Exercise 15.5.7

When evaluating the multiplication formula for φ ( αz ) at a complex number z 0 , one needs to worry about poles and vanishing denominators.

(a)
Let α [ i ] be odd, and assume that z 0 is a pole of neither φ ( z ) nor φ ( αz ) . Prove carefully that Q α ( φ 4 ( z 0 ) ) 0 and that φ ( α z 0 ) = i 𝜀 φ ( z 0 ) P α ( φ 4 ( z 0 ) ) Q α ( φ 4 ( z 0 ) ) .

Exercise 9 of Section 5.4 will be useful.

(b)
Let n be odd, and let p be a prime dividing n . Then let β be a Gaussian prime such that p = β if p 3 ( mod 4 ) and p = β β ¯ if p 1 ( mod 4 ) . Use part (a) to prove carefully that φ ( ϖ β ) ( i , φ ( ϖ n ) ) .

Theorem 15.3.2 will be helpful.

(c)
Let n be odd, and let p be a prime such that p 2 divides n . Also define β as in part (b). Prove that φ ( ϖ β 2 ) ( i , φ ( ϖ n ) ) .

Answers

Proof. (a) Assume that z 0 is not a pole of φ ( z ) . We will prove that Q α ( φ 4 ( z 0 ) ) = 0 implies that z 0 is a pole of φ ( αz ) .

Write u 0 = φ ( z 0 ) , and v 0 = u 0 4 = φ 4 ( z 0 ) . Since Q α ( v 0 ) = 0 , there is some polynomial R and some k , k 1 such that Q n ( v ) = ( v v 0 ) k R ( v ) (we have not proved in the text that Q n has no multiple root) and R ( v 0 ) 0 , so that, with v = u 4 ,

Q α ( u 4 ) = ( u 4 u 0 4 ) k R ( u 4 ) , R ( u 0 4 ) = R ( φ 4 ( z 0 ) ) 0 .

For all z such that both members are defined,

φ ( αz ) = i 𝜀 φ ( z ) P α ( φ 4 ( z ) ) Q α ( φ 4 ( z ) ) = i 𝜀 φ ( z ) P α ( φ 4 ( z ) ) ( φ 4 ( z ) φ 4 ( z 0 ) ) k R ( φ 4 ( z ) ) = v ( z ) ( φ 4 ( z ) φ 4 ( z 0 ) ) k ,

where v ( z ) = i 𝜀 φ ( z ) P α ( φ 4 ( z ) ) R ( φ 4 ( z ) ) is analytic in some neighborhood U 1 of z 0 , since z 0 is not a pole of φ , and since R ( φ 4 ( z 0 ) ) 0 .

Moreover, by Exercise 9 of section 5.4, u P α ( u ) and Q α ( u ) have no common roots. Since φ 4 ( u 0 ) is a root of Q α ( u ) , we have φ 4 ( z 0 ) P α ( φ 4 ( z 0 ) ) 0 , so that

v ( z 0 ) 0 .

If z 0 was a zero of φ ( z 0 ) , then φ ( z 0 ) = 0 and Q α ( φ 4 ( z 0 ) ) = Q α ( 0 ) = 1 0 , in contradiction with the hypothesis, thus φ ( z 0 ) 0 .

We write

φ ( z ) = φ ( z 0 ) + ( z z 0 ) l w ( z ) , w ( z 0 ) 0 ,

where w ( z ) is analytic in a neighborhood U 2 of z 0 , and l is the order of the zero z 0 of φ ( z ) φ ( z 0 ) (thus l = 1 or l = 2 by the proof of theorem 15.3.3).

Then

φ 4 ( z ) φ 4 ( z 0 ) = ( φ ( z ) φ ( z 0 ) ) ( φ 3 ( z ) + φ ( z 0 ) φ 2 ( z ) + φ ( z 0 ) 2 φ ( z ) + φ 3 ( z 0 ) ) = ( φ ( z ) φ ( z 0 ) ) s ( z ) = ( z z 0 ) l w ( z ) s ( z ) = ( z z 0 ) l t ( z )

where s ( z ) = φ 3 ( z ) + φ ( z 0 ) φ 2 ( z ) + φ ( z 0 ) 2 φ ( z ) + φ 3 ( z 0 ) and t ( z ) = s ( z ) t ( z ) . Then s is analytic in a neighborhood U 3 of z 0 , since z 0 is not a pole of φ . Moreover s ( z 0 ) = 4 φ 3 ( z 0 ) 0 . Therefore t ( z ) = s ( z ) w ( z ) is analytic in U 2 U 3 , and t ( z 0 ) = s ( z 0 ) w ( z 0 ) 0 .

Since the poles of φ ( αz ) are isolated, there is a neighborhood U 4 of z 0 such that φ ( αz ) is defined on U 4 { z 0 } .

Then, for all z in the neighborhood U = U 1 U 2 U 3 U 4 of z 0 such that z z 0 , since both members are defined in U ,

φ ( αz ) = v ( z ) t k ( z ) 1 ( z z 0 ) kl , z U .

Since v , t are analytic, and t ( z 0 ) 0 , r ( z ) = v ( z ) t k ( z ) is analytic in a neighborhood of z 0 , and

φ ( αz ) = r ( z ) ( z z 0 ) kl .

Moreover, since v ( z 0 ) 0 , r ( z 0 ) = v ( z 0 ) t k ( z 0 ) 0 . This proves that φ ( αz ) has a pole of order kl 1 .

This proves by contraposition that if z 0 is a pole of neither φ ( z ) nor φ ( αz ) , then Q α ( φ 4 ( z 0 ) ) 0 .

Moreover, since z 0 is not a pole of φ ( αz ) , φ ( α z 0 ) is defined, and since Q α ( φ 4 ( z 0 ) ) 0 , where z 0 is not a pole of φ ( z 0 ) , the right member is defined, thus

φ ( α z 0 ) = i 𝜀 φ ( z 0 ) P α ( φ 4 ( z 0 ) ) Q α ( φ 4 ( z 0 ) ) .

(b) In both cases of the sentence, β is a Gaussian prime dividing n . Thus there is a Gaussian integer α such that

n = βα , α [ i ] .

Since n is odd, β and α are odd.

To apply part (a), it remains to verify that z 0 = ϖ n is a pole of neither φ ( z ) nor φ ( αz ) .

φ ( z ) has no real poles, thus ϖ n is not a pole of φ ( z ) .

If ϖ n was a pole of ϖ ( αz ) , then α ϖ n would be a pole of φ ( z ) . The Theorem 15.3.2 shows that

α ϖ n = ( a + ib ) ϖ 2 , a , b  odd .

But then α = n 2 ( a + ib ) [ i ] , because n , a and b are odd. This contradicts α [ i ] , and this contradiction shows that ϖ n is not a pole of φ ( αz ) .

Therefore, by Theorem 15.4.4 and part (a),

φ ( ϖ β ) = φ ( α ϖ n ) = i 𝜀 φ ( ϖ n ) P α ( φ 4 ( ϖ n ) ) Q α ( φ 4 ( ϖ n ) ) .

This shows that

φ ( ϖ β ) ( i , φ ( ϖ n ) ) .

(c) Now p 2 n . Since β is a Gaussian prime dividing p , β 2 divides n in [ i ] , thus there is some α such that

n = β 2 α , α [ i ] .

We know that n and β are odd, thus α is odd.

Since α is odd, the same proof as in part (b) shows that ϖ n is a pole of neither φ ( z ) nor φ ( αz ) .

Therefore, by Theorem 15.4.4 and part (a),

φ ( ϖ β 2 ) = φ ( α ϖ n ) = i 𝜀 φ ( ϖ n ) P α ( φ 4 ( ϖ n ) ) Q α ( φ 4 ( ϖ n ) ) .

This equality shows that

φ ( ϖ β 2 ) ( i , φ ( ϖ n ) ) .

User profile picture
2022-07-19 00:00
Comments