Lecture 02

IIT Kanpur July 201
20 followers
Jul 19, 2024
[Music]
welcome uh let's recall what we did in
the last lecture we are discussing about
convergence of
series now we discussed about some of
the the tests which involves uh the
where which will enable us to determine
whether the series is convergent or not
so namely all the series which consist
of the non- negative
terms then we
have basically three test one is the
comparison test which is extremely
powerful and if we have two series a n
and BN both of a n and B n they are non-
negative and BN is greater or equal to a
n or a n is lesser equal to some
constant time BN then if BN converges
then a n is going to converge similarly
if a n diverges then BN is going to
diverge summation of BN is going to
diverge and uh as a consequence we can
derive the root test test as well as the
ratio test which we are familiar with so
now let's talk about
the uh complex series now if we have a n
for each n n is a complex number so
where a n is equal to xn + I YN where xn
is the real part of a n and YN is the
imaginary part of a
n then we'll say that the series a n
converges if and only if both the series
xn and YN summation over xn and
summation over YN this
converges now in this case we will say
that a n is up series a n is absolutely
convergent if summation over mod of a n
is converion so now mod of N is a non-
negative real number if this converges
we say that the complex series summation
of a n this is absolutely
convergent so
easily uh one can see that if summation
over mod of a n
convergent then summation a n is also
convergent however the
converse may not be
true so uh all of us we are familiar
with uh the alternating series that's
what the example I have taken you take n
= to sum Over N = 1 to Infinity - 1 ^ n
/ n this is convergent but not
absolutely
convergent because if I take the modulus
then Min -1 ^ n is one and then Su we
know that summation 1 / n is not a
convergent
series now there is a test how to
determine when an alternating series is
going to be convergent or not that is uh
the soal the deine test which says that
you take a n and BN the sequence of
number complex numbers and let the
partial sum of a n that is
bounded and
uh the absolute value of the difference
of the two conjugative BN this is
convergent
and BN converges to zero as n goes to
Infinity then what we get is that summon
over a n BN
convergent for example like in the
previous case so now if I am if we are
taking the summation of this SN if we
take the series to be
- 1 ^ n by n then this
SN a n is equal to -1 the^ n and BN is =
1 by n then S capital of n this is equal
to summation Over N = 1
2 N - 1 the^ n this as you know that
depending on n is odd or even you will
get that this is definitely less or
equal to 1 for all n capital n and that
shows the first condition is satisfied
and in the second condition 1 by n minus
1 by n + 1 this is equal to 1 by n into
n + 1 which is a series of the form 1 /
n² and that is by the comparison test
this I
can do it less or equal to 1 by n² and
this converges and BN of course
converges to zero then a and BN
converg okay so now we'll talk about
some more convergence now if I we have
got two series a n and BN be two
absolutely convergent series which
converges to A and B respectively summon
over a n means the partial sum of a n SN
= to summation
Over N from 1 to
n a n this converges to a similarly the
partial sum for the series BN converges
to
B now what can we say about then we know
that the summation then this series
a n + BN this converges to a +
b and uh now what about the
product so now to discuss that let us
Define CN to be summation of K from 1 to
n a k b of n minus k then in that case
this is the costy product and this
series C n converges to a
b this is the point wise multiplication
of A and
B okay so if one of the series either a
n or BN if that is absolutely convergent
then also we
get the same result that summation of CN
n from one to Infinity that will conver
to AB but if now it is natur natural to
ask suppose we have only the condition
that summation of a n converges and
summation over summation of BN converges
is it necessarily mean that this Cy
product is going to converse to a b in
fact the answer to this question is in
negative so let's look at a very easy
example of this series of two series
which are convergent but their CIP
product may not
converge now for that let us take a
n to be -1 ^ n minus of 1 by square root
of n which is also equal to BN now if I
compute CN then this is going to be now
minus 1 ^ N - 1 is a k minus of 1 into
-1 ^ N - k +
1 uh n -1 - K so now getting cancelled
we'll get that it is -1 ^ n is going to
come out of
it uh and what is left with 1 / root of
K and 1 / < TK of N - K now if you take
the modulus
now if you take the modulus then this
mod of this is equal to
1 therefore this term goes away now the
remaining term whatever inside this all
both of them are non- negative terms 1 /
Ro TK of K is greater than 0 and 1 / n
minus of K
here there is uh I can take this to be
minus of one
so this is also greater or equal
to uh
n uh 1 / < TK of K is greater or equal
to 1 / < TK of
n because K is less or equal to n minus
of 1 and 1 / < TK of n minus of K is
greater or equal to 1 / < TK of n there
and then the sum if I take it all this
terms from the sum I have n minus one
term so this is going to be greater or
equal to 1 for n greater or equal to 2
let's say so however we know that if the
summation CN converges then the this
will imply that CN the sequence CN goes
to zero but here we have got the mod of
C
n is greater or equal to 1 therefore it
can never go to zero therefore this
series does not convert so Cy product
one of the series either a n or BN at
least one of the series has to be
absolutely convergent to get the Cy
product to
converge now another familiar series
which we deal with is the power series
let a n be a sequence of real numbers
then the series of the form a n x the^ n
is called a par series now if I am
taking uh X to be the whatever the x is
this is suppose a particular X then a n
x to the^ n is called the power
series if whenever the series converges
then it makes sense now if if there
exist a capital N such that a n's are
zero after capital N then this is called
a polinomial so in a sense the power
series is nothing but the polinomial
plus three dots
so now what we know is that if a power
series converges for some X
notot then the power series will inverse
for all those X lying on the left of
X Mod of X Mod of X is less than mod of
X not then the power series
converges in this
region okay and similarly if the power
series diverges at some X
not then
any point lying outside the X and minus
X to X not then those series for those X
the power series is going to
diverge so now the question is uh what
can we say that what is the region in
which the power series is going to
converge so natural thought process will
tell that okay so what we will do is
that you will check all such X not for
which a par series converges then we'll
stretch it that means we'll take the
supremum of all the X not for which par
series converges then in that region
suppose it converges here x not then I
will go further and up to this point
where there is a break that from this
point the power series diverges but rest
of the thing converges then in this
region the power series is going to
converge so that idea so how to find
what will be the reason for the
convergence of a power series we have
concept of radius of convergence now we
look at R small R to be limit n goes to
Infinity mod a n ^ 1 by n it may or may
not exist if it exist then we say that
the radius of convergence is 1 / this
small r that is what is the radius of
convergence and then the series
converges absolutely for mod x less than
R so which is essentially showing that
it is that is going to be dominated by a
geometric series and we know
how to compute
determine the convergence of the
geometric
series now consider the series uh x ^ n
n to into 2 ^
n now then if we
take 1 by n ^ 2 / n so this is my
so little r so this is going to the
Limit will exist and be 1 by 2 therefore
this is nothing but R therefore capital
r is going to be 2
now the radius of
convergence is going to be two that
means any X say that mod x is less than
2 The Power series is converge but now
what happened to mod x is equal to 2
there we can see that it we have got two
point + 2 and min-2 now if x is equal to
2 then the series if x is equal to 2
then 2^ n 2 ^ n get cancel we get
n
then now that is a Divergent series now
if x is equal to minus of
2 -2 then if we put that then the series
become -1 to the^ n by
n and that we have seen that this is
series converges so in the radius of
convergence if the radius of convergence
is R then mod x less than R it converges
mod x greater than R it diverges mod x
is equal to R we can't say at some it
may converge or may not converge that is
the essence of the
story
We let's recall that uh a conver
of the sequence of functions so we have
a sequence of functions defined on an
interval I real valued
function and we say that
FN converges to F for some f pointwise
if FNX converges to f ofx as n goes to
Infinity for all X in r that means what
given an Epsilon positive there exist a
n
subot now this n subot will
depend on Epsilon
and on
X for each X there exist a n0 x we can
put for our convenience such that mod of
FN of x minus f of x this is less than
Epsilon for all n greater or equal to
This n z of X now if I can find for a
given Epsilon any Epsilon if I can find
n0
X that is for every X you see here the
n0 X can vary if we vary your X but what
we need for a fixed X we must get a an N
zero and
Subzero uh for which
this
happens now one can see that suppose
what kind of property of the functions
is being preserved with this
convergence the most natural thing to
ask is that if
we
are dealing with a sequence of
continuous function
then will it and it converges to another
function f ofx then is it true that f is
going to be a continuous
function and uh that is uh not true
that's what we know and the example uh
you can
easily see that if you
take f n of X is equal to X to the^ n
where X you take in 0 and
1 then this
converges this converges to here is the
function and this converges to zero as n
goes If X is not equal to
1 because uh X is then in that case X is
less than 1 therefore x the^ n will
Converse to zero now F if x is 1 then
this is going to be f ofx is going to be
uh one because all this FNX FN of uh
this is true for all n therefore now FN
at one is a constant sequence one and
which will converts to one so now this
the limiting function what we are going
to get is uh up to one this is zero and
at one this is one and certainly this is
not a continuous function on Clos
interval
01 so now what extra condition we give
in the convergence which is going to
give us
that the limiting function will also be
a continuous function if we are starting
off with a sequence of continuous
function so now what is uniform
convergence uniform convergence is that
for given Epsilon there you know we are
finding a Cap N Sub 0 which was
depending on X for each fixed X but here
we'll get One n that is uniformly for
all X belongs to I if we manage to get
such a n then we can say that FNX
converges to F ofx
uniformly and uh this uniform
convergence and in other words we can
say that the supremum for a given
Epsilon supremum over X belongs to i f
and x minus F of X Mod is less than
Epsilon for all n greater or equal to n
this and uh this
preserves the
continuity in if FN is a sequence of
continuous function on I and FN
converges to F uniformly then f is going
to be
continuous on
I
okay so now
this is a very brief recall of infinite
series and convergence of sequence of
functions now we would like to recall uh
some basics of remen
integration so REM integration is a
method which uh is tells us to find the
area under a curve
F so now we want to find the area of
this region how to calculate this area
under this region under this curve and
this is what is the prescription was
given by ger mathematician rean Bard
rean uh developed in the mid 19th
century and this is the picture of
Freeman uh okay so now I have a function
f from closed interval AB to R
and uh we are assuming that f is
bounded that means there
exist a
m greater than
zero such that mod of f ofx is less or
equal to M for all X belongs to close
Clos interval
AB so how to do this so as we you recall
that what we do is that uh we divide
this uh the entire interval AB in
[Music]
to a then what we when this is a is X1 X
X2
X3
X4 whatever it is
xn and then this is
um this is some point we keep on
dividing it and this is nothing but xn
and this point is xn minus of one so we
are just partitioning it we are you see
the length between these two can vary
conjugative uh XI and x i + 1 can vary
so we are this is called a partition of
a b so now the reman sum with a given
partition is that so you have a function
f which is defined on this is a this is
B and you have some
partition X3 let let us say
here this is x i this is x i + 1 and
this is
xn so now you choose any point CI
here now at this
point you
have you take
this rectangle
and similarly you choose any point over
here consider all these rectangles you
keep on doing it so now you get the
reman sum which is essentially the
rectangle here it is nothing but Delta
of XI and this is the height of the
rectangle I am choosing here a non-
negative function so for my
picture uh so what we have got is that
uh
that is the remen sum FCI Delta of
XI and uh remen in to define the reman
integration we Define remember that this
is the Partition p
so now this Norm of P I am going to
Define is the maximum of all this
difference so this is this is my X2 this
is my X3 like this so maximum of the
length of the
interval and
uh now we say that f is remain
integrable on AB
if the limit
of are the reman sum corresponding to P
that goes to that exist and then we
denote the reman integration to be a to
b f of x is the limit of Norm p goes to
zero R of
FP thank you
[Music]
[Music]
[Music]
