Homepage Solution manuals Ivan Niven An Introduction to the Theory of Numbers Exercise 4.2.25* ($\sum_{\underset{(a,n)=1}{a=1}}^n (a-1, n) = d(n) \phi(n)$)

Exercise 4.2.25* ($\sum_{\underset{(a,n)=1}{a=1}}^n (a-1, n) = d(n) \phi(n)$)

Show that for all positive integers n ,

a = 1 ( a , n ) = 1 n ( a 1 , n ) = d ( n ) ϕ ( n ) .

Answers

Proof. We note that is the product of two multiplicative functions, so is a multiplicative function.

Put

F ( n ) = a [ [ 1 , n ] ] , a n = 1 ( a 1 ) n .

We show that F is also a multiplicative function. Suppose that n m = 1 , where n , m are positive integers. Let A n denote the set

A n = { a [ [ 1 , n ] ] : a n = 1 }

Then

F ( n ) F ( m ) = ( a [ [ 1 , n ] ] , a n = 1 ( a 1 ) n ) ( b [ [ 1 , m ] ] , b m = 1 ( b 1 ) m ) = ( a A n ( a 1 ) n ) ( b A m ( b 1 ) m ) ,

so

F ( n ) F ( m ) = ( a , b ) A n × A m [ ( a 1 ) n ] [ ( b 1 ) m ] . (1)

Consider the map

φ { A n × A m A nm ( a , b ) c ,

where c [ [ 1 , nm ] ] is the unique integer such that

{ c a ( mod n ) , c b ( mod m ) ,

the existence and unicity of such an integer c being in accordance with the Chinese Remainder Theorem.

  • φ is well defined: since a n = 1 and b m = 1 , using c = a + kn for some integer k , c n = ( a + kn ) n = a n = 1 , and similarly c m = 1 , therefore c nm = 1 , so c A nm .
  • φ is injective (one-to-one): if c = φ ( a , b ) = φ ( a , b ) , then a a ( mod n ) . Since 1 a n and 1 a n , we obtain a = a , and similarly b = b .
  • φ is surjective (onto): if c A nm , we define a as the positive remainder of c modulo n , and b as the positive remainder of c modulo m , so that

    c = qn + a , 1 a n , c = q m + b , 1 b m .

    Since c nm = 1 , then a n = ( c qn ) n = c n = 1 , and b m = ( c q m ) m = c m = 1 , thus a A n and b B m . From c a ( mod n ) and c b mod m , we infer c = φ ( a , b ) , where ( a , b ) A n × A m .

In conclusion, φ is a bijection. Moreover, if c = φ ( a , b ) ,

( c 1 ) nm = [ ( c 1 ) n ] [ ( c 1 ) m ] ( since  n m = 1 ) = [ ( a 1 ) n ] [ ( b 1 ) n ] ).

because a 1 c 1 ( mod n ) , b 1 c 1 ( mod m ) .

Using equality (1), this shows that

F ( n ) F ( m ) = ( a , b ) A n × A m [ ( a 1 ) n ] [ ( b 1 ) m ] (2) = c A nm ( c 1 ) nm (3) = F ( nm ) . (4)

(Formally, if we put f ( c ) = ( c 1 ) n , the bijective change of indices c = φ ( a , b ) gives

( a , b ) A n × A m [ ( a 1 ) n ] [ ( b 1 ) m ] = ( a , b ) A n × A m [ ( φ ( a , b ) 1 ) nm ] = ( a , b ) A n × A m f ( φ ( a , b ) ) = c φ ( A n × A m ) f ( c ) = c A nm ( c 1 ) nm ,

so that the sums (2) and (3) contain the same terms in different orders.)

It remains to compare F ( p α ) and d ( p α ) ϕ ( p α ) , where α 0 and p is a prime number.

First d ( p α ) ϕ ( p α ) = ( α + 1 ) ( p α p α 1 ) .

Next, if we group the terms such that ( a 1 ) p α = d , where d is a divisor of p α (so d = p i , 0 i α ),

F ( p α ) = a [ [ 1 , p α ] ] , p a ( a 1 ) p α = d p α ( a [ [ 1 , p α ] ] , p a , ( a 1 ) p α = d d ) = i = 0 α p i ( a [ [ 1 , p α ] ] , p a , ( a 1 ) p α = p i 1 ) ( d = p i ) = i = 0 α b i p i ,

where

b i = Card B i , B i = { a [ [ 1 , p α ] ] : p a , ( a 1 ) p α = p i } .
  • If i = 0 ,

    B 0 = { a [ [ 1 , p α ] ] : p a , p a 1 } = [ [ 1 , p α ] ] ( ( { p , 2 p , , p α 1 p } { 1 , p + 1 , 2 p + 1 , , ( p α + 1 1 ) p + 1 } ) ,

    so

    b 0 = p α 2 p α 1 .

  • If i = α , for every a B α , p α a 1 , where 1 a p α , so a 1 = 0 . Therefore B α = { 1 } (since 0 p α = p α ), and

    b α = 1 .

  • If 0 < i < α , then every a B i satisfies p i a 1 , so a = k p i + 1 for some integer k , where i > 0 , thus p a . Therefore

    B i = { a [ [ 1 , p α ] ] : ( a 1 ) p α = p i } = { a [ [ 1 , p α ] ] : p i a 1 , p i + 1 a 1 } .

    Taking b = a 1

    | B i | = Card { b [ [ 0 , p α [ [ : p i b , p i + 1 b } = Card { b [ [ 0 , p α [ [ : p i b } Card { b [ [ 1 , p α [ [ : p i + 1 b } = p α i p α i 1 ( since  i < α ) .

    So

    b i = p α i p α i 1 ( 0 < i < α ) .

Then

F ( p α ) = b 0 + i = 1 α 1 b i p i + b α p α = p α 2 p α 1 + i = 1 α 1 ( p α p α 1 ) + p α = p α 2 p α 1 + ( α 1 ) ( p α p α 1 ) + p α = ( α + 1 ) ( p α p α 1 ) = d ( p α ) ϕ ( p α ) .

Since F and are multiplicative functions, for all positive integer n , F ( n ) = d ( n ) ϕ ( n ) , so

a [ [ 1 , n ] ] , a n = 1 ( a 1 ) n = d ( n ) ϕ ( n ) .

User profile picture
2025-01-23 16:06
Comments