Sunday, 26 November 2006

analytic number theory - orthogonality relation for quadratic Dirichlet characters

So I think I solved half of the problem. Suppose that nn is a perfect square. Then chid(n)=1chid(n)=1 unless gcd(d,n)>1gcd(d,n)>1, in which case its 00. So, for gcdd,n=1gcdd,n=1, we are simply pulling out the subset of fundamental discriminants having no common divisor with nn. To quantify the size of this subset, we must first count fundamental discriminants.



The set of fundamental discriminants consists of all square-free integers congruent to 11 modulo 44 (i.e. odd fundamental discriminants) and all such numbers multiplied by 44 and pm8pm8 (i.e. even fundamental discriminants). The odd fundamental discriminants may be counted by considering the series sumtext$d$oddfrac1|d|s.sumtext$d$oddfrac1|d|s. In fact, this is a Dirichlet series. For observe that sumtext$d$oddfrac1|d|s=1+frac13s+frac15s+frac17s+cdots=prodp>2left(1+frac1psright)=fraczeta(s)zeta(2s)left(1+frac12sright)1,sumtext$d$oddfrac1|d|s=1+frac13s+frac15s+frac17s+cdots=prodp>2left(1+frac1psright)=fraczeta(s)zeta(2s)left(1+frac12sright)1,
where fraczeta(s)zeta(2s)=sumin=1nftyfrac|mu(n)|ns,fraczeta(s)zeta(2s)=sumin=1nftyfrac|mu(n)|ns, is the Dirichlet series which generates the square-free numbers. So, by following the definition of fundamental discriminants given above, we can count fundamental discriminants by using the Dirichlet series left(1+frac14s+frac28sright)fraczeta(s)zeta(2s)left(1+frac12sright)1=left(1+frac14s+frac28sright)underbraceprodp>2left(1+frac1psright)lp(s).left(1+frac14s+frac28sright)fraczeta(s)zeta(2s)left(1+frac12sright)1=left(1+frac14s+frac28sright)underbraceprodp>2left(1+frac1psright)lp(s).



Now, to omit those discriminants with gcd(d,m)>1gcd(d,m)>1, we just omit the corresponding factors in lp(s)lp(s). What's missing is prodsubstackp>2 p|mleft(1+frac1psright),prodsubstackp>2 p|mleft(1+frac1psright), so the relative density (compared to all fundamental discriminants dd) is can be quantified by the expression frac1Dcdotprodsubstackp>2 pnmidmleft(1+frac1psright)=prodsubstackp>2 p|mleft(1+frac1psright)1.frac1Dcdotprodsubstackp>2 pnmidmleft(1+frac1psright)=prodsubstackp>2 p|mleft(1+frac1psright)1. As in the proof of the prime number theorem, the main contribution here comes from the simple pole at s=1s=1 (of zeta(s)zeta(s)). In fact, p=2p=2 also fits at s=1s=1 since 1+frac14+frac28=1+frac121+frac14+frac28=1+frac12. Thus, in the end we obtain limXrightarrowinftyfrac1Dsum0<|d|leqXchid(m)=prodp|mleft(1+frac1pright)1,limXrightarrowinftyfrac1Dsum0<|d|leqXchid(m)=prodp|mleft(1+frac1pright)1, as desired.



Now, does anyone know how to explain the other case. I assume it just follows from the symmetry of the roots of unity. But can anyone validate and perhaps clarify this. Thank you.

No comments:

Post a Comment