Chapter 2

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 34

Chapter 2:

1. Arithmetical Functions and Dirichlet Multiplication


Definition 1.1. A real or complex valued function defined on the positive integers
is called an arithmetical function or a number theoretical function.
For example
n(n + 1)
i.e., f : N → R by f (n) = , or
2
f (n) = n, or
f (n) = a, where a is constant.
The Mobius Function µ(n):
The Mobius function µ is defined as follows

ra
µ(1) = 1.

d
en
If n > 1, we can write n = pa11 · pa22 · · · pakk then
m
ar

(−1)k ,

if a1 = a2 = . . . = ak = 1 (i.e., n = p1 · p2 · · · pk ) 
µ(n) =
i.e., n = pa11 · pa22 · · · pakk
Dh

0, Otherwise
For example n is a prime number
N.

i.e., n = p ⇒ µ(p) = −1.


B.

For example n is a power factor


Dr

i.e., n = p2 ⇒ µ(p2 ) = 0.

n 1 2 3 4 5 6 7 8 9 10
µ(n) 1 −1 −1 0 −1 1 −1 0 0 1

The mobius function ares in many different places in number theory.


One X
of its fundamental properties is a remarkably simple formula for the divisor
sum µ(d) extended over the positive divisor of n in this formula [x] denotes the
d|n
greatest integer ≤ x.
Example:
n = 12 = 22 · 3, the divisor of n are 1, 2, 3, 2 · 3, 22 , 22 · 3

X
µ(d) = µ(1) + µ(2) + µ(3) + µ(2 · 3) + µ 22 + µ 22 · 3
 

d|n

= 1 + (−1) + (−1) + (−1)2 + 0 + 0


= 1−2+1
= 0.
2 Dr. B. N. Dharmendra

2
X
(a + b)2 = 2
Cn a2−n bn
n=0
2
= C0 a2 b0 + 2 C1 a1 b1 + 2 C2 a0 b2
put a = 1 and b = −1, then
(1 − 1)2 = 2
C0 (−1)0 + 2 C1 (−1)1 + 2 C2 (−1)2
0 = 2 C0 (−1)0 + 2 C1 (−1)1 + 2 C2 (−1)2
0 = 1−2+1

In general

k
X
(a + b)k = k
Cn ak−n bn
n=0
k k
(a + b) = C0 ak b0 + k C1 ak−1 b1 + · · · + k Ck a0 bk

ra
put a = 1 and b = −1

d
(1 − 1)k = k
en
C0 (−1)0 + k C1 (−1)1 + k C2 (−1)2 + · · · + k Ck (−1)k
m
k
0 = C0 (−1)0 + k C1 (−1)1 + k C2 (−1)2 + · · · + k Ck (−1)k
ar
Dh

Theorem 1.1. If n ≥ 1, we have


  
1 1, if n = 1,
N.

X
µ(d) = =
n 0, if n > 1.
B.

d|n

Proof. The formula is clearly true, if n = 1.


Dr

X
Assume that n > 1 and we write n = pa11 ·pa22 ···pakk in the µ(d) the only non-zero
d|n
terms come from and from those divisors of n which are products of distinct primes,
then
X
µ(d) = µ(1) + µ (p1 ) + µ (p2 ) + · · · + µ (pk−1 ) + µ (pk )
d|n
+µ (p1 p2 ) + · · · + µ (pk−1 pk ) + · · · + µ (p1 p2 · · · pk−1 pk ) .
= 1 + (−1) + (−1) + · · · + (−1) + (−1)2 + (−1)2 + · · · + (−1)2 + · · · + (−1)k .
| {z } | {z }
k times k(k−1)
times
2

= k C0 (−1)0 + k C1 (−1)1 + k C2 (−1)2 + · · · + k Ck (−1)k


= 0.

The Euler Function φ(n):


The Euler function φ is defined as follows

φ(1) = 1.
Theory of Numbers 3

If n ≥ 1, the euler totient function φ(n) is defined to be the number of positive


integers not exceeding n which are relatively prime to n, thus
Xn
0
(1.1) φ(n) = 1
k=1
where the 0 indicates that the sum is extended over those relatively prime to n.

Here is a short table of values of φ(n)

n 1 2 3 4 5 6 7 8 9 10
φ(n) 1 1 2 2 4 2 6 4 6 4

Note:
If n = 2, then the divisors are 1, 2
X X
φ(d) = φ(1) + φ(2) = 1 + 1 = 2 ⇒ φ(d) = 2.

ra
d|2 d|2

d
If n = 3, then the divisors are 1, 3 en
m
X X
φ(d) = φ(1) + φ(3) = 1 + 2 = 3 ⇒ φ(d) = 3.
ar

d|3 d|3
Dh

If n = 6, then the divisors are 1, 2, 3, 6


N.

X X
φ(d) = φ(1) + φ(2) + φ(3) + φ(6) = 1 + 1 + 2 + 2 = 6 ⇒ φ(d) = 6.
B.

d|6 d|6
If n = 8, then the divisors are 1, 2, 4, 8
Dr

X X
φ(d) = φ(1) + φ(2) + φ(4) + φ(8) = 1 + 1 + 2 + 4 = 8 ⇒ φ(d) = 8.
d|8 d|8
X
Theorem 1.2. If n ≥ 1, then φ(d) = n.
d|n

Proof. Let S denote the set S = {1, 2, 3, . . . , n}. We distributed the integers of S
into disjoint sets as follows for each divisor d of n.

Let A(d) = {K ∈ S /(K, n) = d }


i.e., A(d) contains those elements of S which have the g.c.d. d with n. Here
d1 6= d2 ⇒ d1 |n, d2 |n.
i.e., A(d1 ) ∩ A(d2 ) = Φ(empty)
If K ∈ A(d1 ) ∩ A(d2 ) ⇒ (K, n) = d1 & (K, n) = d2 .
i.e., d1 = d2 , which is a contradiction.
A(d) is disjoint or A(d1 ) ∩ A(d2 ) = Φ(empty)
⇒ |A(d1 ) ∩ A(d2 )| = 0

A(d) ⊂ S, ∀ d|n
4 Dr. B. N. Dharmendra

[
(1.2) A(d) ⊂ S
d|n

If K ∈ S and (K, n) = d, then K ∈ A(d)


[
(1.3) S ⊂ A(d)
d|n

From the above equations (1.2) and (1.3), we obtain


[
(1.4) S = A(d).
d|n

The sets A(d) form a disjoint collection whose union is S.


If f (d) denotes the number of integers in A(d), we have
X
(1.5) f (d) = n.

ra
d|n

d
 
K n en
but (K, n) = d if and only if ,
d d
m
K n
i.e., 0 < K ≤ n if and only if 0 < ≤
ar

d d
Dh

There is one−one correspondence between the elements of A(d) and those integers
q satisfying
K n
N.

q= ⇒ 0<q≤
d d
.
B.

n n
q, = 1. The number of such q us φ .
d d
Dr

n
Hence f (d) = φ and also the number of elements in A(d) is n
d

i.e., |S| = |A(d1 )| + |A(d2 )| + . . . + |A(d2 )| − |A(d1 ) ∩ A(d1 ) . . . ∩ A(d1 )|


n = |A(d1 )| + |A(d2 )| + . . . + |A(d2 )| − 0
X
n = |A(d)|
d|n
X n
n = φ
d
d|n
X
n = φ (d) .
d|n

Hence the proof. 


Relation between the Euler function and Mobius function.
Theorem 1.3. If n ≥ 1, then
X n φ(n) X µ(d)
φ(n) = µ(d) or =
d n d
d|n d|n
Theory of Numbers 5

Proof. We know that


n
X
0
(1.6) φ(n) = 1
k=1

The Euler function φ(n) (1.6) can be rewritten the form

n  
X 1
(1.7) φ(n) =
(n, k)
k=1

where k runs through all integers ≤ n. We know that


  
X 1 1, if n = 1,
(1.8) µ(d) = =
n 0, if n > 1.
d|n

ra
Employing the above results (1.4) with (1.6), we get

d
en
n
m
X X
φ(n) = µ(d) [by using equation (1.4) with n replace by (n, k)]
ar

k=1 d|(n,k)
Dh

Xn X
= µ(d)
N.

k=1 d|n, d|k


B.

for fixed divisor d of n. We must sum over all the k in the range which are multiples
n
Dr

of d. If we write k = q d, then 1 ≤ k ≤ n if and only if 1 ≤ q ≤ . Hence the last


d
sum φ(n) can be written as
n
d
XX
φ(n) = µ(d)
d|n q=1
n
X d
X
0
= µ(d) 1
d|n q=1
X n n n o
φ(n) = µ(d) . from equation (1.6) with replace n by
d d
d|n
or
φ(n) X µ(d)
= .
n d
d|n


Y 1

Theorem 1.4. If n ≥ 1, then φ(n) = n 1− .
p
p|n
6 Dr. B. N. Dharmendra

Proof. If n ≥ 1, n = pα1 1 · pα2 2 · pα3 3 · · · pαk k , where α ≥ 1.


Consider

Y 1
 
1
 
1
 
1
 
1

1− = 1− · 1− · 1− ··· 1 −
p p1 p2 p3 pk
p|n

(−1)k
     
−1 −1 −1 1 1
= 1+ + + ··· + + ··· + ··· +
p1 p2 pk p1 p2 pk−1 pk p1 p2 · · · pk
X (−1) X (−1) 2 (−1) k
= 1+ + + ··· + .
pi pi pj p1 p2 · · · pk
X µ (pi ) X µ (pi pj ) µ (p1 p2 · · · pk )
= 1+ + + ··· + .
pi pi pj p1 p2 · · · pk
X µ(d)
=
d
d|n

ra
Y 1

φ(n)
1− =

d
p n en
p|n
or
m
Y 1

ar

φ(n) = n 1− .
p
Dh

p|n
N.


B.

Properties of Euler’s function:


Dr

(i) φ (pα ) = pα − pα−1 for some p, if α ≥ 1.

Proof. We know that

Y 1

φ(n) = n 1−
p
p|n
α
Putting in the above equation n = p , then
Y  1

α
= p 1−
α
p
p|p
 
1
φ(n) = pα 1 −
p
φ(n) = pα − pα−1 .

d
(ii) φ (m n) = φ (m) φ (n) .
φ (d)
Theory of Numbers 7

Proof. We know that


Y 1

φ(n) = n 1−
p
p|n
Replace n by m n, then
Y  1

φ(m n) = m n 1−
p
p|m n
   
1 Y 1
Y
1− 1−
 p p   
 p|m p|n  p|m · p|n
= mn    , ∵ If (m, n) = d, then
Y 1  p|d
 1− 
p
p|d
Y 1
 Y
1

m 1− n 1−
p p
p|m p|n

ra
=  
Y 1

d
1− en
p
p|d
m
φ (m) φ (n)
ar

=
φ (d)
Dh

d
d
N.

φ(m n) = φ (m) φ (n) .


φ (d)
B.


Dr

(iii) φ(m n) = φ (m) φ (n)


Proof. We know that
d
φ(m n) = φ (m) φ (n)
φ (d)
If (m, n) = 1 = d, then
1
φ(m n) = φ (m) φ (n)
φ (1)
φ(m n) = φ (m) φ (n) .

(iv) φ(m n) ≥ φ (m) φ (n)
Proof. We know that
d
φ(m n) = φ (m) φ (n)
φ (d)
d
If φ(d) ≥ d ⇒ ≥ 1, then
φ(d)
φ(m n) ≥ φ (m) φ (n) .

8 Dr. B. N. Dharmendra

(v) a|b, then φ(a)|φ(b).

Proof. Given a|b ⇒ b = a · c

φ(b) = φ (a · c)
d
= φ (a) φ (c) , where (a, c) = d
φ (d)
d
φ(b) = k φ (a) , where k = φ(c) ·
φ(d)
⇒ φ(a)|φ(b).

Dirichlet Product of Arithmetic Function:

ra
Definition 1.2. If f and g are two arithmetical functions. We define the Dirichlet

d
product or Dirichlet Convolution to be the arithmetical function h defined by the
en
equation
m
X n
h(n) = f (d) g
ar

d
d|n
Dh

Notation: We write f ∗ g for h and (f ∗ g) (n) for h(n). The symbol N will be
N.

used for the arithmetical function for which N (n) = n for all n. In this notations
theorem can be stated in the form φ = µ ∗ N.
B.

X n
Dr

φ(n) = µ(d)
d
d|n
X n
φ(n) = µ(d) N
d
d|n
φ(n) = (µ ∗ N ) (n)
or
φ = µ∗N

Multipartite Function:

Definition 1.3.
f (m n) = f (m) f (n), if (m, n) = 1
here f is not identically zero i.e., f (1) 6= 0.

Theorem 1.5. Dirichlet multiplication is commutative and associative. That is any


arithmetical function f ∗ g, we have

f ∗g = g ∗f (Commutative Law),
(f ∗ g) ∗ k = f ∗ (g ∗ k) (Associative Law).
Theory of Numbers 9

Proof. First we note that the definition of f ∗ g can also be expressed as


X n
(f ∗ g) (n) = f (d) g
d
d|n
X
= f (a) g (b)
n=a·b
X
= g(a) f (b)
n=a·b
X n
= g(d) f
d
d|n
(f ∗ g) (n) = (g ∗ f ) (n).
Where a and b vary over all positive integers, whose product is n. This makes the
commutative property it self evident.
To prove the associative property. We let A = g ∗ k and consider f ∗ A = f ∗ (g ∗ k),
we have

ra
X
(f ∗ A) (n) = f (a) A(d)

d
a·d=n
en
m
X X
= f (a) g(b) k(c)
ar

a·d=n b·c=d
Dh

X
(f ∗ A) (n) = f (a) g(b) k(c).
a·b·c=n
N.

In the same way, if we let B = f ∗ g and consider B ∗ k. We are led to the same
formula for B ∗ k = (f ∗ g) ∗ k, we have
B.

X
(B ∗ k) (n) = B(e) k(c)
Dr

e·c=n
!
X X
= f (a) g(b) k(c)
e·c=n a·b=e
X
(B ∗ k) (n) = f (a) g(b) k(c).
a·b·c=n
∴ f ∗ A = B ∗ k,
which means that Dirichlet multiplication is associative. 
We now introduce an identity element for their multiplication.
Definition 1.4. The arithmetical function I given by
  
1 1, if n = 1,
I(n) = =
n 0, if n > 1.
is called the identity function.
X
Example is I(n) = µ(d).
d|n

Theorem 1.6. For all f , we have


I ∗ f = f ∗ I.
10 Dr. B. N. Dharmendra

Proof. We have
X n
(f ∗ I) (n) = f (d) I
d
d|n
= f (1) I(n) + . . . + f (n) I(1)
= 0 + f (n) · 1 [∵ I(1) = 1 and I(n) = 0, if n > 1.]
(f ∗ I) (n) = f (n).

Dirichlet inverse and the Mobius inverse formula:

Theorem 1.7. If f is an arithmetical function with f (1) 6= 0, there is an unique


arithmetical function f −1 is called the Dirichlet inverse of f, such that
f ∗ f −1 = f −1 ∗ f = 1.
1
More over f −1 is given by the recursion formulas f −1 = .

ra
f (1)

d
−1 X n
en
f −1 (n) = f f −1 (d), for n > 1.
f (1) d
m
d|n, d<n
ar

Proof. Given f , we shall show that the equation


Dh

f ∗ f −1 (n) = I(n)


has a unique solution for the function values f −1 (n).


N.

For n = 1, we have to solve the equation


B.

f ∗ f −1 (1) = I(1)

Dr

 
−1 1
X
f (d) f = 1
d
d|1

f (1) f −1 (1) = 1
Since f (1) 6= 0, there is one and only one solution, namely
1
f −1 (1) = .
f (1)
Assume now that the function values f −1 (k) have beenuniquely determined for all
1 < k < n, then we have to solve the equation f ∗ f −1 (n) = I(n)
X n
f (d) f −1 = I(n)
d
d|n
X n
f (1) f −1 (n) + f (d) f −1 = 0
d
d|n, d<n
X n
f (1) f −1 (n) = − f (d) f −1
d
d|n, d<n
−1 X n
f −1 (n) = f (d) f −1 .
f (1) d
d|n, d<n

Since f (1) 6= 0, this establishes the existence and uniqueness f −1 by induction. 


Theory of Numbers 11

Note:
We have (f ∗ g) (1) = f (1) g(1).
Hence if f (1) 6= 0 and g(1) 6= 0, then (f ∗ g) (1) 6= 0.
This fact along with Theorems (1.6), (1.7)and(1.8) tell us that is language of group
theory. The set of all arithmetical functions f with f (1) 6= 0 forms an abelian group
with respect to the operation ∗ the identity element being the function I the reader
can easily verify that
(f ∗ g)−1 = f −1 ∗ g −1 , if f (1) 6= 0 and g(1) 6= 0.
Definition 1.5. We define the unit function u to be the arithmetical function, such
that
u(n) = 1, ∀ n.
Theorem (1.1) states that
  
X 1 1, if n = 1,
I(n) = µ(d) = =

ra
n 0, if n > 1.
d|n

d
en
In the notation of Dirichlet multiplication this becomes µ ∗ u = I.
m
Problem 1.1.
ar
Dh

µ ∗ u = I.
Proof. Consider
N.

X n
(µ ∗ u) (n) = µ(d) u
B.

d
d|n
Dr

X
= µ(d) 1
d|n
(µ ∗ u) (n) = I(n).


Note:
If µ ∗ u = I, then u and µ are Dirichlet inverses of each other
i.e, u = µ−1 and µ = u−1 .
This simple property of the Mobius function, along with the associative property of
Dirichlet multiplication, enables us to give a simple proof of the next theorem.
Theorem 1.8. Mobius inversion formula. Then
X
(1.9) f (n) = g(d)
d|n

if and only if
X n
(1.10) g(n) = f (d) µ .
d
d|n
12 Dr. B. N. Dharmendra

Proof. By using Dirichlet production definition equation (1.9) can be written as


X n
f (n) = g(d) u [∵ u(n) = 1, ∀ n]
d
d|n
f (n) = (g ∗ u) (n)
⇒ f = g ∗u
Multiplying by µ
f ∗ µ = (g ∗ u) ∗ µ
= g ∗ (u ∗ µ) [∵ Associative Law]
= g ∗I
f ∗ µ = g or
g = f ∗µ
X n
i.e., g(n) = f (d) µ .
d

ra
d|n

d
en
Conversely, Again by using Dirichlet production definition equation (1.10) can be
m
written as
ar

X n
Dh

g(n) = f (d) µ
d
d|n
N.

g = f ∗µ
Multiplying by u
B.

g ∗ u = (f ∗ µ) ∗ u
Dr

= f ∗ (µ ∗ u) [∵ Associative Law]
= f ∗I
g ∗ u = f or
f = g ∗u
X n
i.e., f (n) = g(d) u
d
d|n
X
∴ f (n) = g(d). [∵ u(n) = 1, ∀ n]
d|n

The Mangoldt function ∧(n):


We introduce Mangoldt function ∧ which plays a central role in the distribution of
primes

Definition 1.6. For every integers n ≥ 1, we define

log p, if n = pm for some prime p and some m ≥ 1,



(1.11) ∧(n) =
0 otherwise.
Theory of Numbers 13

For example

∧(1) = 0
∧(2) = log (2)
∧(3) = log (3)
∧(4) = log 22


∧(4) = log (2)


∧(6) = 0

Here is a short table of values of ∧(n) The proof of the next theorem shows how

n 1 2 3 4 5 6 7 8 9 10
∧(n) 0 log 2 log 3 log 2 log 5 0 log 7 log 2 log 3 0

d ra
en
this function are naturally from the fundamental theorem of arithmetic.
m
Theorem 1.9. If n ≥ 1, we have
ar
Dh

X
(1.12) log n = ∧(d).
N.

d|n
B.

Proof. Theorem is prove by induction on n. The equation (1.12) is true for n = 1,


Dr

X
log 1 = ∧(d).
d|1
0 = ∧(1)
0 = 0.

Assume that result is true for n > 1, we write

r
Y
n = P1a1 · P2a2 · · · Pkak = Pkak .
k=1

Taking logarithms on both side, we get

r
X
log Pkak

log n =
k=1

r
X
(1.13) log n = ak log (Pk ) .
k=1
14 Dr. B. N. Dharmendra

Now consider the sum on the right hand side of equation (1.12).
X r
X ak
X
∧(d) = ∧ (pm
k )
d|n k=1 m=1
r
X ak
X
= log (pk )
k=1 m=1
Xr
= ak log (pk )
k=1
X
∧(d) = log n. [from equation (1.13)]
d|n
or
X k
X
∧(pi ) + ∧(p2i ) + . . . + ∧(pai i )
 
∧(d) =
d|n i=1

ra
 

d
k
=
X en
log(pi ) + log(pi ) + . . . + log(pi )
 
m
| {z }
i=1
ai times
ar

k
Dh

X
= ai log (pi )
i=1
N.

Xk
= log (pai i )
B.

i=1
Dr

k
!
Y
= log pai i
i=1
X
∧(d) = log n.
d|n


Now we use Mobius inversion to express ∧(n) in terms of the logarithm.
Theorem 1.10. If n ≥ 1, we have
X n X
∧(n) = µ(d) log =− µ(d) log(d)
d
d|n d|n
X
Proof. We know Mobius inversion formula, if f (n) = g(d), then
d|n
X n
(1.14) g(n) = µ(d) f .
d
d|n

The above Theorem (1.9)


X
log n = ∧ (d).
d|n
Theory of Numbers 15

Comparing the above two equations, we get


X n
∧(n) = µ(d) log (i.e., Replace f = log & g = ∧ in equation (1.14))
d
d|n
X
= µ(d) [log (n) − log (d)]
d|n
X X
= µ(d) log (n) − µ(d) log (d)
d|n d|n
X X
= log (n) µ(d) − µ(d) log (d)
d|n d|n
X
= log (n) I(n) − µ(d) log (d)
d|n
X
= 0− µ(d) log (d) [∵ I(1) log(1) = 1 · 0 = 0 & n > 1, I(n) log n = 0 · log n = 0.]
d|n

ra
X
∧(n) = − µ(d) log (d).

d
d|n
en
m

ar
Dh

1.1. Multiplicative Functions. We have already noted that the set of all arith-
metical functions f with f (1) 6= 0 forms an abelian group under Dirichlet multipli-
N.

cation. IN this section we discuss an important subgroup of this group, the so called
multiplicative function.
B.

Definition 1.7. An arithmetical function f is called multiplicative if f is not


Dr

identically zero and if


f (mn) = f (m)f (n), whenever (m, n) = 1.
A multiplicative function f is called completely multiplicative if we also have
f (mn) = f (m)f (n), ∀ m, n.
Example:
(1) Let fα (n) = nα , where α is a fixed real or complex number.

fα (m n) = (m n)α = mα nα = fα (m) fα (n).


This function is completely multiplicative. In particular, the unit function
u = f0 is completely multiplicative.
f0 (m n) = (m n)0 = 1 = m0 n0 = f0 (m) f0 (n)
we denote the function fα by N α and call it the power function.
1
(2) The identity function I(n) = is completely multiplicative function.
n
I(n) 6= 0 is not identically zero. If m, n ∈ N, then
 
1
I(m n) =
n
16 Dr. B. N. Dharmendra

• If m = 1 and n = 1, then

I(mn) = I(1) = 1 = I(m) I(n).

• If m > 1 or n > 1, then

I(m n) = 0 = I(m) I(n). (I(m) = 0 or I(n) = 0.)

I(m n) = I(m) I(n), ∀ m, n.


∴ I(m n) = I(m) I(n), ∀ m, n. is completely multiplicative function.
(3) The Mobius function µ is multiplicative but not completely multiplicative.
• This is easily seen from the definition of µ(n).If either m or n is a prime
square factor the so does m n and both µ(m n) and µ(m), µ(n) are zero.

i.e., µ(m n) = 0 = µ(m) µ(n).

ra
Consider two relatively prime integers m and n. If neither has a square

d
function write m = p1 , p2 , . . . ps and n = q1 , q2 , . . . qt , where pi and qi
en
are distinct primes. Then
m
µ(m) = (−1)s , µ(n) = (−1)t and µ(m n) = (−1)s+t = µ(m) µ(n).
ar
Dh

This show that µ is multiplicative.


N.

i.e., µ(m n) = µ(m) µ(n) (m, n) = 1,


B.

• µ is not completely multiplicative, since


Dr

µ(4) = µ 22 = 0 but µ(2) µ(2) = (−1) (−1) = 1




i.e., µ(m n) 6= µ(m) µ(n) ∀ m, n.


(4) The Euler function φ(n) is multiplicative (See properties of Euler (iii)).

φ(m n) = φ(m) φ(n), (m, n) = 1.

But it is not completely multiplicative . Since

φ(4) = 2, where as φ(2) · φ(2) = 1 · 1 = 1.

i.e., φ(m n) 6= φ(m) φ(n) ∀ m, n.


(5) The ordinary product f g of two arithmetical functions is defined by the usual
formula

(f g) (m n) = f (m n) g(m n)
= f (m) f (n) g(m) g(n)
= f (m) g(m) f (n) g(n)
(f g) (m n) = (f g) (m) (f g) (n).
Theory of Numbers 17

Similarly, the quotient fg is defined by the formula


 
f f (m n)
(m n) =
g g(m n)
f (m) f (n)
=
g(m) g(n)
f (m) f (n)
=
g(m) g(n)
     
f f f
(m n) = (m) (n).
g g g
Theorem 1.11. If f is multiplicative function, then f (1) = 1.
Proof. Given f is multiplicative
f (ab) = f (a) f (b), (a, b) = 1

ra
If a = n and b = 1, then

d
f (n 1) = f (n) f (1)
en
m
f (n) = f (n) f (1)
ar

1 = f (1) or
Dh

f (1) = 1.
N.


B.

Theorem 1.12. Given f (1) = 1, then


Dr

(1) f is multiplicative if and only if


f (pa11 pa22 · · · par r ) = f (pa11 ) f (pa22 ) · · · f (par r ) ∀ pi , and ai ≥ 1.(1 ≤ i ≤ r)
(2) If f is multiplicative, then f is completely multiplicative if and only if
f (pa ) = [f (p)]a ∀ p, and a ≥ 1.
Proof. (1) If r = 2, then
L.H.S = f (pa11 pa22 )
= f (pa11 ) f (pa22 )
L.H.S = R.H.S.
(2) If r = 1, then
L.H.S = f (pa11 ) [∵ a1 > 1 and a1 = 2]
= f p21


= f (p1 p1 )
= f (p1 ) f (p1 )
= [f (p1 )]2
L.H.S = R.H.S.

18 Dr. B. N. Dharmendra

Multiplicative Functions and Dirichlet Multiplication.


Theorem 1.13. If f and g are multiplicative, so is their Dirichlet product f ∗ g.
Proof. Let h = f ∗ g
X n
h(n) = (f ∗ g) (n) = f (d) g
d
d|n

If n = 1 in the above identity becomes


X n
h(1) = (f ∗ g) (1) = f (d) g = f (1) g(1) = 1
d
d|1

⇒ h(1) = 1. Hence h is multiplicative.


If (m, n) = 1, then
X  mn 
h(mn) = f (c) g
c
c|mn

ra
Now every divisor c of mncan beexpressed in the form c = ab, where a|m and b|n.

d
m n en
Moreover (a, b) = 1 ⇒ , = 1 and there is a one to one correspondence
a b
m
between the set of product ab and the divisors c of mn.
ar

Hence
Dh

X  mn 
h(mn) = f (ab) g
ab
a|m b|n
N.

X m n
= f (a) f (b) g g
B.

a b
a|m b|n
m X n
Dr

X
= f (a) g f (b) g
a b
a|m b|n
h(mn) = h(m) h(n)
Hence h is multiplicative. 

Note:
The Dirichlet product of two completely multiplicative functions need not to be
completely multiplicative.
Theorem 1.14. If both g and f ∗ g are multiplicative, then f is a multiplicative.
Proof. We shall assume that f is not multiplicative and deduce that f ∗ g is also not
multiplicative.
Let h = f ∗ g. Since f is not multiplicative there exits positive integers m and n
with (m, n) = 1, such that
f (mn) 6= f (m) f (n).
We choose such a pair m and n for which the product mn is an small as possible.
Case(i): If mn = 1, then
f (1) 6= f (1) f (1),

(1.15) i.e., f (1) 6= 1.


Theory of Numbers 19

h(n) = (f ∗ g) (n)
X n
h(n) = f (d) g
d
d|n
If n = 1  
X 1
h(1) = f (d) g
d
d|1
h(1) = f (1) g(1)
h(1) = f (1) 6= 1 [Since g is multiplicative, then g(1) = 1 and by equation (1.15)]
h(1) 6= 1.
this shows that h is not multiplicative.
Case(ii): If mn > 1, then we have
f (ab) = f (a) f (b) ∀ positive integers a and b with (a, b) = 1 and ab < mn.
Now we argue as in the proof of Theorem (1.13) except that in the sum defining

ra
h(mn), we separate the terms carry to a = m and b = n, then we have

d
X  mn  en
h(mn) = f (ab) g
m
ab
ab|mn
ar

X  mn 
Dh

= f (ab) g + f (mn) g(1)


ab
a|m b|n ab<mn
N.

X m n
= f (a) f (b) g g + f (mn)
a b
B.

a|m b|n ab<mn


X m X n
= f (a) g f (b) g − f (m) f (n) + f (mn)
Dr

a b
a|m b|n a|m b|n
h(mn) = h(m) h(n) + [f (mn) − f (m) f (n)]

Since f (mn) − f (m) f (n) 6= 0 this shows that


h(mn) 6= h(m) h(n)
So h is not multiplicative, this is contradiction.
∴ h = f ∗ g is multiplicative.
Hence f is multiplicative. 
Theorem 1.15. If g is multiplicative, then g −1 is also multiplicative.
Proof. This follows at once from the Theorem (1.14)
If both g and f ∗ g are multiplicative, then f is a multiplicative.
put f = g −1 in the above statement, then
If both g and g −1 ∗ g = I are multiplicative, then g −1 is a multiplicative.

The inverse of a completely multiplicative function:
The Dirichlet inverse of a completely multiplicative function especially easy to
determine.
20 Dr. B. N. Dharmendra

Theorem 1.16. Let f be multiplicative, then f is completely multiplicative if and


only if f −1 (n) = µ(n) f (n) for all n ≥ 1.

Proof. Let g(n) = µ(n) f (n)


If f is completely multiplicative, we have
X n
(g ∗ f ) (n) = µ(d) f (d) f
d
d|n
X  n
= µ(d) f d
d
d|n
X
= µ(d)
d|n
X
= f (n) µ(d)
d|n
= f (n) I(n)

ra
(g ∗ f ) (n) = I(n).

d
en
Since f (1) = 1 and I(n) for n > 1.
m
ar

g ∗ f = I ⇒ g = f −1 .
Dh

Conversely, assume that f −1 = µ(n) f (n)


N.

To show that f is completely multiplicative it suffices to prove that f (pa ) = [f (p)]a


for prime powers. The equation f −1 (n) = µ(n) f (n) implies that
B.

X n
Dr

(g ∗ f ) (n) = µ(d) f (d) f


d
d|n
X  n
I(n) = µ(d) f (d) f
d
d|n
For n > 1, then
X n
0 = µ(d) f (d) f or
d
d|n
X n
µ(d) f (d) f = 0
d
d|n
Hence, taking n = pa , we have
pa pa
   
µ(1) f (1) f + µ(p) f (p) f + . . . + µ(pa ) f (pa ) f (1) = 0
1 p
(1) (1) f (pa ) + (−1) f (p) f pa−1 = 0 ∵ µ p2 = · · · = µ (pa ) = 0
  

f (pa ) − f (p) f pa−1 = 0




or

f (pa ) = f (p) f pa−1



(1.16)
Theory of Numbers 21

By using iteration, replace a by a − 1 in the above equation (1.16), we get

f pa−1
= f (p) f pa−2
 

..
.
2
= f (p) f p1
 
f p
Substituting the above identities in equation (1.16), we obtain
f (pa ) = f (p) f (p) . . . f (p)
| {z }
a times
f (pa ) = [f (p)]a

Hence f is completely multiplicative. 

Example:
The inverse of Euler’s φ function. We that the relation between Euler’s and Mobius

ra
function is

d
φ(n) =
X
µ(d)
n en
d
m
d|n
ar

X n
φ(n) = µ(d) N [∵ N (n) = n]
Dh

d
d|n
φ = µ∗N
N.

φ −1
= (µ ∗ N )−1
B.

φ−1 = N −1 ∗ µ−1
Dr

φ−1 = µ N ∗ u ∵ µ−1 = u and f −1 = µ f ⇒ N −1 = µ N


 

φ−1 (n) = (µ N ∗ u) (n)


X n
φ−1 (n) = (µ N ) (d) u
d
d|n
X
φ−1 (n) = µ(d) N (d) 1 [∵ u(n) = 1]
d|n
X
−1
φ (n) = µ(d) d [∵ N (n) = n] or
d|n
X
φ−1 (n) = d µ(d)
d|n

Theorem 1.17. If f is multiplicative, we have


X Y
µ(d) f (d) = [1 − f (p)]
d|n p|n

X
Proof. Let g(n) = µ(d) f (d)
d|n
Then g is multiplicative (since µ and f are multiplicative), so to determine g(n) it
22 Dr. B. N. Dharmendra

sufficient to compute g (pa ). But


X
g (pa ) = µ(d) f (d)
d|pa

= µ(1) f (1) + µ(p) f (p) + µ(p2 ) f (p2 ) + . . . + µ(pa ) f (pa )


∵ µ(p2 ) = · · · = µ(pa ) = 0, f and µ are multiplicative
 
= 1 + (−1) f (p)
g (pa ) = 1 − f (p)
If n = a pa , then
Q
!
Y
a
g(n) = g p
a
Y
= g (pa ) [since g is multiplicative]
a
Y
g(n) = [1 − f (p)] .

ra
p|n

d
en 
m
Liouville’s Function λ(n)
ar

An important example of a completely multiplicative function is Liouville’s function


Dh

λ(n), which is defined as follows


N.

Definition 1.8.

B.

1, if n = 1
(1.17) λ(n) =
(−1)a1 +a2 +···+ak , if n = pa11 pa22 · · · pakk .
Dr

The definition shows at once that λ is completely multiplicative.

If m = pa11 pa22 · · · pakk and n = q1b1 q2b2 · · · qkbk , then

λ(mn) = (−1)a1 +a2 +···+ak +b1 +b2 +···+bk


= (−1)a1 +a2 +···+ak (−1)+b1 +b2 +···+bk
λ(mn) = λ(m) λ(n).
Example:

(1) If n = 2, then λ(2) = (−1)1 = −1


(2) If n = 3, then λ(3) = (−1)1 = −1
(3) If n = 4, then λ(22 ) = (−1)2 = 1
(4) If n = 6, then λ(2 · 3) = (−1)1+1 = (−1)2 = 1

n 1 2 3 4 5 6 7 8 9 10
λ(n) 1 −1 −1 1 −1 1 −1 −1 1 1
µ(n) 1 −1 −1 0 −1 1 −1 0 0 1

By above table λ(n) = µ(n), if n is no power factor.


Theory of Numbers 23

Theorem 1.18. For every n ≥ 1, we have



X 1, n is a square
λ(d) =
0, otherwise.
d|n

Also,

λ−1 (n) = |µ(n)| , ∀ n.

Proof.
X
Let g(n) = λ(d)
d|n
X n
= λ(d) u
d

ra
d|n

d
g(n) = (λ ∗ u) (n) or en
g = λ∗u
m
ar

Since λ and u are multiplicative, then λ ∗ u is multiplicative. Hence g is multiplica-


Dh

tive.
N.

i.e., g(mn) = g(m) g(n), (m, n) = 1.


B.

So to determine g(n). We need only compute g (pa ) for prime power, we have
Dr

X
g (pa ) = λ(d)
d|pa

= 1 + λ(p) + λ p2 + · · · + λ (pa )


= 1 − 1 + 1 · · · + (−1)a

a 0, if a is odd
g (p ) =
1, if a is even

k
Y
Hence if n = pai i , we have
i=1

k
!
Y
g (n) = g pai i
i=1
k
Y
= g (pai i ) [∵ g is multiplicative]
i=1

1, if ai are even or n is a square
g (n) =
0, if ai is odd or n is a not square
24 Dr. B. N. Dharmendra

We know that
λ−1 (n) = (µ λ) (n) ∵ f −1 = µ f
 

n = pa1i · · · pakk
 
= µ(n) λ(n)

 1, n=1
−1
λ (n) = 1, n is no power f actor.
0, otherwise


 1, n=1
µ2 (n) = (−1)2k = 1, n is square f ree
0, otherwise


 1, n=1
|µ(n)| = (−1)k = 1, n is square f ree
0, otherwise

∴ λ−1 = µ λ = µ2 = |µ(n)| .

ra


d
The Divisor function σα (n) :
en
m
Definition: For real or complex α and any integer n ≥ 1, we define
ar

X
σα (n) = dα
Dh

d|n
N.

The sum of the αth powers of the divisors of n. The function σα (n) so defined are
called divisor functions. They are multiplicative, because
B.

X
σα (n) = dα
Dr

d|n
X n
= N (dα ) u [∵ N (n) = n & u(n) = 1]
d
d|n
σα (n) = (N α ∗ u) (n)

Since N α and u are multiplicative functions, then N α ∗ u is also multiplication.


Hence σα is multiplication.
[∵ N (n) = n is multiplicative function, then N α (mn) = (mn)α = mα mα = N α (m)N α (n), (m, n) = 1.],
When α = 0, σ0 (n) is the number of divisors of n. This is often denoted by d(n).
When α = 1, σ1 (n) is the number of divisors of n. This is often denoted by σ(n).
Since σα (n) is multiplicative, we have
σα pa11 pa22 · · · pakk = σα (pa11 ) σα (pa22 ) · · · σα pakk
 

To compute σα (pa ) . We note that divisors of a prime power pa are 1, p, p2 , . . . , pa ,


hence
σα (pa ) = 1α + pα + p2 α + . . . + pa α
= 1α + pα + (pα )2 + . . . + (pα )a
pα (a+1) − 1
σα (pa ) =
pα − 1
Theory of Numbers 25

If n = pa11 pa22 · · · pakk , then

k
!
Y
σα (n) = σα pai i
i=1
k
Y
= σα (pai i ) [∵ σα is a multiplicative]
i=1
k α (ai +1)
!
Y pi −1
σα (n) = α .
pi − 1
i=1

If α = 0, then

σ0 (pa ) = 10 + p0 + p2·0 + . . . + pa·0

ra
= 1 + |1 + 1 +
{z. . . + 1}

d
a times en
σ0 (pa ) = 1 + a.
m
ar

If n = pa11 pa22 · · · pakk , then


Dh
N.

k
!
Y
σ0 (n) = σ0 pai i
B.

i=1
k
Dr

Y
= σ0 (pai i ) [∵ σα is a multiplicative]
i=1
Yk
σ0 (n) = (1 + ai ) .
i=1

Theorem 1.19. For n ≥ 1, we have

X n
σα−1 (n) = dα µ(d) µ .
d
d|n

Proof. We know that

σα = N α ∗ u
σα−1 = (N α ∗ u)−1
= u−1 ∗ (N α )−1
= µ ∗ µ Nα
σα−1 = µ N α ∗ µ [∵ Dirichlet product is commutative]
26 Dr. B. N. Dharmendra

σα−1 (n) = (µ N α ∗ µ) (n)


X n
= (µ N α ) (d) µ
d
d|n
X  n
= µ(d) N α (d) µ
d
d|n
X n
= µ(d) dα µ [∵ N (n) = n.]
d
d|n
X n
σα−1 (n) = dα µ(d) µ .
d
d|n

d ra
The Bell series of an arithmetical function:
en
m
E.T.Bell used formal power series to study properties of multiplicative arithmetical
ar

functions.
Dh

Definition 1.9. Given an arithmetical function f and a prime p. We denote by


N.

fp (x) the formal power series


B.


Dr

X
(1.18) fp (x) = f (pn ) xn
n=0

and call this the Bell series of f modulo p. Bell series are especially useful when f
is multiplicative.

Example:

(1) Mobius Function µ(n) of a Bell series:


Since µ(p) = −1 and µ(pn ) = 0 ∀ n > 1, We have


X
µp (x) = µ (pn ) xn
n=0
= µ p0 x0 + µ p 1 x1 + µ p2 x2 + · · · + µ (pn ) xn + · · ·
  

2 n
= µ (1) · 1 + µ (p) · x +
| 0 · x + · · ·{z+ 0 · x + · ·}·
=0
= 1 · 1 + (−1) · x
µp (x) = 1 − x.
Theory of Numbers 27

(2) Euler Function φ(n) of a Bell series:


Since φ (pn ) = pn − pn−1 ∀ n > 1, We have


X
φp (x) = φ (pn ) xn
n=0

X
= 1+ φ (pn ) xn
n=1
X∞
pn − pn−1 xn
 
= 1+
n=1
X∞ ∞
X
n n
= 1+ p x − pn−1 xn
n=1 n=1

X ∞
X
= pn xn − pn xn+1 [∵ Replace n by n + 1 in the second summation]

ra
n=0 n=0

d
X∞ X∞ en
= pn xn − x p n xn
m
n=0 n=0
ar


X
(px)n
Dh

= (1 − x)
n=0
1−x
N.

φp (x) = . [∵ By using Geometric series]


1 − px
B.
Dr

(3) Completely Multiplicative Functions of a Bell series:


If f is completely multiplicative function, then
i.e., f (pn ) = [f (pn )]n , ∀ n ≥ 0, so the Bell series fp (x) is a geometric series


X
fp (x) = f (pn ) xn
n=0
X∞
= [f (p)]n xn
n=0
X∞
= [f (p) x]n
n=0


X 1
(1.19) fp (x) = f (pn ) xn = .
1 − f (p) x
n=0

In particular, we have the following Bell series for the identity function I,
the unit function u, the power function N α and Liouville’s function λ.
28 Dr. B. N. Dharmendra

(4) Identity Functions I(n) of a Bell series:


X
Ip (x) = I (pn ) xn
n=0
1
= [By using equation (1.19)]
1 − I(p) x
1
= [∵ I(p) = 0]
1−0 · x
Ip (x) = 1.

(5) Unit Functions u(n) of a Bell series:

ra
X
up (x) = u (pn ) xn

d
n=0 en
1
= [By using equation (1.19)]
m
1 − u(p) x
ar

1
= [∵ u(p) = 1 ∀ p]
Dh

1−1 · x
1
up (x) = .
N.

1−x
B.

(6) Power Functions N α (n) of a Bell series:


Dr


X
Npα (x) = N α (pn ) xn
n=0
1
= [By using equation (1.19)]
1 − N α (p) x
1
Npα (x) = . [∵ N (n) = n ⇒ N α (p) = pα , ∀ p]
1 − pα x

(7) Liouville’s Functions λ(n) of a Bell series:


X
λp (x) = λ (pn ) xn
n=0
1
= [By using equation (1.19)]
1 − λ(p) x
1
λp (x) = . [∵ λ(p) = −1 ∀ p]
1+x
Theory of Numbers 29

(8) Since µ2 (n) = λ−1 (n) the Bell series of µ2 modulo p.


X∞
2
µp (x) = µ2 (pn ) xn
n=0
2 0
x0 + µ 2 p 1 x1 + µ2 p2 x2 + · · · + µ2 (pn ) xn + · · ·
  
= µ p
= µ2 (1) · 1 + µ2 (p) · x + 2 n
| 0 · x + · · ·{z+ 0 · x + · ·}·
=0
2
= 1 · 1 + (−1) · x
µ2p (x) = 1 + x.
Uniqueness Theorem

Theorem 1.20. Let f and g be multiplicative functions. Then f = g if and only if


fp (x) = gp (x) for all primes.
Proof. If f = g, then f (pn ) = g (pn ) ∀ p and all n ≥ 0.

d ra

fp (x) =
X
f (pn ) xn
en
m
n=0

ar

X
= g (pn ) xn
Dh

n=0
fp (x) = gp (x).
N.

Conversely, if fp (x) = gp (x), then


B.

fp (x) = gp (x)
Dr


X X∞
f (pn ) xn = g (pn ) xn
n=0 n=0
f (p ) = g (pn )
n

k
Y
If n = pαi i , then
i=1
k
!
Y
f (n) = f pαi i
i=1
k
Y
= f (pαi i ) [∵ f is multiplication]
i=1
Yk
= g (pαi i )
i=1
k
!
Y
= g pαi i [∵ g is multiplication]
i=1
f (n) = g(n).
Hence the proof. 
30 Dr. B. N. Dharmendra

Bell series and Dirichlet multiplication


The next theorem relates multiplication of Bell series to Dirichlet multiplication.

Theorem 1.21. For any two arithmetical function f and g. Let h = f ∗ g,


then for every prime p we have

hp (x) = fp (x) gp (x).

Proof. Since the divisors of pn are 1, p, p2 , . . . , pn , we have

h (pn ) = (f ∗ g) (pn )
 n
X p
= f (d) g
n
d
d|p
n
pn
X  
n k
h (p ) = f (p ) g
pk
k=0

d ra
n
X  en

(1.20) h (pn ) = f (pk ) g pn−k .
m
k=0
ar

Consider
Dh


X
hp (x) = h(pn ) xn
N.

n=0
B.

∞ X
n
Dr

X  
(1.21) hp (x) = f (pk ) g pn−k xn [from (1.20)]
n=0 k=0

Consider
∞ ∞
! !
X X
n n n n
fp (x) gp (x) = f (p ) x g(p ) x
n=0 n=0

X n
X  
n n n
fp (x) gp (x) = c(p ) x , where c(p ) = f (pk ) g pn−k
n=0 k=0

∞ X
X n  
(1.22) fp (x) gp (x) = f (pk ) g pn−k xn
n=0 k=0

From the above equations (1.21) and (1.22), we get

hp (x) = fp (x) gp (x).

Example:
Theory of Numbers 31

(1) Since σα = N α ∗ u the Bell series of σα modulo p is

(σα )p (x) = Npα (x) up (x)


1 1
= ·
1 − pα x 1 − x
1
(σα )p (x) =
1 − x (1 + pα ) + pα x2
X
(2) Show that 2ν(n) = µ2 (d)
d|n
This example illustrates how Bell series can be use discover identities involv-
ing arithmetical functions.

0 n=1
(1.23) f (n) = 2ν(n) , where ν(n) =
k n = pa11 pa22 · · · pakk .

Here f (1) = 2ν(1) = 20 = 1.

ra
If m = pa11 pa22 · · · pakk and n = q1b1 q2b2 · · · qsbs , then

d
en
f (mn) = 2ν(m+n)
m
ar

= 2k+s
Dh

= 2 k + 2s
= 2ν(m) + 2ν(n)
N.

f (mn) = f (m) f (n), (m, n) = 1.


B.

Hence f is multiplicative and its Bell series modulo p is


Dr


X
fp (x) = f (pn ) xn
n=0

n)
X
= 2ν(p xn
n=0

n)
X
= 1+ 2ν(p xn
n=1
X∞
= 1+ 21 xn
n=1

X
= 1+2 xn
n=1
x
= 1+2
1−x
1+x
fp (x) =
1−x
1
fp (x) = (1 + x) ·
1−x
fp (x) = up (x) · µp (x).
32 Dr. B. N. Dharmendra

By using above Theorem (1.24) which implies f = µ2 ∗ u.

∴ f (n) = µ2 ∗ u (n)

X n
= µ2 (d) u
d
d|n
X
2ν(n) = µ2 (d). [∵ u(n) = 1 ∀ n & equation (1.23)]
d|n

Derivatives of arithmetical functions:

Definition 1.10. For any arithmetical function f , we define its derivative f 0 to be


the arithmetical function given by the equation

(1.24) f 0 (n) = f (n) log(n) f orall n ≥ 1.

Example:

ra
d
(1) Derivative of Identity function en
m
ar

I 0 (n) = I(n) log(n) f orall n ≥ 1


Dh

= I(1) log(1) + I(n) log(n) f orall n > 1


I(1) · 0 + 0 · log(n) [log(1) & I(n) = 0 n > 1]
N.

=
0
I (n) = 0.
B.

(2) Derivative of unit function


Dr

u0 (n) = u(n) log(n), f orall n ≥ 1


= log(n) [∵ u(n) = 1 ∀ n]
X
= ∧(d)
d|n
X n
= ∧(d) u
d
d|n

u0 (n) = (∧ ∗ u) (n) or
u0 = ∧ ∗ u.

This concept of derivative shares many of the properties of the ordinary derivative
discussed in elementary calculus for example, the usual rules for differentiating sums
and products also hold if and product are Dirichlet products.

Theorem 1.22. If f and g are arithmetical functions, we have


(1) (f + g)0 = f 0 + g 0 ,
(2) (f ∗ g)0 = f 0 ∗ g + f ∗ g 0 ,
0
(3) f −1 = −f 0 ∗ (f ∗ f )−1 .
Theory of Numbers 33

Proof. The proof of (1) :


(f + g)0 (n) = (f + g) (n) log(n)
= f (n) log(n) + g(n) log(n)
(f + g)0 (n) = f 0 (n) + g 0 (n) or
(f + g)0 = f 0 + g 0 .
The proof of (2) :
(f ∗ g)0 (n) = (f ∗ g) (n) log(n)
X n  n
= f (d) g log d ·
d d
d|n
X n h  n i
= f (d) g log (d) + log
d d
d|n
X  n  X n n
= f (d) log (d) g + f (d) g log
d d d
d|n d|n

ra
n X n

d
X
= f 0 (d)0 g + f (d) g 0
en
d d
d|n d|n
m
(f ∗ g)0 (n) = f 0 ∗ g (n) + f ∗ g 0 (n) or
 
ar

(f ∗ g)0 = f 0 ∗ g + f ∗ g 0 .
Dh

The proof of (3) : We know that


N.

f ∗ f −1

= I
B.

differentiate the above equation, we get


0
f ∗ f −1 = I0
Dr

0
f 0 ∗ f −1 + f ∗ f −1 ∵ I 0 (n) = 0
 
= 0
0
f ∗ f −1 = −f 0 ∗ f −1
Multiplying by f −1 in the above equation, we get
 0 
f −1 ∗ f ∗ f −1 = −f −1 ∗ f 0 ∗ f −1


0
f −1 ∗ f ∗ f −1 = − f −1 ∗ f 0 ∗ f −1
 
[∵ commutative law]
0
I ∗ f −1 = − f 0 ∗ f −1 ∗ f −1

0
f −1 = −f 0 ∗ f −1 ∗ f −1

0
f −1 = −f 0 ∗ (f ∗ f )−1 . ∵ (f ∗ g)−1 = g −1 ∗ f −1
 


Selberg Identity
Using the concept of derivative we can quickly derive a formula of Selberg which is
sometime used as starting point of an elementary proof of the prime theorem.
Theorem 1.23. The Selberg identity. For n ≥ 1, we have
X n X n
(1.25) ∧(n) log(n) + ∧(d) ∧ = µ(d) log2 .
d d
d|n d|n
34 Dr. B. N. Dharmendra

or

∧0 + ∧ ∗ ∧ = µ ∗ u00
∵ u0 (n) = log(n) & u00 (n) = u0 (n) log(n) = log(n) log(n) = log2 (n).
 

Proof. We know that


∧ ∗ u = u0
differentiate the above equation, we get
0
(∧ ∗ u) = u00
∧0 ∗ u + ∧ ∗ u0 = u00
∧0 ∗ u + ∧ ∗ ∧ ∗ u = u00
Multiplying both side by µ in the above equation, we get
∧ ∗ u ∗ µ + ∧ ∗ ∧ ∗ u ∗ µ = u00 ∗ µ
0

∧0 ∗ I + ∧ ∗ ∧ ∗ I = µ ∗ u00 [∵ Dirichlet product is commutative and u ∗ µ = I]

ra
0 00
∧ +∧∗∧ = µ∗u .

d
en 
m
ar
Dh
N.
B.
Dr

You might also like