\begin{proof}
First suppose $f(x) \to L$ as $x \to \infty$.
Given an $\epsilon > 0$, there exists an $M$ such that for all $x \in S$
where $x \geq M$,
we have $\babs{f(x)-L} < \epsilon$.
Let $\{ x_n \}_{n=1}^\infty$
be a sequence in $S$ such that $\lim_{n\to\infty} x_n = \infty$.  Then there exists an
$N$ such that
$x_n \geq M$
for all $n \geq N$.
Thus
$\babs{f(x_n)-L} < \epsilon$.

We prove the converse by contrapositive.  Suppose $f(x)$ does
not go to $L$ as $x \to \infty$.
This means that there exists an $\epsilon > 0$
such that for every $n \in \N$, there exists an $x \in S$, $x \geq n$, let
us call it $x_n$, such that $\babs{f(x_n)-L} \geq \epsilon$.
Consider the sequence $\{ x_n \}_{n=1}^\infty$.  Clearly 
$\bigl\{ f(x_n) \bigr\}_{n=1}^\infty$ does not converge to $L$.  It remains to note
that $\lim_{n\to\infty} x_n = \infty$ because $x_n \geq n$ for all $n$.
\end{proof}

Using the lemma, we again translate results about sequential
limits into results about continuous limits as $x$ goes to infinity.  That
is, we have almost immediate analogues of the corollaries
in \sectionref{subseq:sequentiallimits}.  We simply allow 
the cluster point $c$ to be either $\infty$ or $-\infty$, in addition
to a real number.  We leave it to
the student to verify these statements.

\subsection{Infinite limit}

Just as for sequences, it is often convenient to distinguish certain
divergent sequences, and talk about limits being infinite
almost as if the limits existed.

\begin{defn}
\index{infinite limit!of a function}\index{limit!infinite}%
Let $f \colon S \to \R$ be a function and suppose 
$S$ has $\infty$ as a cluster point.
We say $f(x)$
\emph{\myindex{diverges to infinity}} 
as $x$ goes to $\infty$
if for every $N \in \R$
there exists an $M \in \R$ such that
\begin{equation*}
f(x) > N
\end{equation*}
whenever $x \in S$ and $x \geq M$.
We write
\begin{equation*}
\lim_{x \to \infty} f(x) \coloneqq \infty ,
\end{equation*}
or we say that $f(x) \to \infty$ as $x \to \infty$.
\end{defn}

A similar definition can be made for limits as $x \to -\infty$
or as $x \to c$ for a finite $c$.  Also, similar definitions can be
made for limits being $-\infty$.  Stating these definitions is left
as an exercise.
Note that
sometimes the slightly incorrect (in this context)
wording
\emph{\myindex{converges to infinity}} is used.
We can again use sequential limits, and an analogue of 
\lemmaref{seqflimit:lemma} is left as an exercise.

\begin{example}
Let us show that $\lim\limits_{x \to \infty} \frac{1+x^2}{1+x} = \infty$.

Proof: For $x \geq 1$, we have
\begin{equation*}
\frac{1+x^2}{1+x} \geq 
\frac{x^2}{x+x}  = 
\frac{x}{2} .
\end{equation*}
Given $N \in \R$, take $M = \max \{ 2N+1 , 1 \}$.
If $x \geq M$, then $x \geq 1$ and $\nicefrac{x}{2} > N$.
So
\begin{equation*}
\frac{1+x^2}{1+x} \geq 
\frac{x}{2} > N .
\end{equation*}
\end{example}

\subsection{Compositions}

Finally, just as for limits at finite numbers,
limits play nice with compositions given a
continuity assumption.

\begin{prop} \label{prop:inflimcompositions}
Suppose $f \colon A \to B$, $g \colon B \to \R$, $A, B \subset \R$, 
$a \in \R \cup \{ -\infty, \infty\}$ is a cluster point of $A$,
and $b \in \R \cup \{ -\infty, \infty\}$ is a cluster point of $B$.
Suppose 
\begin{equation*}
\lim_{x \to a} f(x) = b\qquad \text{and} \qquad \lim_{y \to b} g(y) = c
\end{equation*}
for some $c \in \R \cup \{ -\infty, \infty \}$.
If $b \in B$, then suppose $g(b) = c$.
Then
\begin{equation*}
\lim_{x \to a} g\bigl(f(x)\bigr) = c .
\end{equation*}
\end{prop}

The proof is straightforward, and left as an exercise.  We already
know the proposition when $a, b, c \in \R$ (see Exercises
\ref{exercise:contlimitcomposition} and
\ref{exercise:contlimitbadcomposition}).  Again, the requirement that $g$ is
continuous at $b$, if $b \in B$, is necessary.

\begin{example}
Let $h(x) \coloneqq e^{-x^2+x}$.  Then
\begin{equation*}
\lim_{x\to \infty} h(x) = 0 .
\end{equation*}

Proof:
The claim follows once we know
\begin{equation*}
\lim_{x\to \infty} (-x^2+x) = -\infty
\end{equation*}
and
\begin{equation*}
\lim_{y\to -\infty} e^y = 0 ,
\end{equation*}
which is usually proved when the exponential function is defined.
\end{example}

\subsection{Exercises}

\begin{exercise}
Prove \propref{liminfty:unique}.
\end{exercise}

\begin{exercise}
Let $f \colon [1,\infty) \to \R$ be a function.  Define
$g \colon (0,1] \to \R$ via $g(x) \coloneqq f(\nicefrac{1}{x})$.
Using the definitions of limits directly,
show that $\lim_{x\to 0^+} g(x)$
exists if and only if $\lim_{x\to \infty} f(x)$ exists, in which
case they are equal.
\end{exercise}

\begin{exercise}
Prove \propref{prop:inflimcompositions}.
\end{exercise}

\begin{exercise}
Let us justify terminology.
Let $f \colon \R \to \R$ be a function such that
$\lim_{x \to \infty} f(x) = \infty$ (diverges to infinity).
Show that $f(x)$ diverges (i.e.\ does not converge) as $x \to \infty$.
\end{exercise}

\begin{exercise}
Come up with the definitions for limits of $f(x)$ going to $-\infty$ as $x \to
\infty$, $x \to -\infty$, and as $x \to c$ for a finite $c \in \R$.
Then state the definitions for limits of $f(x)$ going to $\infty$ 
as $x \to -\infty$, and as $x \to c$ for a finite $c \in \R$.
\end{exercise}

\begin{exercise}
Suppose $P(x) \coloneqq x^n + a_{n-1} x^{n-1} + \cdots + a_1 x + a_0$ is a \emph{\myindex{monic polynomial}}
of degree $n \geq 1$ (monic means that the coefficient of $x^n$ is 1).
\begin{enumerate}[a)]
\item
Show that if $n$ is even, then $\lim\limits_{x\to\infty} P(x) = 
\lim\limits_{x\to-\infty} P(x) = \infty$.
\item
Show that if $n$ is odd, then
$\lim\limits_{x\to\infty} P(x) = \infty$ and
$\lim\limits_{x\to-\infty} P(x) = -\infty$ (see previous exercise).
\end{enumerate}
\end{exercise}

\begin{exercise}
Let $\{ x_n \}_{n=1}^\infty$ be a sequence.  Consider $S \coloneqq \N \subset \R$, and
$f \colon S \to \R$ defined by $f(n) \coloneqq x_n$.  Show that
the two notions of limit,
\begin{equation*}
\lim_{n\to\infty} x_n \qquad \text{and} \qquad
\lim_{x\to\infty} f(x) 
\end{equation*}
are equivalent.  That is, show that if one exists so does
the other one, and in this case they are equal.
\end{exercise}

\begin{exercise}
Extend \lemmaref{seqflimitinf:lemma} as follows.
Suppose $S \subset \R$ has a cluster point $c \in \R$, $c = \infty$,
or $c = -\infty$.  Let $f \colon S \to \R$ be a function and suppose
$L = \infty$ or $L = -\infty$.  Show that
\begin{equation*}
\begin{aligned}
&
\lim_{x\to c} f(x) = L
\\
&
\qquad \text{if and only if}
\\
&
\lim_{n\to\infty} f(x_n) = L \enspace \text{for all sequences }
\{ x_n \}_{n=1}^\infty \text{ such that } \lim_{n\to\infty} x_n = c
\text{ and } x_n \in S \setminus \{c\}
\text{ for all } n  .
\end{aligned}
\end{equation*}
\end{exercise}

\begin{exercise}
Suppose $f \colon \R \to \R$ is a 2-periodic function, that is $f(x + 2) =
f(x)$ for all $x$.  Define $g \colon \R \to \R$ by 
$g(0) \coloneqq f(0)$ and for $x \neq 0$ by
\begin{equation*}
g(x) \coloneqq f\left(\frac{\sqrt{x^2+1}-1}{x}\right)
\end{equation*}
\begin{enumerate}[a)]
\item
Find the function $\varphi \colon (-1,1) \to \R$ such that
$g\bigl(\varphi(t)\bigr) = f(t)$, that is, $\varphi^{-1}(x) = 
\frac{\sqrt{x^2+1}-1}{x}$ for $x \neq 0$.
\item
Show that $f$ is continuous if and only if $g$ is continuous and
\begin{equation*}
\lim_{x \to \infty} g(x) = 
\lim_{x \to -\infty} g(x) = 
f(1) = f(-1) .
\end{equation*}
\end{enumerate}
\end{exercise}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

\sectionnewpage
\section{Monotone functions and continuity}
\label{sec:monotonefunc}

%mbxINTROSUBSECTION

\sectionnotes{1 lecture (optional, can safely be omitted unless
\sectionref{sec:ift} is also covered, requires \sectionref{sec:limitatinf})}

\begin{defn}
Let $S \subset \R$.
We say $f \colon S \to \R$ is \emph{\myindex{increasing}}
(resp.\  \emph{\myindex{strictly increasing}}) if $x,y \in S$ with
$x < y$ implies $f(x) \leq f(y)$ (resp.\ $f(x) < f(y)$).
We define
\emph{\myindex{decreasing}} and
\emph{\myindex{strictly decreasing}} in the same way by switching the
inequalities for $f$.

If a function is either increasing or decreasing, we say it is
\emph{monotone}\index{monotone function}.  If it is
strictly increasing or strictly decreasing, we say it is
\emph{strictly monotone}\index{strictly monotone function}.
\end{defn}

Sometimes \emph{\myindex{nondecreasing}}
(resp.\ \emph{\myindex{nonincreasing}}) is used
for increasing (resp.\ decreasing) function to emphasize it is not
strictly increasing (resp.\ strictly decreasing).

If $f$ is increasing, then $-f$ is decreasing and vice versa.  Therefore,
many results about monotone functions can just be proved for, say, increasing
functions, and the results follow easily for decreasing functions.

\subsection{Continuity of monotone functions}

One-sided limits for monotone functions are computed
by computing infima and suprema.

\begin{prop} \label{prop:monotlimits}
Let $S \subset \R$, $c \in \R$,
$f \colon S \to \R$ be increasing,
and
$g \colon S \to \R$ be decreasing.
If $c$ is a cluster point of $S \cap (-\infty,c)$, then
\begin{equation*}
\lim_{x \to c^-} f(x) = \sup \{ f(x) : x < c, x \in S \}
%mbxSTARTIGNORE
\qquad \text{and} \qquad
%mbxENDIGNORE
%mbxlatex \quad \text{and} \quad
\lim_{x \to c^-} g(x) = \inf \{ g(x) : x < c, x \in S \} .
\end{equation*}
If $c$ is a cluster point of $S \cap (c,\infty)$, then
\begin{equation*}
\lim_{x \to c^+} f(x) = \inf \{ f(x) : x > c, x \in S \}
%mbxSTARTIGNORE
\qquad \text{and} \qquad
%mbxENDIGNORE
%mbxlatex \quad \text{and} \quad
\lim_{x \to c^+} g(x) = \sup \{ g(x) : x > c, x \in S \} .
\end{equation*}
If $\infty$ is a cluster point of $S$, then
\begin{equation*}
\lim_{x \to \infty} f(x) = \sup \{ f(x) : x \in S \}
%mbxSTARTIGNORE
\qquad \text{and} \qquad
%mbxENDIGNORE
%mbxlatex \quad \text{and} \quad
\lim_{x \to \infty} g(x) = \inf \{ g(x) : x \in S \} .
\end{equation*}
If $-\infty$ is a cluster point of $S$, then
\begin{equation*}
\lim_{x \to -\infty} f(x) = \inf \{ f(x) : x \in S \}
%mbxSTARTIGNORE
\qquad \text{and} \qquad
%mbxENDIGNORE
%mbxlatex \quad \text{and} \quad
\lim_{x \to -\infty} g(x) = \sup \{ g(x) : x \in S \} .
\end{equation*}
\end{prop}

Namely, all the one-sided limits exist whenever they make
sense.  Therefore, for monotone functions, when we say
the left-hand limit $x \to c^-$
exists, we mean that $c$ is a cluster point of $S \cap (-\infty,c)$,
and same for the right-hand limit.

\begin{proof}
Let us assume $f$ is increasing, and we will show the first
equality.  The rest of the proof is very similar and is left as an
exercise.

Let $a \coloneqq \sup \{ f(x) : x < c, x \in S \}$.  If $a = \infty$,
then given an $M \in \R$, there exists an $x_M \in S$, $x_M < c$, such that $f(x_M) > M$. 
As $f$ is increasing, $f(x) \geq f(x_M) >  M$ for all $x \in S$ with $x > x_M$.
Take $\delta \coloneqq c-x_M > 0$ to satisfy the definition of the limit
being infinity.

Next suppose $a < \infty$.
Let $\epsilon > 0$ be given.  Because $a$ is the supremum and
$S \cap (-\infty,c)$ is nonempty, $a \in \R$ and
there exists an
$x_\epsilon \in S$,
$x_\epsilon < c$,
such that $f(x_\epsilon) > a-\epsilon$.  As $f$ is increasing,
if $x \in S$ and $x_\epsilon < x < c$, we have
$a-\epsilon < f(x_\epsilon) \leq f(x) \leq a$.  Let
$\delta \coloneqq c-x_\epsilon$.  Then for $x \in S \cap (-\infty,c)$
with $\sabs{x-c} < \delta$,
we have $\babs{f(x)-a} < \epsilon$.
\end{proof}

Suppose $f \colon S \to \R$ is increasing, $c \in S$, and
that both one-sided limits exist.
Since $f(x) \leq f(c) \leq f(y)$
whenever $x < c < y$, taking the limits we obtain
\begin{equation*}
\lim_{x \to c^-} f(x) \leq f(c) \leq \lim_{x \to c^+} f(x) .
\end{equation*}
Then $f$ is continuous at $c$ if and only if both limits are equal
to each other (and hence equal to $f(c)$).  See also
\propref{prop:onesidedlimits}.
See \figureref{fig:figinccont} to get an idea of what a discontinuity
looks like.


\begin{cor} \label{cor:continterval}
If $I \subset \R$ is an interval and $f \colon I \to \R$ is 
monotone and not constant, then $f(I)$ is an interval if and only if $f$
is continuous.
\end{cor}

Assuming $f$ is not constant is to avoid the technicality
that $f(I)$ is a single point: $f(I)$ is a single
point if and only if $f$ is constant.  A constant function is 
continuous.

\begin{proof}
Without loss of generality, suppose $f$ is increasing.

First suppose $f$ is continuous.  Take two points
$f(x_1), f(x_2)$ in $f(I)$ and suppose
$f(x_1) < f(x_2)$.
As $f$ is increasing, $x_1 < x_2$.  By the
\hyperref[IVT:thm]{intermediate value theorem},
given $y$ with $f(x_1) < y < f(x_2)$, we find
a $c \in (x_1,x_2) \subset I$ such that $f(c) = y$, so $y \in f(I)$. 
Hence, $f(I)$ is an interval.

Let us prove the reverse direction by contrapositive.
Suppose $f$ is not continuous at $c \in I$,
and that $c$ is not an endpoint of $I$.
Let
\begin{equation*}
%mbxlatex \begin{aligned}
a
%mbxlatex &
\coloneqq \lim_{x \to c^-} f(x) = \sup \bigl\{ f(x) : x \in I, x < c \bigr\} ,
%mbxSTARTIGNORE
\qquad
%mbxENDIGNORE
%mbxlatex \\
b
%mbxlatex &
\coloneqq \lim_{x \to c^+} f(x) = \inf \bigl\{ f(x) : x \in I, x > c \bigr\} .
%mbxlatex \end{aligned}
\end{equation*}
As $c$ is a discontinuity, $a < b$.
If $x < c$, then $f(x) \leq a$, and
if $x > c$, then $f(x) \geq b$.  Therefore,
no point
in $(a,b) \setminus \bigl\{ f(c) \bigr\}$ is in $f(I)$.
There exists $x_1 \in I$ with $x_1 < c$, so
$f(x_1) \leq a$, and there exists $x_2 \in I$ with $x_2 > c$,
so $f(x_2) \geq b$.  Both $f(x_1)$ and $f(x_2)$ are in $f(I)$,
but there are points in between them that are not in $f(I)$.
So $f(I)$ is not an interval.  See \figureref{fig:figinccont}.

When $c \in I$ is an endpoint, the proof is similar and is left as an exercise.
\end{proof}

\begin{myfigureht}
\myincludepdft{figinccont}{%
A diagram of a graph of a function with a jump discontinuity and the effect
on the image of the function.
The function is defined on an interval marked I.  Points x sub 1, c, and
x sub 2 are marked inside I and come in that order from left to right.
The function increases continuously until x equals c,
where the function approaches a, that is, the limit of f of x as x approaches
c from below is a.  The value of f at c is f of c, which is larger than a
and smaller than b, where b is the limit of f of x as x approaches c from
above.  Then the function keeps rising continuously until the end of the
interval I.  The image of f is split in three pieces.
The lowest piece is
an interval whose upper bound is a, and this interval contains
f of x sub 1.
The next piece is simply the point f of c.
The third, the highest, piece is an interval starting at b
and this interval contains f of x sub 2.}
\caption{Increasing function $f \colon I \to \R$ discontinuity at
$c$.\label{fig:figinccont}}
\end{myfigureht}

A striking property of monotone functions is that they cannot have
too many discontinuities.

\begin{cor} \label{cor:monotcountcont}
Let $I \subset \R$ be an interval and
$f \colon I \to \R$ be monotone.  Then $f$ has at most
countably many discontinuities.
\end{cor}

\begin{proof}
Let $E \subset I$ be the set of all discontinuities
that are not endpoints of $I$.  As there are
only two endpoints, it is enough to show that $E$ is countable.
Without loss of generality, suppose $f$ is increasing.
We will define an injection $h \colon E \to \Q$.
For each $c \in E$,
both one-sided limits of $f$ exist as $c$ is not an endpoint.
Let
\begin{equation*}
%mbxlatex \begin{aligned}
a
%mbxlatex &
\coloneqq \lim_{x \to c^-} f(x) = \sup \bigl\{ f(x) : x \in I, x < c \bigr\} ,
%mbxSTARTIGNORE
\qquad
%mbxENDIGNORE
%mbxlatex \\
b
%mbxlatex &
\coloneqq \lim_{x \to c^+} f(x) = \inf \bigl\{ f(x) : x \in I, x > c \bigr\} .
%mbxlatex \end{aligned}
\end{equation*}
As $c$ is a discontinuity, $a < b$.  
There exists a rational number $q \in (a,b)$, so let $h(c) \coloneqq q$.
Suppose $d \in E$ is another discontinuity.
If $d > c$, there
exists an $x \in I$ with $c < x < d$, and so $\lim_{x \to d^-} f(x) \geq b$.
Hence, the rational number we choose for $h(d)$ is different from $q$,
since $q=h(c) < b$ and $h(d) > b$.
Similarly if $d < c$.  After making such a choice for
every element of $E$, we have a 
one-to-one (injective) function into $\Q$.  Therefore, $E$ is countable.
\end{proof}

\begin{example} \label{example:countdiscont}
Denote the largest integer less than or equal to $x$
by $\lfloor x \rfloor$.
Define $f \colon [0,1] \to \R$ by
\begin{equation*}
f(x) \coloneqq
x +
\sum_{n=0}^{\lfloor 1/(1-x) \rfloor}
2^{-n} ,
\end{equation*}
for $x < 1$ and $f(1) \coloneqq 3$.
It is an exercise to show that $f$ is strictly increasing, bounded, and
has a discontinuity at all points $1-\nicefrac{1}{k}$ for $k \in \N$.  In particular,
there are countably many discontinuities, but the function is bounded and
defined on a closed bounded interval.  See \figureref{fig:countdiscont}.
\begin{myfigureht}
\myincludegraphics{increasing-discont-fig}{%
A graph of a function on the interval from 0 to 1
that is composed of straight line segments sloped upwards.  There is
a first segment going across half the graph starting at 
y equal to 1.5 and ending at y equal to 2.  Then there is an upward
jump discontinuity and another much shorter line segment
also sloped upwards.  This set of smaller and smaller upward jumps and
shorter and shorter upward sloped line segments keeps going until
at x equal to 1 we get to y equal to 3.}
\caption{Strictly increasing function on $[0,1]$ with countably many
discontinuities.\label{fig:countdiscont}}
\end{myfigureht}

Similarly, one can find an example of a monotone function discontinuous on a dense set
such as the rational numbers.  See the exercises.
\end{example}


\subsection{Continuity of inverse functions}


A strictly monotone function $f$ is one-to-one (injective).
To see this fact,
notice that if $x \neq y$, then we can assume $x < y$.  Either $f(x) <
f(y)$ if $f$ is strictly increasing or $f(x) > f(y)$ if $f$ is strictly
decreasing, so $f(x) \neq f(y)$.
Hence, $f$ must have an inverse $f^{-1}$ defined on its range.

\begin{prop} \label{prop:invcont}
If $I \subset \R$ is an interval and $f \colon I \to \R$ is strictly
monotone, then the inverse $f^{-1} \colon f(I) \to I$ is continuous.
\end{prop}

\begin{proof}
Suppose $f$ is strictly increasing.  The proof is almost
identical for a strictly decreasing function.
Since $f$ is strictly increasing, so is $f^{-1}$.  That is, if $f(x) <
f(y)$, then we must have $x < y$ and therefore
$f^{-1}\bigl(f(x)\bigr) < f^{-1}\bigl(f(y)\bigr)$.

Take $c \in f(I)$.
If $c$ is not a cluster point of $f(I)$, then $f^{-1}$ is continuous at $c$
automatically.  So let $c$ be a cluster point of $f(I)$.
Suppose both of the following one-sided limits exist, that is, they both
make sense:
\begin{align*}
x_0 & \coloneqq \lim_{y \to c^-} f^{-1}(y) =
\sup \bigl\{ f^{-1}(y) : y < c, y \in f(I) \bigr\}
=
\sup \bigl\{ x \in I : f(x) < c \bigr\} , \\
x_1 & \coloneqq \lim_{y \to c^+} f^{-1}(y) =
\inf \bigl\{ f^{-1}(y) : y > c, y \in f(I) \bigr\}
=
\inf \bigl\{ x \in I : f(x) > c \bigr\} .
\end{align*}
We have $x_0 \leq x_1$ as $f^{-1}$ is increasing.
For all $x \in I$ where $x > x_0$, we have $f(x) \geq c$.  As $f$ is strictly increasing,
we must have $f(x) > c$ for all $x \in I$ where $x > x_0$.  Therefore,
\begin{equation*}
\{ x \in I : x > x_0 \} \subset \bigl\{ x \in I : f(x) > c \bigr\}.
\end{equation*}
The infimum of the left-hand set is $x_0$, and the infimum of the right-hand
set is $x_1$, so we obtain $x_0 \geq x_1$.
So $x_1 = x_0$, and $f^{-1}$ is continuous at $c$.

If one of the one-sided limits does not exist (that is, when it does not
make sense to take the limit from that side), the argument is similar
and is left as an exercise.
\end{proof}

\begin{example}
The proposition does not require $f$ itself to be continuous.  Let
$f \colon \R \to \R$ be defined by
\begin{equation*}
f(x) \coloneqq
\begin{cases}
x & \text{if } x < 0, \\
x+1 & \text{if } x \geq 0. \\
\end{cases}
\end{equation*}
The function $f$ is not continuous at $0$.
The image of $I = \R$ is the set 
$(-\infty,0)\cup [1,\infty)$, not an interval.
Then $f^{-1} \colon (-\infty,0)\cup [1,\infty)
\to \R$ can be written as
\begin{equation*}
f^{-1}(y) =
\begin{cases}
y & \text{if } y < 0, \\
y-1 & \text{if } y \geq 1. 
\end{cases}
\end{equation*}
It is not difficult to see that $f^{-1}$ is a continuous function.  See
\figureref{invcontfig} for the graphs.
\begin{myfigureht}
\myincludepdft{invcontfigAB}{%
Two graphs.  On the left, an upward sloping diagonal line that goes from
bottom left to the origin (not including the origin),
then jumps up and then continues diagonally up.
On the right graph, the first line giving the
graph is the same, but then the graph skips a short
interval and the line starts again on the horizontal
axis going diagonally up again.}
\caption{Graph of $f$ on the left and $f^{-1}$ on the right.\label{invcontfig}}
\end{myfigureht}
\end{example}

Notice what happens with the proposition if $f(I)$ is an interval.
In that case, we could simply
apply \corref{cor:continterval} to both $f$ and $f^{-1}$.  That is, if
$f \colon I \to J$ is an onto strictly monotone function and $I$ and $J$ are intervals,
then both $f$ and $f^{-1}$ are continuous.  Furthermore, $f(I)$ is an
interval precisely when $f$ is continuous.

\subsection{Exercises}

\begin{exercise}
Suppose $f \colon [0,1] \to \R$ is monotone.  Prove $f$ is bounded.
\end{exercise}

\begin{exercise}
Finish the proof of \propref{prop:monotlimits}.
Hint: You can halve your work by noticing that if $g$ is decreasing,
then $-g$ is increasing.
\end{exercise}

\begin{exercise}
Finish the proof of \corref{cor:continterval}.
\end{exercise}

\begin{exercise}
Prove the claims in \exampleref{example:countdiscont}.
\end{exercise}

\begin{exercise}
Finish the proof of \propref{prop:invcont}.
\end{exercise}

\begin{samepage}
\begin{exercise}
Suppose $S \subset \R$, and $f \colon S \to \R$ is an increasing
function.  Prove:
\begin{enumerate}[a)]
\item
If $c$ is a cluster point
of $S \cap (c,\infty)$, then
$\lim\limits_{x\to c^+} f(x) < \infty$.
\item
If $c$ is a cluster point of $S \cap (-\infty,c)$
and $\lim\limits_{x\to c^-} f(x) = \infty$, then
$S \subset (-\infty,c)$.
\end{enumerate}
\end{exercise}
\end{samepage}

\begin{exercise}
Let $I \subset \R$ be an interval and $f \colon I \to \R$ a function.
Suppose that for each $c \in I$, there exist $a, b \in \R$ with
$a > 0$ such that $f(x) \geq a x + b$ for all $x \in I$
and $f(c) = a c + b$.  Show that $f$ is strictly increasing.
\end{exercise}

\begin{exercise}
Suppose $I$ and $J$ are intervals and
$f \colon I \to J$ is a continuous, bijective (one-to-one and onto)
function.  Show that $f$ is strictly monotone.
\end{exercise}

\begin{exercise}
Consider a monotone function $f \colon I \to \R$ on an interval $I$.  Prove that there exists
a function $g \colon I \to \R$ such that
$\lim\limits_{x \to c^-} g(x) = g(c)$ for all $c$ in $I$ except the
smaller (left) endpoint of $I$, and such that
$g(x) = f(x)$ for all but countably many $x \in I$.
\end{exercise}

\begin{exercise}
\leavevmode
\begin{enumerate}[a)]
\item
Let $S \subset \R$ be a subset.  If $f \colon S \to \R$ is increasing and
bounded,
then show that there exists an increasing $F \colon \R \to \R$
such that $f(x) = F(x)$ for all $x \in S$.
\item
Find an example of a strictly increasing bounded $f \colon S \to \R$ such that
an increasing $F$ as above cannot be strictly increasing no matter how we
choose it.
\end{enumerate}
\end{exercise}

\begin{exercise}[Challenging] \label{exercise:increasingfuncdiscatQ}
Find an example of an increasing function $f \colon [0,1] \to \R$
that has a discontinuity at each rational number.  Then show that the image
$f\bigl([0,1]\bigr)$ contains no interval.  Hint: Enumerate
the rational numbers and define
the function with a series.
\end{exercise}

\begin{exercise}
Suppose $I$ is an interval and $f \colon I \to \R$ is monotone.
Show that $\R \setminus f(I)$ is a countable union of disjoint intervals.
\end{exercise}

\begin{exercise}
Suppose $f \colon [0,1] \to (0,1)$ is increasing.  Show that for every
$\epsilon > 0$, there exists
a strictly increasing $g \colon [0,1] \to (0,1)$ such that
$g(0) = f(0)$, $f(x) \leq g(x)$ for all $x$, and $g(1)-f(1) < \epsilon$.
\end{exercise}

\begin{exercise}
Prove that the \myindex{Dirichlet function}
$f \colon [0,1] \to\R$, defined by $f(x) \coloneqq 1$
if $x$ is rational and $f(x) \coloneqq 0$ otherwise, cannot be written as a
difference of two increasing functions.  That is, there do not exist
increasing $g$ and $h$ such that $f(x) = g(x) - h(x)$.
\end{exercise}

\begin{exercise}
Suppose $f \colon (a,b) \to (c,d)$ is a strictly increasing
onto function.  Prove that there exists a $g \colon (a,b) \to (c,d)$,
which is also strictly increasing and onto, and $g(x) < f(x)$ for all $x \in
(a,b)$.
\end{exercise}