0% found this document useful (0 votes)
58 views9 pages

Analysis 3 Homework

Uploaded by

Ikram Bbezzit
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
58 views9 pages

Analysis 3 Homework

Uploaded by

Ikram Bbezzit
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 9

Analysis 3’s homework

Bouhafara Nour EL Houda A2


January 4, 2024

1 Solutions of the problems :


1.1 Problem 4 : (I would like to Begin with the 4th one)
Let (an )n≥0 be a decreasing sequence of positive numbers with limit 0, and let bn = an − 2an+1 + an+2 ≥ 0 ∀n ≥ 0

Showing that:
+∞
X
nbn = a1
n=0
Lets start with calculating the partial sum sn :
Pn Pn Pn Pn Pn
sn = k=0 kbk = k=0 (kak − 2kak+1 + kak+2 ) = k=0 kak + k=0 kak+2 − k=0 2kak+1

= 0 ∗ a0 + 1 ∗ a1 + 2a2 + 3a3 + 4a4 + 5a5 + 6a6 + · · · + nan +

0 ∗ a2 + 1 ∗ a3 + 2a4 + 3a5 + 4a6 + · · · + (n − 2)an + (n − 1)an+1 + nan+2

−2 ∗ 0 ∗ a1 − 2 ∗ 1 ∗ a2 − 2 ∗ 2a3 − 2 ∗ 3a4 − 2 ∗ 4a5 − 2 ∗ 5a6 − · · · − 2(n − 1)an − 2nan+1

= a1 + 2a2 + 4a3 + 6a4 + 8a5 + 10a6 + · · · + (2n − 2)an + (n − 1)an+1 + nan+2 − 2 ∗ 0 ∗ a1 − 2 ∗ 1 ∗ a2 − 2 ∗ 2a3 −
2 ∗ 3a4 − 2 ∗ 4a5 − 2 ∗ 5a6 − · · · − 2(n − 1)an − 2nan+1

= a1 + (n − 1)an+1 + nan+2 − 2nan+1 = a1 − (n + 1)an+1 + nan+2


P+∞
We know that : n=0 nbn = limn→+∞ sn

Since (an )n≥0 be a decreasing sequence of positive numbers Then : a0 > a1 > · · · > an > an+1 > an+2

Thus we have :
a1 − (n + 1)an+2 + nan+2 < a1 − (n + 1)an+1 + nan+2 < a1 − (n + 1)an+1 + nan+1

a1 − an+2 < a1 − (n + 1)an+1 + nan+2 < a1 − an+1

We have that :
lim an = 0
n→+∞

Thus:
lim (a1 − an+2 ) = lim (a1 − an+1 ) = a1
n→+∞ n→+∞

Then: limn→+∞ (a1 − (n + 1)an+1 + nan+2 ) = a1


So:
+∞
X
nbn = a1
n=0

1
1.2 Problem 6 :
We have : f0 (x) = ex

And : fn+1 (x) = xfn′ (x)

Let’s show now that :


+∞
X fn (1)
= ee
n=0
n!

Using Taylor expansion we have :

x2 x3 xn xn
P+∞
f0 (x) = 1 + x + 2! + 3! + ··· + n! + ··· = n=0 n!

xn ′ nxn
P+∞ P+∞
f1 (x) = x n=0 ( n! ) = n=0 ( n! )

nxn ′ n2 x n
P+∞ P+∞
f2 (x) = x n=0 ( n! ) = n=0 ( n! ) ...

k n xk
P+∞
Thus :fn (x) = k=0 ( k! )

By substituting x by 1 we get :
+∞
X kn
fn (1) = ( )
k!
k=0

kn
P+∞
k=0 ( k! )
P+∞ fn (1) P+∞
n=0 n! = n=0 n!

Note that n is independent from k so :


+∞ X+∞
X kn
n=0
k!n!
k=0
+∞
X ek
= = ee
k!
k=0

As required

1.3 Problem 7 :

For any integer n≥ 1, let < n > denote the closest integer to n

Evaluating:
+∞ <n>
X 2 + 2−<n>
n=1
2n
Here we will try to remove the ” <> ” :

Notice that between (n + 1)2 and n2 there are 2n integers ∀n ∈ N (n2 + 2n + 1 − n2 + 1 − 2 = 2n: we subtract
−2 to eliminate the first and the last term )

Logically from these 2n n (call them k) of them are close to n : < k >= n
and the remaining n (call them l) are close to n + 1 : < l >= n + 1

Therefore, for any n ∈ N the positive integers k that have < k >= n are
n2 − (n − 1), n2 − (n − 2), . . . , n2 , n2 + 1, n2 + 2, . . . , n2 + n

2
Now ∀n ∈ N we have :

2
nX +n
2<k> + 2−<k>
2k
k=n2 −(n−1)
2
nX +n
2n + 2−n
=
2k
k=n2 −(n−1)
2
nX +n
n −n 1
= (2 + 2 )
2k
k=n2 −(n−1)

This is nothing but a geometric sum : so let’s simplify it :

2 (2−1 )2n − 1
= (2n + 2−n ) (2−(n −(n−1))
)( 1 )
2 −1
2
2n−1−n (2−2n − 1)
= (2n + 2−n ) ( 1 )
2 −1
2 2
n −n 2−n −n−1
− 2n−1−n
= (2 + 2 )( 1 )
2 −1
2 2
= (2n + 2−n ) (2n−n − 2−n −n
)

2 2 2 2 2 2 2 2
= 22n−n − 2−n −2n
= 21−(n−1) − 21−(n+1) = (21−(n−1) + 21−n ) − (21−n + 21−(n+1) )

Now let un≥1 be a sequence such that ∀n ∈ N :

2 2
un = 21−(n−1) + 21−n
Thus for ∀n ∈ N :
2
nX +n
2<k> + 2−<k>
= un − un+1
2k
k=n2 −(n−1)

This implies that :


+∞ <n> +∞
X 2 + 2−<n> X
= uj − uj+1
n=1
2n j=1

Passing by the limit of the partial sum we get :

lim (u1 − ui+1 ) = 3


i→+∞

1.4 problem 5 :
Let un≥1 be a sequence of positive integers. For an integer m let Vm denote the number of terms of un≥1 that are
less than or equal to m

1
P
Showing that if n≥1 un is convergent, then

Vn
lim =0
n→+∞ n

3
Vn P 1
Suppose that : limn→+∞ n ̸= 0 and let’s try to prove that n≥1 un is divergent
Vn Vn
Since limn→+∞ n ̸= 0 so ∃ ϵ > 0 such that: ∀n (large) n ≥ ϵ
Vn
This implies that : ∃ N such that : ∀n ∈ N we have the same result which is : n ≥ ϵ

Let
N
X 1
SN =
u
n=1 n

be the partial sum


N VN VN
X 1 X 1 X 1
SN = ≥ ≥ϵ
n=1
u n n n
n=N n=N
P+∞ 1
We know that the harmonic series n=1 n is divergent multiplying by ϵ doesn’t change the divergence =⇒ SN is
divergent
P+∞ 1
P+∞ 1
Since SN is the partial sum of n=1 un =⇒ n=1 un is divergent too

As required ( i used contraposition)

4
2 More Convergence Tests (Solutions):
2.1 Test 1 : (Kummer’s Test)
P P
Let n≥0 un be a positive series , and let n≥0 dn be a positive divergent series . Define the Kummer characteristic
Kn by :
1 un 1
Kn = − ,n ≥ 0
dn un+1 dn+1

Supposing that : limn→+∞ Kn = K − ∞ ≤ K ≤ +∞

Showing that :
P
(a) if K > 0 , then n≥0 un is convergent .
P
(b) if K < 0 , then n≥0 un is divergent .

Case (a) : If K > 0 : let r be a real number such that 0 < r < L then there must be an integer N ≥ 0 ( rank )
such that :
1 un 1 1 1
− > r ⇐⇒ un − un+1 > r un+1
dn un + 1 dn+1 dn dn+1

∀n ≥ N . Given any positive integer m , we have :


1 1
uN − uN +1 > r uN +1
dN dN +1
1 1
uN +1 − uN +2 > r uN +2
dN +1 dN +2
1 1
uN +2 − uN +3 > r uN +3 . . .
dN +2 dN +3

1 1
uN +m−1 − uN +m > r uN +m
dN +m−1 dN +m

Summing side by side we get :


1 1
uN − uN +m > r (uN +1 + . . . + uN +m ) = r(SN +m − SN )
dN dN +m

PN +m
where Sn is the partial sum for k=N +1 uk , thus :
1 1 1
r SN +m < r SN + uN − uN +m < r SN + uN . . . (1)
dN dN +m dN

r SN + d1n un
let M1 = r

from (1) we can remark that : Sn < M1 ∀n ≥ N , thus : the sequence of partial sums (Sn )n≥0 of the series
P+∞
n=0 un is bounded and therefore the series is convergent .

Case (b) : If K < 0 : this means that ∃ N ≥ 0 such that :


1 un 1
− ≤ 0 ∀n ≥ N
dn un+1 dn+1
1 1
un ≤ un+1 ∀ n ≥ N
dn dn+1

5
this implies that :
1 1
uN ≤ un ∀ n ≥ N
dN dn

1 M2
Let M2 = dN uN so : un ≥ 1 = dn M2
dn

P+∞ P+∞
for n ≥ N and because n=0 dn diverges the series n=0 un diverges by the comparaison test

this completes the proof as required .

2.2 Test 2 :
Let (un )n≥0 be a sequence of positive real terms satisfying :
 
un
lim n ln = l − ∞ ≤ l ≤ +∞
n→+∞ un+1
P
Showing that : the series n≥0 un converges if l > 1 and diverges if l < 1

Suppose that l exists with : −∞


P ≤ l ≤ +∞ P . Using the previous Test (Kummer’s Test) . And without loss of
1
generality let n ≥ 1 and consider n≥1 dn = n≥1 n ln n
 
1 un 1
K = lim − ∀n ≥ 1
n→+∞ dn un+1 dn+1
 
un
K = lim n ln n − (n + 1) ln (n + 1)
n→+∞ un+1
  
un 1
= lim n ln n − (n + 1) ln n − (n + 1) ln 1 +
n→+∞ un+1 n
    
un 1
= lim ln n n − (n + 1) − (n + 1) ln 1 +
n→+∞ un+1 n
        
un 1 1
= lim ln n n − 1 − 1 − n ln 1 + − ln 1 +
n→+∞ un+1 n n
      n  
un 1 1
= lim ln n n − 1 − 1 − ln 1 + − ln 1 + =l−1
n→+∞ un+1 n n
n
Because : limn→+∞ ln 1 + n1 = 1 and limn→+∞ ln 1 + n1 = 0


P
This implies that : if l > 1 then K > 0 and the series n≥1 un converges by Kummer’s Test . similarly if l < 1
then K < and the series diverges.

As required .

2.3 Test 3 : (Raabe’s Test)


Let un≥0 be a sequence of positive real numbers satisfying
un
lim n ( − 1) = l
n→+∞ un+1
Supposing that l exists
P
Showing that : the series n≥0 un converges if l > 1 and diverges if l < 1. It is simple to prove this using the
Kummer’s Test

6
1
P P
Consider n≥0 dn = n≥0 n this series is a positive divergent series . we know that ∀n ≥ 0 :

1 un 1
K = lim ( − )
n→+∞ dn un+1 dn+1

un
K = lim (n − (n + 1))
n→+∞ un+1
un
= lim (n ( − 1) − 1) = l − 1
n→+∞ un+1
P+∞
If l > 1 then K > 0 and the series n=0 converges by Kummer’s Test . Similarly, if l < 1 then K < 0 then
P+∞
n=0 diverges

2.4 Test 4 : (Gauss’s Test)


Let (un )n≥0 be a sequence of positive real numbers satisfying :
un β θn
= λ + + 1+ϵ
un+1 n n
for n sufficiently large , where λ , β , ϵ > 0 are real constants and |θn | ≤ M , M > 0

showing that :
P
(a) the series n≥0 un converges if λ > 1 or (λ = 1 and β > 1)
P
(b) the series n≥0 un diverges if λ < 1 or (λ = 1 and β ≤ 1)

Let’s study the case of λ ̸= 1 :


un
Notice that : limn→+∞ un+1 = λ , hence by the ratio test the series converges if λ > 1 and diverges if λ < 1

Now when λ = 1 :
 
un
Notice that : limn→+∞ n un+1 − 1 = β , hence by Raabe’s Test the series converges if β > 1 and diverges if
β<1

Raabe’s Test can’t determine the case of β = 1 , so :


un 1 θn
For β we have : un+1 =1+ n + n 1+ϵ ϵ>0

1
P
Let’s compare this series with n≥1 vn where vn = n ln n this series diverges ( Bertrand series with α = 1 , β = 1 )

un vn 1 θn (n + 1) ln (n + 1)
− = 1 + + 1+ϵ −
un+1 vn+1 n n n ln n
 
θn n + 1 ln (n + 1)
= 1+ϵ − −1
n n ln n
   ϵ 
1 1 n
= 1+ϵ θn − (n + 1) ln 1 +
n n ln n

n nϵ
but limn→+∞ (n + 1) ln 1 + n1 = limn→+∞ ln 1 + n1 + 1 + n1 = 1 and limn→+∞
  
ln n = ∞ , ϵ > 0 and
(θn )is bounded (|θn | ≤ M , M > 0)

1 nϵ

therefore : for sufficiently large n (θn − (n + 1) ln 1 + n ln n ) is negative

Thus ∃ N positive integer such that :

7
 ϵ 
1 n
θn − (n + 1) ln 1 + < 0 ∀n ≥ N
n ln n
un vn
=⇒ − < 0 ∀n ≥ N
un+1 vn+1
un vn
=⇒ < ∀n ≥ N
un+1 vn+1

P+∞ P+∞
Since : n=1 vn diverges , so by the Logarithmic Comparison Test n=1 un diverges too.

And this complets the proof .

2.5 Test5 : (Duhamel’s Test)


Let (un )n≥0 be a sequence of positive real numbers satisfying :
 
un+1 β 1
=1− +o . . . (1)
un n n
Where β is a constant
P
Showing that : the series n≥0 un converges if β > 1 and diverges If β < 1

If β > 1 (resp β < 1) consider β ′ ∈ (1 , β) (resp β ′ ∈ (β , 1))



To exploit the previous hypothesis . Let’s consider the sequence an = ln (nβ un ) and let bn = an+1 − an ,
then :
   
1 un+1
bn = β ′ ln 1 + + ln
n un
Substituting by (1) we get :
    
′ 1 β 1
bn = β ln 1 + + ln 1 − + o
n n n

Using DL :
"  2 #
β′
   
1 β 1 β 1
= +O − +o +O − +o
n n2 n n n n

1 2 1 1 1 1 1 1
      
We know that : o n =o n2 , n o n =o n2 , o n2 =O n2 , O(O(un )) = O(un ) ,
un 1

O n = o(un ) and limn→+∞ n = 0 thus :

  2   2    2  
β 1 1 β 1 β 1
− +o = o −2 o + =O
n n n n n n n2
Therefore :
β′−β β′−β
 
1
un = +o ∼
n n n

Thus for the convergence case we have : β ′ − β < 0 =⇒


P
n≥0 un → −∞ which means an → −∞ which
β′ −β ′
P+∞
means n un → 0 which means un = o(n ) hence the result which is : n=0 un converges because : (β ′ > 1)
′ P+∞
Similarly β ′ − β > 0 gives us n β un → +∞ so n−β = O(un ) hence the result which is : n=0 un diverges
because : (β ′ < 1)

this completes the proof.

8
2.6 Test 6 : (Schlomilch Condensation Test)
Let (un )n ≥ 1 be a decreasing sequence of positive real numbers , and let (nk )k≥1 be an increasing sequence of
positive integers such that :
nk+1 − nk
∃C > 0, ∀k ≥ 1 : ≤ C . . . (1)
nk − nk−1

P P
Showing that : the series n≥1 un converges if and only if the series n≥1 (nk+1 − nk ) unk converges .
P
Since the series n≥1 un has positive decreasing terms , ∀ n ,

(nk+2 − nk+1 ) unk+1


< (nk+1 − nk ) unk+1 ≤ unk + unk+1 + unk+2 + · · · + unk − 1 ≤ (nk+1 − nk ) unk
C

Where in the first inequality we used (1).


By summing the inequality above over n:
+∞ +∞ +∞
1 X X X
(nk+1 − nk ) unk < un ≤ (nk+1 − nk ) unk
C n=1 n=1 n=1

Applying the Comparison Test we get the result.

You might also like