28.04.2014 Views

Mock-modular forms of weight one - UCLA Department of Mathematics

Mock-modular forms of weight one - UCLA Department of Mathematics

Mock-modular forms of weight one - UCLA Department of Mathematics

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

20 W. DUKE AND Y. LI<br />

we arrive at<br />

(52) 〈g ψ , g ψ 〉 = −H(p)<br />

∑<br />

A∈Cl(F )<br />

Since ψ 2 is non-trivial, we have equation (50).<br />

ψ 2 (A) log( √ y A |η(τ A )| 2 ).<br />

Remarks.<br />

i) By the same procedure, <strong>one</strong> can analyze the inner product between any <strong>weight</strong> <strong>one</strong><br />

new<strong>forms</strong>. In particular, if g(z), h(z) ∈ S 1 (p) arises from different types <strong>of</strong> Galois representations,<br />

then 〈g, h〉 = 0.<br />

ii) This proposition is really a proven case <strong>of</strong> Stark’s conjecture in the abelian, rank <strong>one</strong><br />

case.<br />

Next we prove the existence <strong>of</strong> the special pre-image.<br />

Proposition<br />

∑<br />

4.2. Let ψ be a non-trivial character <strong>of</strong> Cl(F ). Then there exists ˜g ψ (z) =<br />

n≥n 0<br />

r + ψ (n)qn ∈ M − 1 (p) with shadow g ψ (z) such that<br />

(i) When χ p (n) = 1 or n < − p+1 , the coefficient 24 r+ ψ<br />

(n) equals to zero.<br />

(ii) For n ≤ 0, the coefficients r + ψ<br />

(n) are <strong>of</strong> the form<br />

r + ψ<br />

(n) = −β<br />

∑<br />

A∈Cl(F )<br />

ψ 2 (A) log |u(n, A)|,<br />

where β ∈ Q depends only on p, and u(n, A) ∈ O × H<br />

depends only on n and A.<br />

(iii) Let σ C ∈ Gal(H/F ) be the element associated to the class C ∈ Cl(F ) via Artin’s isomorphism.<br />

Then it acts on the units u(n, A) by<br />

σ C (u(n, A)) = u(n, AC −1 ).<br />

Pro<strong>of</strong>. From Lemma 3.3, we know that the dimension <strong>of</strong> S + 1 (p), denoted by d + , is 1 2 (H(p) −<br />

1)+d − . Let {g 1 , g 2 , . . . , g d+ } be a basis <strong>of</strong> S + 1 (p). Suppose we have chosen n 1 , n 2 , . . . , n d+ > 0<br />

such that χ p (n k ) ≠ −1 and the matrix<br />

P := [c(g j , n k )] 1≤j,k≤d+<br />

is invertible. Then for any ψ, there exists ˜g ψ (z) ∈ M − 1 (p) with shadow g ψ (z) and principal<br />

part coefficients r + ψ (−n) = 0 for all positive integer n not in the set {n 1, n 2 , . . . , n d+ } by<br />

Proposition 3.5. Furthermore, these r + ψ<br />

(−n) can be uniquely determined from solving a<br />

d + × d + system <strong>of</strong> equations. We will show that such ˜g ψ (z) can be made to satisfy the<br />

proposition.<br />

First, we can choose {n j : 1 ≤ j ≤ d + } such that n j ≤ (p + 1)/24. Let {g 1 , g 2 , . . . , g d+ } to<br />

be a q-echelon basis. Then these n j ’s can be chosen to be bounded by the supremum <strong>of</strong> the<br />

set<br />

{ord ∞ h(z) : h(z) ∈ S 1 + (p)}.<br />

Now, the square <strong>of</strong> a cusp form in S 1 (p) is in S 2 (p), the space <strong>of</strong> <strong>weight</strong> 2 cusp form <strong>of</strong> trivial<br />

nebentypus on Γ 0 (p). This gives rise to a holomorphic differential form on the <strong>modular</strong> curve<br />

X 0 (p). Atkin showed in [1] that ∞ is not a Weiestrass point <strong>of</strong> the <strong>modular</strong> curve X 0 (p),<br />

i.e. ord ∞ f(z) is no more than the genus <strong>of</strong> X 0 (p). Using the Riemann-Hurwitz formula (see<br />

[37]), we know that the genus <strong>of</strong> X 0 (p) is bounded by (p + 1)/12. So the n j ’s can be chosen<br />

to be bounded by (p + 1)/24.<br />

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!