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

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

MOCK-MODULAR FORMS OF WEIGHT ONE 11<br />

c n,u = 0 for u ≤ 0. Applying the Rankin-Selberg unfolding trick to 〈 ˜Q m (·, s), e n (·)〉 gives us<br />

∫ ∞<br />

〈 ˜Q m (·, s), e n (·)〉 = (4π|m|) 1/2 e −2π|m|y (4π|m|y) s−1 c n,m W sgn(m)<br />

0<br />

= (4π|m|) 1/2 Γ ( s − 1<br />

c − it ) (<br />

2 n Γ s −<br />

1<br />

+ it )<br />

2 n<br />

(21)<br />

n,m ( )<br />

Γ s − sgn(m)<br />

2<br />

2 ,s n− 1 2<br />

(4π|m|y) dy<br />

y<br />

The last step uses the Mellin transform <strong>of</strong> the W -Whittaker function [2, eq. (8b)] and<br />

the substitution s n = 1/2 + it n . When Re(s) > 1, the sum defining D(z, s) is absolutely<br />

convergent [35, Satz 8.1 ], since y 1/2 Q m (z, s) ∈ ˜D 1 (M, ν) for Re(s) > 1. When 0 < Re(s) ≤ 1,<br />

we can write D(z, s) as<br />

D(z, s) = (4π|m|) ∑ 1/2 Γ ( s + 1 − 1<br />

c − it ) (<br />

2 n Γ s + 1 −<br />

1<br />

+ it )<br />

2 n s − sgn(m)<br />

n,m (<br />

)<br />

2<br />

e n (z) ·<br />

n∈N<br />

Γ s + 1 − sgn(m)<br />

(s − 1<br />

2<br />

2 )2 + t 2 n<br />

Since t 2 n = λ n − 1/4 and ∑ n>M λ−2 n converges [35, Satz 8.1], we can apply Cauchy-Schwarz<br />

inequality to see that the sum on the right hand side above converges absolutely on compact<br />

subsets <strong>of</strong> {s ∈ C : Re(s) > 0, s ≠ 1/2}. At s = 1/2, the first R terms in the sum produce a<br />

simple pole since t n = 0 for all 1 ≤ n ≤ R. The rest <strong>of</strong> the sum still converges absolutely. So<br />

the right hand side above gives the analytic continuation <strong>of</strong> D(z, s) to Re(s) > 0.<br />

For the continuous spectrum, the contribution C(z, s) can be treated similarly. For any<br />

cusp ι, we can write the Fourier expansion <strong>of</strong> E ι (z, s) at infinity in the following form (for<br />

ι = ∞, see [20, eq. (76)’])<br />

E ι (z, s) = y s + ψ ι (s)y 1−s + ∑ ψ ι,m (s)W sgn(m)<br />

,s− 1 (4π|m|y)e 2πimx ,<br />

2 2<br />

m≠0<br />

where ψ ι (s) and ψ ι,m (s) are products <strong>of</strong> gamma factors and Selberg-Kloosterman zeta functions.<br />

It is well-known that The Eisenstein series can be analytically continued in s to the<br />

whole complex plane. When Re(s) > 1/2, the poles <strong>of</strong> E ι (z, s) are in the interval s ∈ (1/2, 1]<br />

[35, Satz 10.3 ]. On the line Re(s) = 1/2, E ι (z 0 , s) is holomorphic in s for any fixed z 0 ∈ H<br />

[35, Satz 10.4]. So both ψ ι (s) and ψ ι,m (s) admit analytic continuation to Re(s) > 0 and are<br />

holomorphic on Re(s) = 1/2. Using the same unfolding trick above, we can evaluate<br />

〈 ˜Q m (·, s), E ι (·, 1 + ir)〉 = ψ ( 1 2 ι,m + ir) (4π|m|) 1/2 Γ ( s − 1 − ir) Γ ( s − 1 2 ( ) + ir) 2 .<br />

2<br />

Γ s − sgn(m)<br />

2<br />

Then C(z, s) can be written as<br />

C(z, s) = ∑ ι<br />

(22)<br />

= ∑ ι<br />

∫<br />

1 ∞<br />

4π −∞<br />

∫<br />

1 ∞<br />

4π −∞<br />

(<br />

ψ 1<br />

ι,m + ir) (4π|m|) 1/2 Γ ( s − 1 − ir) Γ ( s − 1 2<br />

( ) + ir) 2<br />

E<br />

2 ι (z, 1 + ir)dr<br />

Γ s − sgn(m)<br />

2<br />

2<br />

(<br />

ψ 1<br />

ι,m + ir) (4π|m|) 1/2 Γ ( s + 1 − 1 − ir) Γ ( s + 1 − 1<br />

(<br />

2<br />

)<br />

+ ir) 2<br />

2<br />

Γ s + 1 − sgn(m)<br />

2<br />

(s − sgn(m) )<br />

2<br />

·<br />

(s − 1 − ir)(s − 1 + ir) · E ι(z, 1 + ir)dr.<br />

2<br />

2 2

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

Saved successfully!

Ooh no, something went wrong!