Review of Calculus

differential

-rates of change
-

speed
-

acceleration

I
f(x) f(a) f(a)(x a)

W
= + -

"
+

If "(a)(x a)
-

·
·

-
-
7 S
a

Integral
"continuous addition"
integration
:

#
rixidy- area under the
curve between
a and b
multivariable calculus

in
-rates of
change many variables

I FRE

I
-
IR

* approximation by some plane

f(x y) f(a fx(a b)(X a)

=
b) b)
fy(a b)(y
+ - +
+ -

, , ,

gral double
that stands
integral
for volume

#
S f(x y)dxdy
,
=
volume

/ under the
graph
of the
region D
 Recap of series
real #

term
A
A ceries is a cum with and IR

we call Swan ->

the Nth partial sum

If the
pim SN=S exists in IR
an
converges and we

If to and Iris
write
San= S
example : a

rEIR , then art= Er

with a
n
Indeed ,
Sw =
Earm= a +art arN=

Nimo a
*
= it a
1- r

e
 THE OREMS
theorem if
: an
converges ,
then
him an =
0

Consequence (test for

divergence)

if
him ,
an Fo
,
then E ,
an
diverges

theorem (integral test) say it is a continous positive function decreasing

on (1 c) let an
,
=

f (n)
then a an A

S
-I
an

-...
n1

as
=

! S

11 If I f(x)dx converges ,
then so does an
,

2) if I f(x)dx does E
diverges ,
then so an

example let p>0 be a real number

,

C
if if
↳ converges >I and
diverges pal
n 1
=

Indeed let
decreasing positive [1 8)
f(x) continuous and on
=
o
, ,
,

I lim"
!
ax=
nax ox11p)
=

t -
c

limioxn"=yim
1 ,
=

/endivergen
if =1 I Hax =

em(x) -

Him int =
c
 value of
8 f(x) ax says nothing about the value of
,
an

it
pick seveninteger, the
if ratio a

E an
conotenever gent of enter overgent
A series

Fact : It San is

Alternating Series test

say we have a series f) bn= b ,

b + by bat ...

* with each bu> 0

H
*
if but bu and bu is
nmbr= 0
,
then
convergent

example

(n 1 -
=

+
-...

It is
convergent by the series test it , In ni = 0

But not
absolutely convergent !I

151n*1 = diverges
 the ratio test

let
San be a series

1) if nml)=< 1 ,
then & also is absolutely convergent

a) if
(a) =

21 or s
,
then an is divergent

Remark
if
m /)=1 ,
then the test tests
nothing

example

So
Let XER and consider
.
it =

1 +

x
+ +
+
+ ...

with an
=

*
o

11 : O

So the ratio test implies that for all xEIR ,

So is absolutely convergent

Example

Let XIR and consider

S(-1)n 4 4 2 +
-

= x - +
-

...

n =
1

Here an= s-132 = x

=
/x1 (n) =
(x
(ii) mon
IX

if IXK 1 -1) is
,
then divergent
if X
=

if IX/<1 ,
then C-1(n- A is absolutely convergent

if X
=

1
,
then 1 -

+ -
+
...

converges
 Math 222 Lecture -

What is calculus ?

Differentical

a re
·

Tangent lines and fogration="continuous

linear
approximation addition

i
f(x) =
f(a) +

f(a)(X -

a)

Highes order approximation If(xdX= f(x) between

·

under
area

i
and b

f(x) =
f(a) f(a)(x a)
+
-
+ fatal
 Multivariable
Differential
Integral
·
Rates o
Motion
charge
volocity und
Intragration="continuous
addition
us

acceleration
with direction
·
Linear
approximation

--
C
"
f(x, y)=f(a b) + ,

fx(a b)(x a) ,
-
(f(x y)dxdy
,
=

b)(y b)
"valum" under f(x y)
+fy(a
,
-

the &
over
region
in the
(xy)-plane
-
 Recap on Series

Aries is a sun an with anot

N

We call Sw=Ean the N* partial sum .

n 0
=

If lim SN =

S exists in RR , we
say
N - 1
A

Ean and write an=s

n
=
0
converges
it
Otherwise , we
say diverges .

1512 1 with areIR than

If at 8 and , ,

Ear =

*
n 0
=

Indeed ,
Sear *
- axarx ...

rNx
ap
al
-

=
-

I - ~

and
Is * Er if 1511 .

Thm)-Theorem) If an theu ater an= 8

converges ,

Consequence test for

divergence) :

If ant 8, then an
diverges
The (The intrgral test)
Say & is a continuous , decreasing , positive function
on
18, 2) and let antf(n) .

Then
1 .
If Sof(xdX converges ,
so doos
n = 0
an

2 P SfIXdx does
En
.

dierges ,
so

Let real number Than

g
E
p>8 be a .

it converges
if pul and
diverges if

p=1
Let f(x) =

E continuous , docreusing (since >01 and ,

positive an [ind

Pp I SY +
= Espxpl- -

p
-I

Sp
-
iP8s

iPpdx =

x))= (n) =
0 -
 Volum of
Rak (Remork)
convergent fixdx

says nothing about value of Ian 7

-

n=

i
pickle evening
is

rational number et

odd integer
p=24x1 is

ident
nate- ?? We have no
 Lecture #2 09/05/2023

Last time Ratio test

:

example
divergent
↓
, is

2)
En ,
is
convergent

cratio test on these series(

inconclusive

In/ it =
ratio test will not

1)

I
be useful
result
is
"1"

2) Iantise/ne) :

Fixe :

(ii) & ratio test

be useful
will not

* so ,
ratio test is inconclusive if

him (a)= I
 The root test

Let an be a series

1) if
him "anl=2<I is
absolutely convergent

2) if im "Tanl=1 is divergent
n-g

* Remark :

if lim ~lan)=1
r ,
the test is inconclusive
n +9

Ex
Let Xt IR .
Consider use root test to

figure out for which

Ex ( )
x does the

converges/diverges
series
or

is inconclusive
an
=

x ()"
n

.
/X
lan)= ")"1
r .: x
:

1x1

1) is
If IXI<1 absolutely convergent
2) if IX1>1 is
divergent
3) if (x1 =
1 test is inconclusive
 for 1x1=1 we get E (n)"

nim (1)= him

iin him = <0

the terms of this series do to 0 the is

not so series
divergent
*
converge ,

for x =-1 it is trickier

,

Often we want to estimate the sum of the series in case it's convergent
,

Ex
Let Irk1
we , assume that
M
because we can't divide by e

infinite sum

a . -r

In cale of series do not have such

most
convergent ,
we a

closed formula .

How well is In
We still can try to estimate :
approximating s?

partial
S =

E an Set RN =

IS - Sal
sum ,

by SN- ,
an
 test
Integral estimate

Let
f be continuous
,
positive and
decreasing [1
,
3)

suppose
f(n)
=

an

if
1 8f(x)dx
,
is convergent ,
Series is
convergent
and we can eltimate our

then , n Crate of
convergence) .

D ,gixax RN
/gixsax

- +1
i
Alternating series test estimate

decreasing
Let
F be be a series
Sun but 1
that

= b
but

I
,
series is
absolutely
convergent
and
nm bu =
0

then ,
(RN) < bn+1 -
RN=1S-sol

Alternating series :

Ex Consider abn When the an's

,
are
Alternating
is
& Hin an signs .

n
=
1
-
an

n
I
bu =
bu>0

antnrge
Assumpti satisfied

O we will use the

alternating series

N bu+ 1 test estimate to

could
in
I small but
b compute up
32 have an impact
a
I
= 0 0052 to 2 decimal places
2
.

192 small
a enough
0 00090
-a
.

↳ that is how we know

55
=

+ 5-
+
2
=

0 22396
.

which integer to use (R31 by=

=
0 . 00098
for the test
* add/subs 0 . 00098 to 0 22396
. does not

change the first two decimal places

(1) 0 2
=

so , 1n
 Power series

Hef a power series is a series of the form :

S (no x =

Co +
(1x +

(2 x2+
:

n en ...

=
0
an

we view x as a variable
,

stick in 1 x value
,
get a .

A The In's are called the Coefficients

More will also consider the series

generally ,
we
of the form :

an(X-al ,
for a fixed a EIR

->
we call it a power series centered at a

Converges for some values of X and

diverges
for some other values of X .

# & c .
x" is
convergent
if IXI <
n 1
=

is if IxI>
divergent I

if
is divergent IxI= 1
is
divergent if Ix=-1

n
** converges for all xEIR
 <example Continued)

3 (use ratio test !!! )

=
3)
an

|a+)= /(34) =

((x -
3) .) =
1x-51 . (x-31

if Ix-31<1 it
converges
2XL1

if 1x-3K1 it
diverges
x<2 or x>1
if x =
2
,
x =
& test is inconclusive .

if x = 2 We have :

I converges by
alternating series test

if x =
d
,
we have

Ent diverges by
integral test

,
* To sum
up

converges if and only if 24x<4 Xt [2 2)

,
 Lectwo

& A series
n =
0
an is
absolutely convergent
if
Elan is
convergent .

The If a series is absolutely convergent ,

it is

convergent .

Alternating Spries Yest Say we have a series

H"bu =

bo-b,+ bi-bet---

with bu: Then if

·
babu for all n

im b
=

Then
Estelbe converges -

Ent =

I - * -

**
-

-
 This is
convergent by
the
alternating serves test
since
= for all n

and
s
=

0 .
But it is not absolutely
convergent since

EIF)= E * diverges .

The Ratio Test Eran Let be a series .

1 If
=LI then
.

,
an is
n 0
=

absolutely convergent .

2 If
.

h =I or , then can
is
divergent .

Rub If the limit =

1 the ration test says nothing
, ,
the serves
i .
e
may converge or
diverge .

E Let XE IR ,
and consider
3

= 1+ +
* +x
art---
With an s we compute
n = 1

a) =x=o
So by the Ratio Tosts

St is absolutely convergent -

Eg Let xeIR ,
consider
/G * *
1
=
x
-

* E-
+
+ -
-

u =

For an=(y
t we compute

/11m =
*
18)
x
+
In x ,
=

Ix
=

So by the Ratio Test , E HH *** absolutely

,
is

if Ix21 and if (x)> 1

convergent divergent .

E x
=

1 ,
 -* converges by
tost
the alternating sories

IP x
=

-1 s

Est =-

* diverges .

know from last line

We
(a) E * diverges
(b) E converges
,
.

The Ratio test us

gives

calling) ny1
(d)
= (n)" Fo I
The Root Fost Let Ean be a series .

I If
.

anl=L1 ,
the
can
is absolutely
convergent .
 IP
2
tanl=1 thou
diverges
.

or
,
an .

m
&
b Just like with the Ratio Yosts i

he al=1 ,

you can't doduce anything .

consider
Say Xe ,

"( & x
n 1
=

For <" hour

an x ,
wo

** 1x/x =

absolutely convergent
So by the Root Tost, &x(m)" 1
is

u
=

(x iP Ix) I
P and
divergent >
.

Far x = * 1 , the Root Test tells us nothing .

Take x
=

1 s

E() diverges by the Divergence Test since

(n= Ines (1 ]* ~
-
I
lin =

H- D

(by a Prov Jala course and /or an oracle

If x =
-

1 ,
we howe

, "()" is tricking e
 Lecture 3
& Often hard to
compute the value of a
convergen
E An
n =
0

But we con
estimate
s
= E an
n 0
=

by its
partial W
suns

SN
=

Ear
this estimate ?
How
good is

Set Rr= S-SN

Tost Estimate
Integral Let be a continuous ,

positives decreasing function on [0 1) such that

,

Sot an= a d can dreges konverge e

an
,

Sf(xdX = ( f(xdx
 Picture of test
Aside
integral

-
--Go

ar_an
_

ar

f(x)

01 2 3 4 5

Alternating Sosies Test Estimate

Let
ot'bu be a series with busO ,
bux, bu ,

and bot 0 Iso Etsbu converges by

the alt

Ser test) .
Than
(Rn =

1s 3w) -

-
bN x)
Find the value of
E
1 an
n
=

to & decimal places

up
-

Note the
alternating series test and estimate
apply since

A
hims antis
A = and =

0
an(nx2) angnes)
Want (Rul=IS-Suls Yu *
(N+2) to be small .
If N
=

3
, we have Scratch work

Sw =

1 8 86 * N bN +1

=0 88182)
O !2 0 123
=

=
.

auch I * = =8 821 .

(Rs)= 15-selsbr 2
=
so O Go
.

& 0 00018
3
E
.

subtracting
so

Adding or

=0 000/8
0 00018
.

does not the first S 1 =+

change -6
-

two drei mal places in

- 6 881821
841821s
.

0 .

so

" 0 88
=

S s
=
.

Dowis Spring

L A is series of the form

power series a

& CX =

(
+

CX + Cex+---
n 0
=

where we view X as variable The Ins are called .

the compficients of the serves More generally .

,
We will also consider

(n(x a) -

n
=

for fixed real number a call it series

power
a
,

controd at a.

Rank A series for

may
Dx for s
owes I some

values of and
diverge

y xe converges
-
o

(What about x
= *
1 ?) (Lectwob)

Et, converges
for all XeIR Chacture 2)

for what values of

E XER does ,

S n
converge
?
S
the matic test Set Then
We
apply .
an .

(x 3)n+
= 3)xx) 3)(ii)
-

⑰ =((x -
=

(x -

(x- 374
T - (x-31
Ratio tost iIX-311 , 2x<
converg
out i . e .

is 1x-31>I i x <2 x= !
and
divergent ,
.. or

What about X=2 4 , .

If x
=

2 , we get , convergent
by Alt series test -

x 4
get E, diergart
=

, we .

So * (x -
3)
2 :
n =
1 i converges =x

and diverges otherwise .

 Lecture #3 09/07/2023

For which x does the series ono(x-1" converge?

we will use the roof test

an =
n"(x 1) -

ranl :

"Wex-in) =

n .
/x-11

divergel

I
if x #1 then "Fanl me to as
,

so
& n (x-1)"
diverges
n 0

diverse feelept
=

converges of
n+ O
if x
=
1
,
then "Fanl =

n .
0 =
0 -
to O "1"

38
& no(x-1)n converges
n =
0

Theorem

Sanx-al
For a power series
,
there are 5 possibilities

1) the series only when x=a of is O

converges -> radius
convergence .

2) it
converges for all possible values of X EIR - radius of is
convergence

5) there is a positive number R such that the levies converges

if x is closer to than R and it it is further
a
diverges x

from a than R .

Remark :

In case 5 , there are a possibilities on whether the series

converges or
diverges at a R

Def in the 3rd case

,
we call the number R the

radius of in the series

convergence .
 The interval of
convergence

is the let of all xEIR for which the series

converges .

[ I

C ⑧ S
a a R
a R
+
-

Recap of examples

Radius of Integral of
series convergence
convergence

Enr(x-D O 21 13 ,
=

[13

& C (
-
-
,
c) =

iR
n =
1 X !

* n
3
I (2 1) ,
n 1
=

Ex D (-1 1) ,
 When a ceries
Cux" has a positive

radius of it defines function f(x)

convergence ,
then a

f(x) =

Ex
defined the interval of
on
convergence

-Sometimes ,
we have a
simpler formula for f(x)

:
for xt (-1 1)
,

if then we that the series is a

Cux
=

f(x) say

power series representation of

fix) on I

example
Find a power series representation of fix - Fx

-is
and its interval of convergence

= 1- xe

= -

+x +

&x+ ...

it is
convergent if and only if - - x <2

1x1< E

the interval of
convergence is S-E E ,
 Theorem if the power series Cr(x-al

has radius of R>O then the function

convergence

-
f(x1 E(n(x-a) a) a)
=
= (0 -
(x - +
((x - +
((x -
)

is differential (hence continuous) on (a-R ,

a +
R) and

changes
1) f(x) =

,
n-cn(X -
a) - =

n =
0
(n 1)(n + (x
+
-

a) originall new

( 2((X a) 3(z(X a)2

=

+ + - +
,
-
...

ne
we removed the
Co

2) (f(x1dX =
2 +
(o(x -a) + (x a)-
+ (x -
a) +

...

=+ (x-ant
on lame open
interval
-

The series 1) and 2) are

convergent on (a-R ,
a +
R)

Remark In about the end points

general we cannot say anything
:

in the above theorem .

example gix x
=

1 +
x +
x+ ...

what is
-
suggested

f((x)
=

Fxx =

1 +

2x +
3x2+ ...

=En .
x -
n
=
0
(n 1)x+

convention /

Example
f(x)
=

&
n 0
=
0!

xo
=

=
1
1

So
i
fix) :

f(0) =
1

ex
f(x)
=

↳ Because it has all properties

 Taylor Series

Q1 :

When doel a function have a power series representation ?

Q2 :

How to find the power series representation of a

given function ?

Suppose that a function

f that is represented by a power series .

fix)
E
((X-a)" for
=
x

I interval

How can we determine (nS ?

f(x) =
(0 +

(, (x a)- +

(z)X -

a)2 +
(3(X a)3+ -

...

Co
f(a)
=

f((x) (, 2(z(X a) (3(X a)2+

=
+
- -
+ ...

C1 =

f'(a)

8"(x) =
2) +
2 .

3(z(X -
a) +
...

in
general
:
a

=a In =

Ma
because we get
f(n(x)
=
a f(x)
2/2

the with derivative .

13
Nial
=
By convention

0! =
1

infinitely

Lef Let o be a function differentiable at a a IR .

The taylor series off at a is the power seriel

Mal Taylor series f

(x-al of centered at a

at a 0
=

,
then it is called the Maclauren Series of f .

for K >0 an Integer

Tr (x)
= cal(x-a =

f(a)
+

fi(a)(X
-
a) +

(a)(x-a)2 + ...
+a) (x -
a

is called the off

degree K taylor polynomial at a

Example
Consider ex
f(x) =

& f(x) =
ex =

f(x) ,
for any n, 0

a f(x) =
ex

so , the Maclauren Series of ex

f(x)
=

It! n
xn T01x)= 1

n
=

0 T, (x) =

1 +
X

and the first polynomial T2(x) =

1 +
x +
T3(x) 1 =
+
x +
- (x) =

1 +
x +

j
x
+
 Spicture)

Write
Rx(x)
f(x) Tr(x)
=
-

it is called the K-th remainder

of the Taylor series of f .

iffix Mcal-alr the

Remark :

on

Theorem :

Let Tk(x
f(x) Rx(x)
= +

Ik degree of
Taylor polynomial fat
is the K a

If the him RxCx1=0 for Ix-al<R for some R>O ,

D
,
K -

then
fix-falix-an ,
for Ix-aR
 K+ A
How do we show R2(x) -> 0 ?

Taylor Inequality

if
If(k+ (x)) = ,
for some constant M>0 for all x + [a-r ,
a +
r] for

some rs 0 .
Then the ,
remainder rxxl satisfies
Irx Ix-alle

for all Ix-aler

Example :

How well does To approximate f(x)

=

ex on C-1 17 ,

T =

1 +
x +

e +

I +
We want to bond R1(X) =

ex-T11x) on C-1 13
,

f(x) exf()(x)
=
=
ex

so levie for xe [1 1 ,

The Taylor inequality :

for IXI1 We have

1R(x)) < ? 1x15 = o

=
0 023.

Let's show that ex= it *

for all xEIR

we showed earlier (the ratio test) that

it is absolutely
for all xtIR
convergent

In particular for all xEIR

him I =
o (*)
 to
we will
apply Taylor's inequality f(x) ex
=

for >, O

f(r(x) = ex

fexs vsso we will look at [-r r] ,

If "(x))
( *+
ex er for +[ r]
=

= x
-
r
,

By Taylor's Inequality ,

**
IX
1Rx(x)) - Y :

er
o
fix x + [ -r r],
1Rk(x1) =

Thus , ex= it x for all x = [-r r],

Since r was arbitrary ,

ex=
Eit x
for all KEIR
example Do Sincy)
,
CO11X1 have Taylor series representations around O?

let f(0) 0
=

f(x) sincxl
=

f((x) =
20/(X)
f((0) =
1

g"(x) Sincx f "(0) 0

=

= -

8"(X) (0)(X) 8 "(0) -1

= -
=

0
18 Sincx) ""
(0)
(x)
8
=

So the Taylor series of Sincel at 0

x -

5,
x3 +

5, x+ ...
=

i x Entire since

/g
***
Since
f(K) is either I Sincel or Icosixl (x)) =)

Taylor's inequality gives us IRkcx1l < o

we conclude

sincer-
xant
ai

Similar
argument gives
:

cos(X)
= x

=1 -

= + + ...

We can also see this as follows :

-
x*
=
n
1- 11 -xzn
E Int
.

cosixl =
Sincx) =

n =
0 (2011 !

-i xan
Lecture #5 110912023

Let f(x)
=

sincx) For example (sin)

T , (x) =
X Tz(x) =

X if IX) _ =
28 6 .

Isin(x) j (213
Since
f(x)=-c0s1X) x 1 0 02
- =

= . = .

Ifsxl for all xER

For example (101)

By Taylor's inequality
1 -

2 =
T2(x) =

Ts(x)

(sin(x1 -

x1 =

(R > (x)) >

-
5! /X3

By Taylor's Inequality

(R3(x)) =
1 ! Ix14

In particular for (xI = I

(201(x1 -
(1 -

1 =
32 = 0 :0016
 Example staylor expansion
S
**

In (1 +

x) =
x
for IXIC
of the logarithm
looks like the

=x -

2 1- +
+ ...
-> alternating series

To see this recall that

,

#x
a
In (x+ 11
plugin -x( x)
=

-x x
-
=
+

di
for /C I

* Recall! so
-
.

-In x
I
-

xn
n 0
=

to find the
value of the

constanti Plug

In(1xx-faxanegral arti t
,
r

the value
Of "y

the x" for IXI < 1

n

C =
(n(1 0)+
=

(n(1) =
0

So Inc1+x)= x for IXK1

,

*Remark :

Substituting X-1 for x above we

get

Incx1= x-112 for IX-11 <1

Recall :

b>n

(B) =

#
ways to an -element

sublet from b elements

(2) =

# ways 6 choose 2 elements

from 31 ,
2
,
3 ,
13 =
6

(b -n+1)
(B) n b-n b (b-1) ....
.

=
=

!
n!

=3 1
3
(2) 6
=
=
=

The binomial theorem s for bil

integerl

(1 +

x) * =
(B) -
x

Now , consider an
arbitrary b and :

x) B IXI 1
f(x) (1
=
+

f(x) (1 x)b f(0) 1

=
= +

- b
f((x) b(1 x)b f((0)
=
=
+

-2
f"(x) b(b 1)(1 x)b
f"(0) b(b 1)
=
-
=
+ -

g((x) b(b 1) (b 1)(( x)b

f((0) b(b 1) (b 1)
-

n + ... n
= - -
-

...
- - + +
-
=
. .

(b
((0) b(p-1) 1)
=
..... -
nt

Notation :

for every real b

(2) =
b(b 1) -
.
. ... (b -
n +
1)
(8) = 1
n!
 ben
Note if b n then this coincides the
is an
integer with previous
:

notation .

if b is an
integer (n>b1

then (h) = 0

So , the Taylor Series of CI+X* is

( x

⑰ We will justify that CI+x=

Go (B1x /xI < 1

1) Apply the ratio test to the Taylor series) to see that it is abs ·

convergent for 1x1<1)

I
b(-1 x (b .. - ti)(b-nxx *

(a
(n 111
/x) /X
.! =
+
=
=

=
x
by-11 .. X (b An+11.

x
"

for
so ,
we
get absolutely convergent IXI1

Now we need to justify equality in

Consider (B)x IXI1

gix
:
:

1)
&b(b 1)
n

-(4) xn
-

(a)nxn
+

re
gix =
-

=
+
=

n = 1

- b(b-1) ..... (b n) -

I x
=

n!
n 1
look at :
=

b(b- ) ..... (b -n + 1)
x
x-g(x) =

(n-11 !

b(b-1) ..... (b n) -
b ..... (b -
n +
1)
So (1+x) x x
gix) =

g(x) +x-g(X) =
n !
+
(n-11 !
continued)
b(b-11 .... (b
I
-n + 1)
x
.

1) (b) t
=b +

E [b(b- :...
,
(n-11 !

bIn
-

1)(=
b(b 1) (b n
1)
- - +
...

b +
= +

(n-1) !

b(b ..... (n -
n+

b
-

y b
b g(x)
=
.
=

!
+

so ,

gi(x) =
b Y
Consider (1 x)
b

y
-

h(x1 gix)
=
.
+
=

b
hisx) b(1xx) bg((x)
- -

b)1 x)
-

g(x)
= -
+
+

b b +

b(1
-

g(x) = b)( x)
-

x)
=
-
-

g(x)
+
+

O
So n(x) =

constant =

h301 =
(H01-3 .

g(x) = /

10 / 1 + x)
*
for IXK1
g(x)
=
Special cale :
b =

I * Remark :
It
might happen that
f is an infinitely
differentiable function
,
its Taylor Series
x =
(lx
↳
n
=

0
gir ,
sol xa
e

converges on come interval but

fix+80) (x-

example n if x 0 on this interval

g,x
=

G8 if x
=
0

f("(0) = 0

we can show that I is int diff. at O and

I So its

is O
Taylor levies

< ↳! 0x =

0 +
0+ ... o

but , 8 is
not
constant O
 Lecture #6 09/19/2023

Remark In if
:

general ,

fixlax gix
=

dax

for IXKR

then

f g(x)
.
=

((0 +
x, x +
xx +
... ) )dot
.

d, x +
dex2+ ... )

Codo+(Cod +Cido(x (Codi ( ,d, +(2do) <

=

,
+
+
x + ...

is a power series that

converges for IXIR

Example :

find the first three non-zero terms of the power series

of sinxl ·

(OS(X)

sincx) -

COS(X)
=

(x - 5 + +
...
) .
(1 -
+ -

... )

=x
+

) I -

i)x +

(a! 25 +

!
+

5! )x5+ ...

x
-
5! x3 +
x5+ ...
Recap on power series
lof applications)

1) to compute limits of expressions involving trigonal functions :

sincx)
lim Note that :
x- 0(x)
sincxl =
x
-
+ +
...

m (1-! )
x

xlin = + ...

x
=

1 -

- + + ...

=
1 +
0

x2
2) f(x) =

e
-

* this function is important in Stats

, probability Signal processing
, ,
etc .

There is no closed formula for

Sexax
using Taylor Series

-Bone ... *
e

SO ,

Sexax =
C +
on xant

so

i so iiit is anantaerivate
or it
 Vectors & Geometry

48P
Let P = IR3
·
2e
Y
p =

(x y ,
,
z)

The vector from O to P

!
-
o is the vector
(2) = - All vectors are
starting at the
origin

we can add vectors

i =

(a ,
b ,
c) =

(x , y ,
z) Y+ =
(a +
x
,
b +
y ,
c +

z)

also vector by
we can
multiply a a number

I =

(a ,
b c) & E IR
,

2 .
T =

(2-a ,
a -
b
,
a. c)

of the vector
we can also compute the length
F (a ,
=
b ,
c)
III =

Nc +

Ate-11a .

El =

181 .
1151

11511 =

0 if and only if I =

*
-

if F 0 =
# /unitrector) - Ill = 1
The standard balis vectors

=(i) al
mernizens
-

"
(i) =

xi +

y + c

Bef
:

Given two vectors Their dot product : - a number

I . F =

x , xc +

x , x2
+

21 -

22

r
(ii) w (ii)
=
=

fasy to check
:
-Y .
= IVIR
-
.

( F +
=
Y. + is .

-I . =
i it .

if CER (C . ) . =
c .

( -)
,

Theorem :

En
if & the between I and I
is angle ,

then : .
= Ilrll .

11 WIl 2018

By the law of cosines

i
I- ↑
*

-El =

11511 : 11 W-2 /EIIEI ·

CO18

on the other hand

↑
i
⑦ 11 - 12 =
(w-r) (w . ) .
=
w .
-
. - . +
F I .

=NIl2 +112-2 m .

so , i .
E = 11 11 11511-c01(8)
·
M
 Special cale if & =
E then cost =
0

w I .
=

for non-zero vectors I I

,

.I 0 if and only if =
E only way the expression

dir
=

would be =

and only if i and I

are
orthogonal/perpendicularl

Example

(E) :s
orenagonal
to
(i) (i) ()= =
0
=2
+
- -

⑧
21

use the dot product to determine the projection

will
we

of I onto it

# n
prin
projec =

c .
i
= ity in

1
=
(0)(8)

lengthIcolored)
of interval

c =
11511-(01(8)

ThVI
Cost not
,

III -
=
-
by the previous theorem

the

-
#C (can

is called the
be 1)
scallar proje
↳ formula for project /I outo )
denoted by
competit is the
signed length of the projection
Example

Let us find the

projection

proje
for ↑
=

(1 ,
1 ,
2) w =

( -
2
,
3 , 1)

F F
-
=

(1 ,
1 ,
2) .

( -

2 , 3 1)
,
=
-
2 +
3 +
2 =
3

11 W(12 =
1 212
-

+ 32 + 12 =

11
proje
= 1 -2 , 3 , 11 =

(2 ,, )
Lecture #7 09/21/2023

Hef
Let r =

(a ,92 as) =

(b ,, b bs) both EIR3

, ,

The cross-product of i and i

formula
Y x
=

(acbs -

Azbz ,
Azb , -a , bs ,
aibz- acb , )

-dit
(at Tae)-saubs-asbil-labs-asbij) + labe-anb , E)=

* Remark this only makes sense in IRS

:

in IR" ,
we can take the "cross-product" of n-l vectors .

Example
& =

(1 ,
2
,
3) w =

(2 ,
-1
,
1)

- x =
det
(iii) :

=(21 -
31 -

11) + -
(1 -

1 -

32)j +

(1) -
1) -
2 .
2)
54 58
58 15 , 5 , 5)
= + - =
Properties

110x ( w) +
=
ex + x
and

( w) x +
=
rxu +
x

2) (Cr) x
=

<(0xw)
and

↓x((w) =

c(rx =)

3) Ex = -
wxe in particular &x=-Ext so , x =
8

4) Y (Ex)
.
=

0 and 5 .
(*x) =
0

proof of 1) let

i =

(a as) (b ,, b bs)
, as
=

, ,

I (0x)
. =

(a 92 as) .

(Acbs -

Aske Asb , a, by a be a bil

, , , ,

=A , acbs / -

abs + ab , -

/by
aca , +

asabe I ab , =
O

The second part follows from the first one

5) if & is the angle between I and I

in
L 115x wil 1IVI /I WII Sinc
=

· ·

ol
, I

in particular in I and I are parallel ,

then 8 0
=

11x1 Ex

5
=
0 So =
0
,

L .

11 all sing-area of parallelogram

of if x 0
,
then the direction of Exh is
given by the

hand rule
right .

I in the plane of this page

Ex is pointing up
Ext is down
pointing .

1 , 5 , 6) together uniquely characterize ix i

Lines & Planes

say that p: (Xo 40 20)

, ,
I =

(a , b c) ,

The P in the direction I

line
passing through

Can be represented

f(+1 =

p +
+ I .

=(x0 +
at
, yo + bt
,
20 +

ct)

in coordinates x(t) =

x0 +
at # Remark
:

two lines filt=p+ tr

y(t)
=

yo
+
bt f2(+) =
Pc +
+ E
z(t) =
20 +
(t are parallel if I and i are

parallel /non-zero multiples

of each other) .

Parametrizing planes

The plane through P :

/Xo , y ,
20) with normal vector
n (a =

,
b, c)

/ a
9
is the let of all &EIR3 (9-) .
5 =

.....
ie if q =

(x y , z)
....
,

#
coordinates
- -

- (X-X0 , y-y0 z-20) (a b c) 0

(origin)
=
·

, ,
,
·-

a(x -x0) +

b(y y0) - +

((z -20) =

Or

ax +
by +

(2 =

d =
axo +
bxo +

(20
important !!!

The
plane given by equation ax+ by + (2
=
d

has 5 =
<a ,
b ,
c) as a normal vector .

Example
Let Us find the equation of the
plane containing 3 points

P =

(1 ,
3 2) G (3=

,
-1
,
6) R
=

(5 ,
2 , 01
,

R
*.
·

-
this is the unique plane
the free
containing
..
vectors
+

->
O Q &,
P PR

with a normal vector

->

n
=

Pa x PR

i =

(2 ,
-
1
,
1)x(d ,
-
1
,
-

2)
=(12 , 20 11) ,

so the plane is :

(12 , 20 19),
.

(X -1 ,
y 3,
- z -

2) =
0

12(x 1) -
+

20(y 1) -
+

11(2 9) - =
0
Example
Let us find the distance from the
point p=(1 ,
2 ,
1)
to the plane
· P x 1
y 2
=

.
+ +

I A normal vector to

-
I

! the plane & =

<1 ,
1 , 11

Q
·

and the plane pallel through G :

(1 , 0 , 01

The distance from P to the plane is

the
length of the projection of P to

compP1= /10
Il proje II I /
Onl= 2 1) (1 1,131
.
=
,
, ,

11 (1 ,
1 ,
1) 11

= 10 -
1 +
2 1 . +
11) =
3 = 55
r3 5
Lecture 8 09/26/2023

Example x
2 +

y2 =
1

is a circle of radius 1 centered at 10 , 01

E
g
.
.

x y2 22 1
=

+ +

defines the sphere of radius

........
....

I centered at 10 0, 01 ⑧

-
,

visualized

Generally to sketch a surface in IRB it is ,

useful to look at its cross-sections/travel

by setting x , y ,
2 to concrete numbers
E g
.
. Sketch
y

x2 z2 0 I2
y2
=

+ +

we start by looking at 2-traces

I I

2 0 -> x2 2 0 X
y
=
= +

2 =
1 + x2 +
y2
=
1

2 =
-
1 -> x +

y
=
1

2 = = 2 + x2 +
y
=
1

Z
2

y-traces 0 x 22 0
=

y
=

O
x
=

= 2

I I 2
1 22 0
y x
=
+
-
=

22 =

x +
1

2 = =
V1 +
X

yI 2
=

22 =
x
2 + 1

2
=

IN1

X-traces ,
same as y-traces 2

- .-
..

I
-
Y
.......,
-
- -...
-

givel 2 cones
X
....
-

.......
-
~ ...
~
-

-
 E g
.
.
Sketch 2 =
x +

2- traces
Y

2
2 0 x
2
y 2 2 x y2 2
=
=
=
= -

y I
=
X x = r
x 2 +

2 =

1y 2 =
x2 1 -
2 -
= 1 x2 y 2 -
= -
1 X

2
x2 x2 1
y x
2
1
=
= +
+

x
=

Y1 + y
-
=

2 = -
2x2 yz - = - 2

I r
x2 2
y
=
+

y-traces

0 2 x2
y
=
=

112 x2 1
=

y
-
=

22 x2
y
I
=
= -
1

X-tracel

x
=

02 =
-

x2

12 1 y2
-

x
= = -

x =
=

22 =
2
-

x2
-
2

Hi
It saudle
y
 vector-valued functions

parametric curves in IRS

Def F :

IR -> IR3 /IRTIR" or IRT /R

Y
)
is called a vector-value function

Can be written as

F( + ) =

(x) +
) ,
y) +
1
,
z) +

1)
=x( + 1 . +
y( 1 + .

j + z(+ 1 .
E

it defines the parametric curve Grit) :

tt1RY =

[(x1+ , x1t) ,
21 +) : +

+R)
f g
. .

ri +) =

(10) (+ ) -(in(+1) * Remark :

Parametric curves

have a natural direction

given by increasing
values of t

parametric
we can re-parametrice them for
so the corresponding
curve in this case is the circle ·
instance by Fi(+) r (-H Changing the
=

direction .

E g .
.
H= (c0 , (2+) ,
Sink2t)
,
E .

g . (H =

(cr1(+) ,
sinc+1 ,
+
Also has the circle as

d
its parametric curve the parametric
curve is helix

t g .
. Y(+ = <x1) ,
x(+) ,
2( )
+
x(+) =

2 +
(0)(EH)-(011 + 1
x1 1
=

2 +
(0) (2+)-(inct)
(Et1

⑧
21+ 1 =

sin
 Def For F /R-IR3
:
the limit of FCH at a tIR
,

him rsts =

(1im xst
,
him xst ,
limest
if it exists , I is continuous a if
flal He st
=

at
,

Def The derivative of I :

IRMIRY at at1R / if it exists) ,

is Psal =

himh(r(a +
4) -

r(a))
....a

~
- ------

+
* Note that :

ric+= (x '( +
,
y'x + ,
c +
1)

(H =

/im (8) + n) +
-
F(+1)
n+0

I
-n
-
- lim n(x) + n) -x1H y(t h) x|t) n(+
+

,
+ -

,
+
h) -
1)
x+
neo

=(x'(t) ,
y't) ,
z(+ 1)
t .

9 .

F(+) =

(1011+ ) ,
Sinc
,
+(
& (t =

(-linc) ,
10111 ,
1 (

Def if F : IRTIRS IS differentiable and (to) FO

.

then the
tangent vector to the parametric curve

I at t is I'Stol

The
tangent line at to the unit
tangent vector at to

rstol t 'Stal
vi all
+ .
Lecture 10 10/03/2023

if vst) =
(x(t) ,
y(t) ,
z(+ 1) then the
,

arclength between and b :

a

1 =

la x 'S +

x' 2'(2 dt +

example Compute the length of the full turn

of the helix

↑ (1) =

(c) (t)
,
sinc ,
t

&
r')
=

(-sinCt) ,
105(7) ,
1)

we compute full turn /0 2π)

,

2π

! .
(-sinst)) + /CO11H"+1 &t

= 111 at
=

2πE

* Remark :

if the curve overlaps itself then the overlap is

,

counted multiple times .

&

(1) =
(10112+ sinc2+1)

!*
,

*
**
115'still at =

( esin12+1 ,
<cos12+1) :

( in t)+d10s CHAA

! "Adt =2 .

CT = &H =
2 .

/circumference of a circlel
* Remark :
Sometimes
I x '(+2 y ' ( 1
+
+ +
2'(+12 At

can be difficult to compute

if x '(+2 +
y'C+2 +
2'72 is not a square
=

Difficult

Eg find of an ellipse

b
(a) +

(*) =
1
a

we can use the parametrization

r(+) =

(acost ,
brin1+1) +t [0 24] ,

rilt) =

)- asinIt ,
boist1)

!* asin(t) + b 2 10s2 (H)

↳
dt
this does not have a "closed form"
if a b

it is called an elliptic integral .

Def
The function function I
arclength of vector valued

sit :

" I swill de
Harene* Your teaching Time +
"

This function measure the distance along the parametric curve CH

from F(0) to F(H) ·

 Example
& (H =

(10 1 11) sinc +

, ,

r'(t)
=

1-sinst) ,
1091+)
,
1)

11t
=) H15's will du

Edu =
R A .

Along the curve we have

travelled E .

Observation :
S'1 =
115'(+) /l

Def We say that CH is parametrized by arclength

if IIr'(Hll =

Note that this mean s...

11t
=) Fr swilldu =1 du =

Sometimes it may be difficult to find the arclength parametrication of a curve .

S =

S(t) if we can write += tCs Altis the inverse of 1) *

(c) r(+ (s)) the F.

r, is
parametrication of
=

arclength
because :

11 r, xsll =
1r( +(11) -
+(11)

Sa = /I still -= +(l)
=Il rs+ 1) . ll

=( * ( .
11 +

(1) I =
1

M why it is = 1
 Example find
arclength parametrication of the helix

*
Recall

8(+ =

(10) (1) sinct) t) S1H = Et .

, ,

solve for t =

* ->
writting
+
as a function of 1)

is+ (1) =

5 =

(cos( ) ,
sin (E) ,
(1)

Curvature

.
S
>
=> Unit tangent vectors

· · D
* around ob
,
the unit tangent vectors are changing direction faster than at oa

↑
-
Recall that if r : RMIR
,

↑ (+
the unit tangent vector TH) =

115'st11l

·b the curve is more curved then 1IT'still

=

speed at which T'is changing .

would be called flat

· a
Higher dimensions ,
curve

The bigger the speed of change -

more curved

Def if I :
IR-IR is parametriced by the

arclength ,
then the curvature

k(s) =
11 T'xxill =

/I l
What if is by
our curve not parametrized arclength?

[(+)
we can compute S(H
,
T()

- rature antcomputeee
k(t) =
 Lecture II 10/05/2023

t g
.
. Let 5 :

(acos(t) ,
asinc+1)

v'st) =

(- asin/t ,
acos(t)) a

IIr'll= acsiniC+) ,
accos"(H) =
a

T1 = = (-since (01(+1) ,

T'(H =
( - 101(H1 ,
-sin+1)

1IT'(tll =

COSH +(in'(+)= 1

k(+) =
a Does not depent on I

* smaller the radius , curvature

larger the
 E g.
.
Consider the line

9 =

(x0 y0 , ,
20) +
t -

(a ,
b ,
c)

f'st) =

(a ,
b ,
c) 11 r'(+1) =
a2 +
b2 +
x

(a , b, c)
T(+) =

a +
b +
(2 T'( =

(0 0 01
, ,

11 T'( +) 11 =

k(H) =
0 for all to

Theorem :
if : RAIR is a
doubly differentiable
vector valued function .

k(+)= Ir (H) x r "( +) ll

11 v'( +) 113

Proof : we will omit (t) for simplicity of notation

T = SO ↓
=

11r'll .
T =
S'T

S-arclength function of our curve

SO r
=

S"T +
S'T'

r'xr" =
((T) x (s"T +
(T)
=(ST) x(s"t) +
(s' +)x(s'T') vector croll-product of itself =
8

-"I
=0 +
1S'lTxT'
=

I'll TxT' .

Angle between
T & T T & T
=1 al are
orthogonal

Ilr'x roll= I'll

-
I T'll I'l HTl It'll ⑨ Number
x
sin(o)
. =
.
. .

110112 .
1 .
11 I'll .

The latter follows from the fed that l1T11=1

 III'll
Recap : Il r'Il

Ilrxrll= I l .
Il T'll=lr' .
E
11 r'x r"ll
k
- end of proof
=
11 'll3

E .

g .

Let us find the curvature of

↑( +) =

(+ +
2 +

3)
,
,

ri(t) = (1 ,
2+ ,
3 +
2)
r"(t) =

(0 ,
2
,
6 t)

11r(t)11 =
12 +
9+ 2 +
9 +2

I
I

I
i k 2)i (6 + 08 (2-0)E
2
r(t) x r"(t) = =

(2 + -
61 -
3 + .
-
-
+

det 1 27 37 2 =

6t2Y -

6t +
2k
O 26t

118'() x = "(t) 11 =
36+9 +
36+ 2 +
1

k( +) = 36 + 1 +
36 +
2 +
1

19+1 +
1 +
2 +

1 13
 A special is for
:

case a plane curve

y
=

f(x)

we view this al a 3D curve with z =

0 and define
,

[(x) =

(x ,
f(x) ,
0)

r(x) (1 f(x) 01
n
=

, ,

r"(X) =

(0 , f"(x) ,
01
Ilr'(XIII =
1 +

f11X12

af o)
↑ (x) x "(X) = 10-07-10-018 + (f(x)- 01E
=f "(x)E

If "(x1
3
ir(x1 x = "(xiIl= If " (x1) SO , k(x) =

1 +

f(X)
 E The curvature of a parabola (y 2
g x
=
.

at point X

2
3
k(x) =
1 +

1x2 depends on x /curvature I everywhere

the max is at x 0 =

,
it equall to 2 k(01

TNB frame

r'( +)
↑ :

IRT IRS T(+) =

11 r'(r)1)

11 T H+D1) = 1 /O T T .
=
0

Def The principal unit normal vector Junit normal rector

Ns+ =

Fin
The binormal rector

& (t =
F( + 1 xN(+)
&"

Remark
-
FCt
vectors
,
Ni
orthogonal
,
BC
to
are

each
unit

other
~
t 9
.
↑(1) =
/(01(+) ,
sinc ,
H

v'( =
1-cin1t ,
101111 14
,

IIr'(H)ll =
Sin +10P(+1
=

F1H =

( sinc I cost I
, ,

Y'st) ) =
-
cuit)
,
-cinit ,
of

117'sIll= col"Ct + cint +02 =

so .

NCH =
= crit ,
since
,
of

- (-2011) ,
sinct ,
ol

af
Si-Faxn-

= ( sinst
- cust .
 Interpretations

we think of r :
IR+IRs as a motion of an object in IR

FC+ is its position at time +

-
vi+= r(t) is its velocity
-

act) = r" (H) is its acceleration

we have the scalars versions of the above

11H) :
distance travelled

S1H=1" llr'swill du arclength or

distance travelled

sit) =
(Ir'still speed

s"() :
SHH) Scalar acceleration

Remark
*

1
1) speed 11 velocity
=

21 "H) be 11ast)))
might not
=

S"(t =

aq 11 I still

11 a it11) =

11 still
Theorem r: IR-IR3 Vector valued function

for which I is doubly differentiable

& Ht) =

S "(t)4( +) +
k(+) -

/ +2 . N(t)

proof we umit t for simplicity of notation

I =
i = I
so r =

S' . F
fil
↳ derive to
get

a =
&" =
S . F + S .
F
11 I'll
= S"F +
S'll +'ll . N since = . k(t) =
II 'll
=

S . + (s') -

k . =11 I'll

Is' ⑪

End of proof
 12/10/2023
Lecture 12

Helix
Example : Me
r( +) =
(10)) +
)
,
(int) ,
+)

rist) =
)-cinst) ,
1013+1 1) I'll =

E
,
↓
recall from last

TH =

/* ** (IA)(

NIH =

(-10S1H) -sinct) 01 these two

2
a
, ,

are equal
alt) =

v"(t) =

( -

x01(+) ,
-
(int) ,
0) =

N(t)

This means
:

s"(t) = 0 and (s'(+1) " kH+ = 1

3
by the theorem discussed last class .

* Note that S'CH=Ir's)/l =

E (s' does not
depend on +

so indeed S"(t) =

0
,

Le Derivative of a scalar function

&
2
(S ' ( +11 =

=
2

so ,

k(+ =

We can allo compute k(+) directly since :

ki + -

!" : =
I
 functions of several variables

Given a function f(x y) ,

its domain the let of ex

is all , y) on the plane

such that f(x y) ,

makes sense

E .

g
f(x , y) =
1 -

x2 -

yz
+

m(y +
I)

we need

1 x2 y 20

I
·
-
-
x +

y
2 = 1

y
+

I >0 y > -

domain
:

all (x y ,
:
x +

y
=
1 "IIII) :
Domain

"IIIIIIn
y >
-

The of fix , y) is a surface

graph
in IR3 given by
we can represent such
f(x y) =
z
level curves
,
Surfaces using their .

#/ A level curve of fix , y) is a curve

f(x , 4) =
C for some (

A collection of level curves is called

a ccuntan map .
E .

g
Let's /ketch a contour map of

f(x y) ,
=

Ny M :
domain

M
Domain :

(We need)

x 0
Y ,
-

x >, 0 and y > 0

Or

M
x= 0 and Y= 0

contour map :

1 2
=

first ,
lets consider c
=

C = 1

f(x , y)
=

0 =

Xy
1 =

f(x y) ,
=
1 =

My -
y =
I

Xy G
f(x
-
y) 2 y
=
=
=

We can also consider function IRS+IR

F(x , y ,
z)
IRC+
or more
generally IR
:

we can represent them by level surfaces

(x , y , 2) f(x y z) C
: =

, ,
E .

g .

f(x , y , 2)
=
1000 -
x2 -

42 - z

f(x y 2) 0 - x2 + y 2 22 1000

-
+
= =

, ,

f(x y , 2) ,
=
500 - x2 +
y2 +
22
=

300 ·
-

f(x , y , z)
=
1000 - x2 +

y2 +
22 =
0

↳ No solutions for 1x1000

 Limits & Continuity

Recall

f :
IR IR a tIR ,
LEIR

Lim f(x) = L
x+a

means for every 370 there is >0

,

If(x)-21 < E Whenever /x-al > &

~
BAD

~
EL = &
O

-
-
⑧

a
I
a

Def . Let F :
IRTIR and (a b) GIR2
,

L E IR

we lay that the limit of f(x y) , as (3)

-
. . . . . .

......
. . . . .

1x , y) approaches (a , b) is L
~

lim f(X , 4) =
L
x , y4(a , b)

I
if for every 230
,
there exists a 830 (a , b)
,

such that (f(x y)-2/<E

,
whenever

(x -al +

(y - b) < &