Datasets:
Tasks:
Text Generation
Modalities:
Text
Sub-tasks:
language-modeling
Languages:
English
Size:
100K - 1M
License:
\chapter{Differentiation} | |
\section{Definition} | |
\prototype{$x^3$ has derivative $3x^2$.} | |
I suspect most of you have seen this before, but: | |
\begin{definition} | |
Let $U$ be an open subset\footnote{We | |
will almost always use $U = (a,b)$ or $U = \RR$, | |
and you will not lose much by restricting | |
the definition to those.} | |
of $\RR$ and let $f \colon U \to \RR$ be a function. | |
Let $p \in U$. | |
We say $f$ is \vocab{differentiable} at $p$ | |
if the limit\footnote{Remember we are following the convention in | |
\Cref{abuse:limit}. | |
So we mean ``the limit of the function $h \mapsto \frac{f(p+h)-f(p)}{h}$ | |
except the value at $h=0$ can be anything''. | |
And this is important because that fraction | |
does not have a definition at $h = 0$. | |
As promised, we pay this no attention.} | |
\[ \lim_{h \to 0} \frac{f(p+h) - f(p)}{h} \] | |
exists. | |
If so, we denote its value by $f'(p)$ and refer | |
to this as the \vocab{derivative} of $f$ at $p$. | |
The function $f$ is differentiable if it is differentiable | |
at every point. | |
In that case, we regard the derivative $f' \colon (a,b) \to \RR$ | |
as a function it its own right. | |
\end{definition} | |
\begin{exercise} | |
Show that if $f$ is differentiable at $p$ | |
then it is continuous at $p$ too. | |
\end{exercise} | |
Here is the picture. | |
Suppose $f \colon \RR \to \RR$ is differentiable | |
(hence continuous). | |
We draw a graph of $f$ in the usual way and consider values of $h$. | |
For any nonzero $h$, what we get is the slope of the \emph{secant} | |
line joining $(p, f(p))$ to $(p+h, f(p+h))$. | |
However, as $h$ gets close to zero, | |
that secant line begins to approach a line | |
which is tangent to the graph of the curve. | |
A picture with $f$ a parabola is shown below, | |
with the tangent in red, and the secant in dashed green. | |
\begin{center} | |
\begin{asy} | |
import graph; | |
size(8cm); | |
real f(real x) { return (x-2)*(x-2)/2 - 0.1; } | |
graph.xaxis("$x$"); | |
graph.yaxis("$y$"); | |
draw(graph(f,-1,5,operator ..), blue, Arrows); | |
pair P = (3, f(3)); | |
dot("$(p, f(p))$", P, dir(-20), red); | |
draw((1.8,-0.8)--(4.2, 1.6), red); | |
label("Slope $f'(p)$", (4.2, 1.6), dir(-25), red); | |
pair Q = (4.3, f(4.3)); | |
dot("$(p+h, f(p+h))$", Q, dir(-20), deepgreen); | |
draw(P--Q, dashed+deepgreen); | |
label("Slope $\frac{f(p+h)-f(p)}{h}$", P--Q, 1.5*dir(165), deepgreen); | |
\end{asy} | |
\end{center} | |
So the picture in your head should be that | |
\begin{moral} | |
$f'(p)$ looks like the slope of the tangent line at $(p, f(p))$. | |
\end{moral} | |
\begin{remark} | |
Note that the derivatives are defined | |
for functions on \emph{open} intervals. | |
This is important. | |
If $f \colon [a,b] \to \RR$ for example, | |
we could still define the derivative at each interior point, | |
but $f'(a)$ no longer makes sense | |
since $f$ is not given a value on any open neighborhood of $a$. | |
\end{remark} | |
Let's do one computation and get on with this. | |
\begin{example} | |
[Derivative of $x^3$ is $3x^2$] | |
Let $f \colon \RR \to \RR$ by $f(x) = x^3$. | |
For any point $p$, and \emph{nonzero} $h$ we can compute | |
\begin{align*} | |
\frac{f(p+h) - f(p)}{h} &= \frac{(p+h)^3 - p^3}{h} \\ | |
&= \frac{3p^2h + 3ph^2 + h^3}{h} \\ | |
&= 3p^2 + 3ph + h^2. | |
\end{align*} | |
Thus, | |
\[ \lim_{h \to 0} \frac{f(p+h)-f(p)}{h} | |
= \lim_{h \to 0} (3p^2+3ph+h^2) = 3p^2. \] | |
Thus the slope at each point of $f$ is given by the formula $3p^2$. | |
It is customary to then write $f'(x) = 3x^2$ | |
as the derivative of the entire function $f$. | |
\end{example} | |
\begin{abuse} | |
We will now be sloppy and write this as $(x^3)' = 3x^2$. | |
This is shorthand for the significantly more verbose | |
``the real-valued function $x^3$ on domain so-and-so | |
has derivative $3p^2$ at every point $p$ in its domain''. | |
In general, a real-valued differentiable function | |
$f \colon U \to \RR$ naturally gives rise to derivative | |
$f'(p)$ at every point $p \in U$, | |
so it is customary to just give up on $p$ altogether | |
and treat $f'$ as function itself $U \to \RR$, | |
even though this real number is of a ``different interpretation'': | |
$f'(p)$ is meant to interpret a slope (e.g.\ your hourly pay rate) | |
as opposed to a value (e.g.\ your total dollar worth at time $t$). | |
If $f$ is a function from real life, the units do not even match! | |
This convention is so deeply entrenched I cannot uproot it | |
without more confusion than it is worth. | |
But if you read the chapters on multivariable calculus | |
you will see how it comes back to bite us, | |
when I need to re-define the derivative to be a \emph{linear map}, | |
rather than single real numbers. | |
\end{abuse} | |
\section{How to compute them} | |
Same old, right? | |
Sum rule, all that jazz. | |
\begin{theorem} | |
[Your friendly high school calculus rules] | |
In what follows $f$ and $g$ are differentiable functions, | |
and $U$, $V$ are open subsets of $\RR$. | |
\begin{itemize} | |
\ii (Sum rule) If $f,g \colon U \to \RR$ then | |
then $(f+g)'(x) = f'(x) + g'(x)$. | |
\ii (Product rule) If $f,g \colon U \to \RR$ then | |
then $(f \cdot g)'(x) = f'(x) g(x) + f(x) g'(x)$. | |
\ii (Chain rule) If $f \colon U \to V$ and $g \colon V \to \RR$ | |
then the derivative of the composed function | |
$g \circ f \colon U \to \RR$ is $g'(f(x)) \cdot f'(x)$. | |
\end{itemize} | |
\end{theorem} | |
\begin{proof} | |
\begin{itemize} | |
\ii Sum rule: trivial, do it yourself if you care. | |
\ii Product rule: for every nonzero $h$ and point $p \in U$ | |
we may write | |
\[ | |
\frac{f(p+h) g(p+h) - f(p) g(p)}{h} | |
= \frac{f(p+h) - f(p)}{h} \cdot g(p+h) | |
+ \frac{g(p+h)-g(p)}{h} \cdot f(p) | |
\] | |
which as $h \to 0$ gives the desired expression. | |
\ii Chain rule: this is where \Cref{abuse:limit} | |
will actually bite us. | |
Let $p \in U$, $q = f(p) \in V$, so that | |
\[ (g \circ f)'(p) = \lim_{h \to 0} \frac{g(f(p+h)) - g(q)}{h}. \] | |
We would like to write the expression in the limit as | |
\[ \frac{g(f(p+h)) - g(q)}{h} | |
= \frac{g(f(p+h)) - g(q)}{f(p+h) - q} | |
\cdot \frac{f(p+h) - f(p)}{h}. \] | |
The problem is that the denominator $f(p+h)-f(p)$ might be zero. | |
So instead, we define the expression | |
\[ | |
Q(y) = \begin{cases} | |
\frac{g(y) - g(q)}{y - q} & \text{if } y \ne q \\ | |
g'(q) & \text{if } y = q | |
\end{cases} | |
\] | |
which is continuous since $g$ was differentiable at $q$. | |
Then, we \emph{do} have the equality | |
\[ \frac{g(f(p+h)) - g(q)}{h} | |
= Q\left( f(p+h) \right) \cdot \frac{f(p+h) - f(p)}{h}. \] | |
because if $f(p+h) = q$ with $h \ne 0$, | |
then both sides are equal to zero anyways. | |
Then, in the limit as $h \to 0$, | |
we have $\lim_{h \to 0} \frac{f(p+h)-f(p)}{h} = f'(p)$, | |
while $\lim_{h \to 0} Q(f(p+h)) = Q(q) = g'(q)$ by continuity. | |
This was the desired result. \qedhere | |
\end{itemize} | |
\end{proof} | |
\begin{exercise} | |
Compute the derivative of the polynomial $f(x) = x^3 + 10x^2 + 2019$, | |
viewed as a function $f \colon \RR \to \RR$. | |
\end{exercise} | |
\begin{remark} | |
Quick linguistic point: | |
the theorems above all hold at each individual point. | |
For example the sum rule really should say that | |
if $f,g \colon U \to \RR$ are differentiable at the point $p$ | |
then so is $f+g$ and the derivative equals $f'(p) + g'(p)$. | |
Thus $f$ and $g$ are differentiable on all of $U$, | |
then it of course follows that $(f+g)' = f' + g'$. | |
So each of the above rules has a ``point-by-point'' form | |
which then implies the ``whole $U$'' form. | |
We only state the latter since that is what is used in practice. | |
However, in the rare situations where you have a function | |
differentiable only at certain points of $U$ rather | |
than the whole interval $U$, you can still use the below. | |
\end{remark} | |
We next list some derivatives of | |
well-known functions, | |
but as we do not give rigorous definitions | |
of these functions, we do not prove these here. | |
\begin{proposition} | |
[Derivatives of some well-known functions] | |
\listhack | |
\begin{itemize} | |
\ii The exponential function $\exp \colon \RR \to \RR$ | |
defined by $\exp(x) = e^x$ is its own derivative. | |
\ii The trig functions $\sin$ and $\cos$ | |
have $\sin' = \cos$, $\cos' = -\sin$. | |
\end{itemize} | |
\end{proposition} | |
\begin{example} | |
[A typical high-school calculus question] | |
This means that you can mechanically compute | |
the derivatives of any artificial function obtained by using the above, | |
which makes it a great source of busy work | |
in American high schools and universities. | |
For example, if | |
\[ f(x) = e^x + x \sin(x^2) \qquad f \colon \RR \to \RR \] | |
then one can compute $f'$ by: | |
\begin{align*} | |
f'(x) &= (e^x)' + (x \sin(x^2))' & \text{sum rule} \\ | |
&= e^x + (x \sin(x^2))' & \text{above table} \\ | |
&= e^x + (x)' \sin(x^2) + x (\sin(x^2))' & \text{product rule} \\ | |
&= e^x + \sin(x^2) + x (\sin(x^2))' & (x)' = 1 \\ | |
&= e^x + \sin(x^2) + x \cdot 2x \cdot \cos(x^2) & \text{chain rule}. | |
\end{align*} | |
Of course, this function $f$ is totally artificial and has no meaning, | |
which is why calculus is the topic of widespread scorn in the USA. | |
That said, it is worth appreciating that calculations like | |
this are possible: it would be better to write the pseudo-theorem | |
``derivatives can actually be computed''. | |
\end{example} | |
If we take for granted that $(e^x)' = e^x$, | |
then we can derive two more useful functions | |
to add to our library of functions we can differentiate. | |
\begin{corollary} | |
[Power rule] | |
Let $r$ be a real number. | |
The function $\RR_{>0} \to \RR$ by $x \mapsto x^r$ | |
has derivative $(x^r)' = rx^{r-1}$. | |
\end{corollary} | |
\begin{proof} | |
We knew this for integers $r$ already, | |
but now we can prove it for any positive real number $r$. | |
Write | |
\[ f(x) = x^r = e^{r \log x} \] | |
considered as a function $f \colon \RR_{>0} \to \RR$. | |
The chain rule (together with the fact that $(e^x)' = e^x$) | |
now gives | |
\begin{align*} | |
f'(x) &= e^{r \log x} \cdot (r \log x)' \\ | |
&= e^{r \log x} \cdot \frac rx = x^r \cdot \frac rx = rx^{r-1}. | |
\end{align*} | |
The reason we don't prove the formulas for $e^x$ and $\log x$ | |
is that we don't at the moment even have a rigorous | |
definition for either, or even for $2^x$ if $x$ is not rational. | |
However it's nice to know that some things imply the other. | |
\end{proof} | |
\begin{corollary} | |
[Derivative of $\log$ is $1/x$] | |
The function $\log \colon \RR_{>0} \to \RR$ | |
has derivative $(\log x)' = 1/x$. | |
\end{corollary} | |
\begin{proof} | |
We have that $x = e^{\log x}$. | |
Differentiate both sides, and again use the chain rule\footnote{There | |
is actually a small subtlety here: | |
we are taking for granted that $\log$ is differentiable.} | |
\[ 1 = e^{\log x} \cdot (\log x)'. \] | |
Thus $(\log x)' = \frac{1}{e^{\log x}} = 1/x$. | |
\end{proof} | |
\section{Local (and global) maximums} | |
\prototype{Horizontal tangent lines to the parabola | |
are typically good pictures.} | |
You may remember from high school | |
that one classical use of calculus | |
was to extract the minimum or maximum values of functions. | |
We will give a rigorous description of how to do this here. | |
\begin{definition} | |
Let $f \colon U \to \RR$ be a function. | |
A \vocab{local maximum} is a point $p \in U$ | |
such that there exists an open neighborhood $V$ of $p$ | |
(contained inside $U$) | |
such that $f(p) \ge f(x)$ for every $x \in V$. | |
A \vocab{local minimum} is defined similarly.\footnote{Equivalently, | |
it is a local maximum of $-f$.} | |
\end{definition} | |
\begin{definition} | |
A point $p$ is a \vocab{local extrema} | |
if it satisfies either of these. | |
\end{definition} | |
The nice thing about derivatives is that they pick up all extrema. | |
\begin{theorem} | |
[Fermat's theorem on stationary points] | |
Suppose $f \colon U \to \RR$ is differentiable | |
and $p \in U$ is a local extrema. | |
Then $f'(p) = 0$. | |
\end{theorem} | |
If you draw a picture, this result is not surprising. | |
\begin{center} | |
\begin{asy} | |
import graph; | |
size(7cm); | |
real f(real x) { return 2-(x-2)*(x-2)/3; } | |
graph.xaxis("$x$"); | |
graph.yaxis("$y$"); | |
draw(graph(f,-1,5,operator ..), blue, Arrows); | |
pair P = (2, f(2)); | |
dot("$(p, f(p))$", P, dir(90), red); | |
draw( (0.3,f(2))--(3.7,f(2)), red ); | |
\end{asy} | |
\end{center} | |
(Note also: the converse is not true. | |
Say, $f(x) = x^{2019}$ has $f'(0) = 0$ | |
but $x=0$ is not a local extrema for $f$.) | |
\begin{proof} | |
Assume for contradiction $f'(p) > 0$. | |
Choose any $\eps > 0$ with $\eps < f'(p)$. | |
Then for sufficiently small $|h|$ we should have | |
\[ \frac{f(p+h)-f(p)}{h} > \eps. \] | |
In particular $f(p+h) > f(p)$ for $h > 0$ | |
while $f(p-h) < f(p)$ for $h < 0$. | |
So $p$ is not a local extremum. | |
The proof for $f'(p) < 0$ is similar. | |
\end{proof} | |
However, this is not actually adequate | |
if we want a complete method for optimization. | |
The issue is that we seek \emph{global} extrema, | |
which may not even exist: | |
for example $f(x) = x$ (which has $f'(x) = 1$) | |
obviously has no local extrema at all. | |
The key to resolving this is to use \emph{compactness}: | |
we change the domain to be a compact set $Z$, | |
for which we know that $f$ will achieve some global maximum. | |
The set $Z$ will naturally have some \emph{interior} $S$, | |
and calculus will give us all the extrema within $S$. | |
Then we manually check all cases outside $Z$. | |
Let's see two extended examples. | |
The one is simple, and you probably already know about it, | |
but I want to show you how to use compactness to argue thoroughly, | |
and how the ``boundary'' points naturally show up. | |
\begin{example} | |
[Rectangle area optimization] | |
Suppose we consider rectangles with perimeter $20$ | |
and want the rectangle with the smallest or largest area. | |
\begin{center} | |
\begin{asy} | |
size(4cm); | |
draw( (0,0)--(7,0)--(7,3)--(0,3)--cycle ); | |
label("$10-x$", (3.5,0), dir(-90)); | |
label("$x$", (0,1.5), dir(180)); | |
\end{asy} | |
\end{center} | |
If we choose the legs of the rectangle to be $x$ and | |
$10-x$, then we are trying to optimize the function | |
\[ f(x) = x(10-x) = 10x-x^2 \qquad f \colon [0,10] \to \RR. \] | |
By compactness, there exists \emph{some} global maximum | |
and \emph{some} global minimum. | |
As $f$ is differentiable on $(0,10)$, | |
we find that for any $p \in (0,10)$, a global maximum | |
will be a local maximum too, and hence should satisfy | |
\[ 0 = f'(p) = 10 - 2p \implies p = 5. \] | |
Also, the points $x = 0$ and $x = 10$ lie in the domain | |
but not the interior $(0,10)$. | |
Therefore the global extrema (in addition to existing) | |
must be among the three suspects $\{0, 5, 10\}$. | |
We finally check $f(0) = 0$, $f(5) = 25$, $f(10) = 0$. | |
So the $5 \times 5$ square has the largest area | |
and the degenerate rectangles have the smallest (zero) area. | |
\end{example} | |
Here is a non-elementary example. | |
\begin{proposition}[$e^x \ge 1+x$] | |
For all real numbers $x$ we have $e^x \ge 1+x$. | |
\end{proposition} | |
\begin{proof} | |
Define the differentiable function | |
\[ f(x) = e^x - (x+1) \qquad f \colon \RR \to \RR. \] | |
Consider the compact interval $Z = [-1,100]$. | |
If $x \le -1$ then obviously $f(x) > 0$. | |
Similarly if $x \ge 100$ then obviously $f(x) > 0$ too. | |
So we just want to prove that if $x \in Z$, we have $f(x) \ge 0$. | |
Indeed, there exists \emph{some} global minimum $p$. | |
It could be the endpoints $-1$ or $100$. | |
Otherwise, if it lies in $U = (-1, 100)$ | |
then it would have to satisfy | |
\[ 0 = f'(p) = e^p - 1 \implies p = 0. \] | |
As $f(-1) > 0$, $f(100) > 0$, $f(0) = 0$, | |
we conclude $p = 0$ is the global minimum of $Z$; | |
and hence $f(x) \ge 0$ for all $x \in Z$, hence for all $x$. | |
\end{proof} | |
\begin{remark} | |
If you are willing to use limits at $\pm \infty$, | |
you can rewrite proofs like the above in such a way | |
that you don't have to explicitly come | |
up with endpoints like $-1$ or $100$. | |
We won't do so here, but it's nice food for thought. | |
\end{remark} | |
\section{Rolle and friends} | |
\prototype{The racetrack principle, perhaps?} | |
One corollary of the work in the previous section is Rolle's theorem. | |
\begin{theorem} | |
[Rolle's theorem] | |
Suppose $f \colon [a,b] \to \RR$ is a continuous function, | |
which is differentiable on the open interval $(a,b)$, | |
such that $f(a) = f(b)$. | |
Then there is a point $c \in (a,b)$ such that $f'(c) = 0$. | |
\end{theorem} | |
\begin{proof} | |
Assume $f$ is nonconstant (otherwise any $c$ works). | |
By compactness, there exists both a global maximum and minimum. | |
As $f(a) = f(b)$, either the global maximum | |
or the global minimum must lie inside the open interval $(a,b)$, | |
and then Fermat's theorem on stationary points finishes. | |
\end{proof} | |
I was going to draw a picture until I realized xkcd \#2042 has one already. | |
\begin{center} | |
\includegraphics[scale=0.66]{media/xkcd-rolles.png} | |
\\ \scriptsize Image from \cite{img:xkcd_rolles} | |
\end{center} | |
One can adapt the theorem as follows. | |
\begin{theorem} | |
[Mean value theorem] | |
Suppose $f \colon [a,b] \to \RR$ is a continuous function, | |
which is differentiable on the open interval $(a,b)$. | |
Then there is a point $c \in (a,b)$ such that | |
\[ f'(c) = \frac{f(b)-f(a)}{b-a}. \] | |
\end{theorem} | |
Pictorially, there is a $c$ such that the tangent at $c$ | |
has the same slope as the secant joining $(a, f(a))$, to $(b, f(b))$; | |
and Rolle's theorem is the special case where that secant is horizontal. | |
\begin{center} | |
\begin{asy} | |
import graph; | |
size(7cm); | |
real f(real x) { return x*x/2 - 0.2; } | |
graph.xaxis("$x$"); | |
graph.yaxis("$y$"); | |
draw(graph(f,-2,2.5,operator ..), blue, Arrows); | |
pair A = (-1, f(-1)); | |
pair B = (2, f(2)); | |
dot("$(a, f(a))$", A, dir(A-B), deepgreen); | |
dot("$(b, f(b))$", B, dir(10), deepgreen); | |
draw(A--B, deepgreen); | |
label("Slope $\frac{f(b)-f(a)}{b-a}$", A--B, dir(120), deepgreen); | |
draw(A--(A.x,0), deepgreen+dashed); | |
draw(B--(B.x,0), deepgreen+dashed); | |
label("$a$", (A.x,0), dir(-90), deepgreen); | |
label("$b$", (B.x,0), dir(-90), deepgreen); | |
real c = (A.y-B.y) / (A.x-B.x); | |
pair C = (c, f(c)); | |
dot("$(c, f(c))$", C, dir(-70), red); | |
draw( (c-1, f(c)-c)--(c+1, f(c)+c), red ); | |
\end{asy} | |
\end{center} | |
\begin{proof} | |
[Proof of mean value theorem] | |
Let $s = \frac{f(b)-f(a)}{b-a}$ be the slope of the secant line, | |
and define | |
\[ g(x) = f(x) - s x \] | |
which intuitively shears $f$ downwards so that the | |
secant becomes vertical. | |
In fact $g(a) = g(b)$ now, so we apply Rolle's theorem to $g$. | |
\end{proof} | |
\begin{remark} | |
[For people with driver's licenses] | |
There is a nice real-life interpretation of this I should mention. | |
A car is travelling along a one-dimensional road | |
(with $f(t)$ denoting the position at time $t$). | |
Suppose you cover $900$ kilometers in your car | |
over the course of $5$ hours | |
(say $f(0) = 0$, $f(5) = 900$). | |
Then there is \emph{some} point at time in which | |
your speed at that moment was exactly $180$ kilometers per hour, | |
and so you cannot really complain | |
when the cops pull you over for speeding. | |
\end{remark} | |
The mean value theorem is important because it lets | |
you relate \textbf{use derivative information | |
to get information about the function} | |
in a way that is really not possible without it. | |
Here is one quick application to illustrate my point: | |
\begin{proposition} | |
[Racetrack principle] | |
Let $f, g \colon \RR \to \RR$ be two differentiable functions | |
with $f(0) = g(0)$. | |
\begin{enumerate}[(a)] | |
\ii If $f'(x) \ge g'(x)$ for every $x > 0$, | |
then $f(x) \ge g(x)$ for every $x > 0$. | |
\ii If $f'(x) > g'(x)$ for every $x > 0$, | |
then $f(x) > g(x)$ for every $x > 0$. | |
\end{enumerate} | |
\end{proposition} | |
This proposition might seem obvious. | |
You can think of it as a race track for a reason: | |
if $f$ and $g$ denote the positions of two cars (or horses etc) | |
and the first car is always faster than the second car, | |
then the first car should end up ahead of the second car. | |
At a special case $g = 0$, this says that if $f'(x) \ge 0$, | |
i.e.\ ``$f$ is increasing'', | |
then, well, $f(x) \ge f(0)$ for $x > 0$, which had better be true. | |
However, if you try to prove this by definition from derivatives, | |
you will find that it is not easy! | |
However, it's almost a prototype for the mean value theorem. | |
\begin{proof} | |
[Proof of racetrack principle] | |
We prove (a). Let $h = f-g$, so $h(0) = 0$. | |
Assume for contradiction $h(p) < 0$ for some $p > 0$. | |
Then the secant joining $(0, h(0))$ to $(p, h(p))$ has negative slope; | |
in other words by mean value theorem there is a $0 < c < p$ | |
such that | |
\[ f'(c) - g'(c) = h'(c) = \frac{h(p)-h(0)}{p} = \frac{h(p)}{p} < 0 \] | |
so $f'(c) < g'(c)$, contradiction. | |
Part (b) is the same. | |
\end{proof} | |
Sometimes you will be faced with two functions which you cannot | |
easily decouple; the following form may be more useful in that case. | |
\begin{theorem} | |
[Ratio mean value theorem] | |
Let $f, g \colon [a,b] \to \RR$ be two continuous functions | |
which are differentiable on $(a,b)$, | |
and such that $g(a) \neq g(b)$. | |
Then there exists $c \in (a,b)$ such that | |
\[ f'(c)(g(b)-g(a)) = g'(c)(f(b)-f(a)) \] | |
\end{theorem} | |
\begin{proof} | |
Use Rolle's theorem on the function | |
\[ h(x) = \left[ f(x)-f(a) \right] \left[ g(b)-g(a) \right] | |
- \left[ g(x)-g(a) \right] \left[ f(b)-f(a) \right]. \qedhere \] | |
\end{proof} | |
This is also called Cauchy's mean value theorem | |
or the extended mean value theorem. | |
\section{Smooth functions} | |
\prototype{All the functions you're used to.} | |
Let $f \colon U \to \RR$ be differentiable, | |
thus giving us a function $f' \colon U \to \RR$. | |
If our initial function was nice enough, | |
then we can take the derivative again, | |
giving a function $f'' \colon U \to \RR$, and so on. | |
In general, after taking the derivative $n$ times, | |
we denote the resulting function by $f^{(n)}$. | |
By convention, $f^{(0)} = f$. | |
\begin{definition} | |
A function $f \colon U \to \RR$ is \vocab{smooth} | |
if it is infinitely differentiable; | |
that is the function $f^{(n)}$ exists for all $n$. | |
\end{definition} | |
\begin{ques} | |
Show that the absolute value function is not smooth. | |
\end{ques} | |
Most of the functions we encounter, | |
such as polynomials, $e^x$, $\log$, $\sin$, $\cos$ | |
are smooth, and so are their compositions. | |
Here is a weird example which we'll grow more next time. | |
\begin{example} | |
[A smooth function with all derivatives zero] | |
Consider the function | |
\[ f(x) = \begin{cases} | |
e^{-1/x} & x > 0 \\ | |
0 & x \le 0. | |
\end{cases} | |
\] | |
This function can be shown to be smooth, | |
with $f^{(n)}(0) = 0$. | |
So this function has every derivative at the origin | |
equal to zero, despite being nonconstant! | |
\end{example} | |
\section{\problemhead} | |
\begin{problem} | |
[Quotient rule] | |
Let $f \colon (a,b) \to \RR$ and $g \colon (a,b) \to \RR_{>0}$ | |
be differentiable functions. | |
Let $h = f/g$ be their quotient | |
(also a function $(a,b) \to \RR$). | |
Show that the derivative of $h$ is given by | |
\[ h'(x) = \frac{f'(x) g(x) - f(x) g'(x)}{g(x)^2}. \] | |
\end{problem} | |
\begin{problem} | |
For real numbers $x > 0$, how small can $x^x$ be? | |
\end{problem} | |
\begin{problem} | |
[RMM 2018] | |
\gim | |
Determine whether or not there exist | |
nonconstant polynomials $P(x)$ and $Q(x)$ with | |
real coefficients satisfying | |
\[ P(x)^{10} + P(x)^9 = Q(x)^{21} + Q(x)^{20}. \] | |
\end{problem} | |
\begin{problem} | |
\gim | |
Let $P(x)$ be a degree $n$ polynomial with real coefficients. | |
Prove that the equation $e^x = P(x)$ has at most $n+1$ real solutions in $x$. | |
\end{problem} | |
\begin{problem} | |
[Jensen's inequality] | |
Let $f \colon (a,b) \to \RR$ be a twice differentiable function | |
such that $f''(x) \ge 0$ for all $x$ | |
(i.e.\ $f$ is \emph{convex}). | |
Prove that | |
\[ f\left( \frac{x+y}{2} \right) | |
\le \frac{f(x) + f(y)}{2} \] | |
for all real numbers $x$ and $y$ in the interval $(a,b)$. | |
\end{problem} | |
\begin{problem} | |
[L'H\^{o}pital rule, or at least one case] | |
Let $f,g \colon \RR \to \RR$ be differentiable functions | |
and let $p$ be a real number. | |
Suppose that | |
\[ \lim_{x \to p} f(x) = \lim_{x \to p} g(x) = 0. \] | |
Prove that | |
\[ \lim_{x \to p} \frac{f(x)}{g(x)} | |
= \lim_{x \to p} \frac{f'(x)}{g'(x)} \] | |
provided the right-hand limit exists. | |
\end{problem} | |