Math 726: L-functions and modular forms Fall 2011

Week 2, lecture 7: Modular Forms and Congruence Subgroups

Instructor: Henri Darmon Notes written by: Juan Ignacio Restrepo

Recall

Previously we dealt with one dimensional representations of GQ and we established a corre-

spondence, via Class Field Theory, between them and primitive Dirichlet characters. Also,
we mentioned the correspondence between one dimensional representations of GK (where K
is a number field) and primitive Hecke characters, although we said much less about this.
Now we will deal with Artin representations of GQ of dimension greater than 1, focusing
on the two dimensional representations. What object would they be in correspondence with,
analogous to the Dirichlet and Hecke characters in the one dimensional case? The Modular
Forms! So, we are going to talk about Modular Forms.

More on the θ function.

2
For the functional equation of ζ(s) we used the function θ(t) = Pn ∈Z e−πn t , for which the

√
relation θ 1t = tθ(t) holds. In all this analysis, θ : R>0 → R. It turns out to be better to
consider it as a function on the right-half plane C<[s] > 0, θ : C<[s]>0 → C, so for convenience,
we are going to rotate it to the PoincarXupper-half plane H := {z ∈ C : =[z] > 0} by defining
2
the function θ̃(z) = θ(−iz), so θ̃(z) = eπin z .
n∈Z

Lemma 1. For all z ∈ H the following relations hold:

a) θ̃(z) = θ̃(z + 2)

−1 √
= e−2πi/8 z θ̃(z), where the natural branch for the square root is chosen.


b) θ̃ z

Proof. a) is clear. For b), we will check that it is true for the set iR>0 , which has accumulation
points. Since both sides of the equation are analytic functions on H, if they agree on a set
with accumulation points they will agree on the whole domain.  Every z ∈ iR>0 can be
1 √
expressed as iy with y > 0, so the relation reduces to θ = yθ(y), which we know is
y
true.

1
    
1 2 −1 0 −1
We can think of z 7→ z + 2 as z 7→ (z) and z →
7 as z 7→ (z) acting
0 1 z 1 0
as Mbius transformations.
  The two matrices  generate a subgroup of the modular group
a b
SL2 (Z) := : a, b, c, d ∈ Z, ad − bc = 1 . θ̃ has nice transformation properties un-
c d
der this group action, which we will see later.

Subgroups of SL2 Z
The most important subgroups of SL2 (Z) have their own names. They are
      
a b a b ∗ ∗
Γ0 (N ) := ∈ SL2 (Z) : ≡ (mod N )
c d c d 0 ∗
      
a b a b 1 ∗
Γ1 (N ) := ∈ SL2 (Z) : ≡ (mod N )
c d c d 0 1
      
a b a b 1 0
Γ(N ) := ∈ SL2 (Z) : ≡ (mod N ) .
c d c d 0 1

We have the chain of subgroups Γ(N ) / Γ1 (N ) / Γ0 (N ) ≤ SL2 (Z). Note that Γ(N ) is the
kernel of the projection map SL2 (Z) → SL2 (Z/N Z), so Γ(N ) / SL2 (Z). We can identify the
quotients as follows:

SL2 (Z)/Γ0 (N ) = P1 (Z/N Z)

   
a b a b
7→ (a : c) = (∞)
c d c d

Γ0 (N )/Γ1 (N ) = (Z/N Z)×

 
a b
7→ d (or a)
c d

Γ1 (N )/Γ(N ) = Z/N Z
 
a b
7→ b.
c d

The first quotient is not a group, since Γ0 (N ) is, in general, not a normal subgroup of
SL2 (Z). However, there is a left action of SL2 (Z) on the cosets which passes to P1 (Z/N Z).
 
a b
The action of on (A : C) results in (aA + bC : cA + dC). The two morphisms
c d
Γ0 (N ) → (Z/N Z)× and Γ(1) → Z/N Z described earlier are well defined and they have

2
kernels Γ1 (N ) and Γ(N ) respectively. Note that ad ≡ 1 (mod N ) because ad − bc = 1 and
N | c. Also (Z/N Z)× is abelian, so the map g 7→ g −1 is an automorphism. This implies that
 
× a b
the morphism Γ0 (N ) → (Z/N Z) can be defined as 7→ a as well if we want.
c d

Modular Forms
Definition 1. A modular form of weight k, level N and character E is a holomorphic
function f : H → C satisfying
    
a b az + b
f (z) = f = E(d)(cz + d)k f (z) (1)
c d cz + d
 
a b
for all ∈ Γ0 (N ), together with suitable growth conditions at the cusps (which we
c d
will mention later).

Remark: We could have chosen to write E(a) instead. Everything would work the same as
long as we are consistent with our choice.
Note that if N = 1, Γ0 (N ) = SL2 (Z) and all characters are trivial, (1) reduces to
    
a b k a b
f (z) = (cz + d) f (z) ∀ ∈ SL2 (Z).
c d c d

More about these growth conditions at ∞.

 
a b
We know that ∈ Γ0 (N ) so f (z + 1) = f (z), so we have a Fourier expansion for
c d
f (z) as X
f (z) = an q n , q = e2πiz .
n∈Z

This is the so-called q-expansion of f .

Definition 2. We say that f is meromorphic (resp. holomorphic, resp. vanishes) if the

q-expansion of f (z) lies in C((q)) (resp. C[[q]], resp. qC[[q]]).

The values for a modular form of level 1 are predetermined by the values the function
takes on at the region
   
1 1 1
F := z ∈ C : − ≤ <[z] < , kzk > 1 ∪ z ∈ C : − ≤ <[z] ≤ 0, kzk = 1 .
2 2 2

3
 [
A fundamental region for Γ1 (N ) is FΓ1 (N ) := γF , a finite union of hyperbolic
γ∈Γ1 (N )\SL2 (Z)
triangles whose vertices lie in P1 (Q). This construction is not particular for Γ1 (N ); it works
for all subgroups Γ such that there exists a positive integer n such that Γ(N ) ≤ Γ ≤ SL2 (Z).
 
a b
Weight k “Slash operators”: If γ ∈ SL2 (Z), γ = , we define
c d

f |k γ(z) = (cz + d)−k f (γ(z)).

Some properties of the slash operator:

1) If f |k γ = f for all γ ∈ Γ1 (N ), then f satisfies (1) for all γ ∈ Γ1 (N ).

2) f |k (γ1 γ2 ) = (f |k γ1 ) |k γ2 .

3) If f |k γ = f for all γ ∈ Γ and α ∈ SL2 (Z), then f |k α is invariant under α−1 Γα.

In particular, if α(∞) = r/s, then f |k α is invariant under some finite index subgroup
of SL2 (Z). As a consequence, ∃w ∈ Z such that f |k α(z) = f |k α(z + w). The minimal w
with this property is called the width of the cusp r/s relative to Γ. Since the function is
w-periodic, it has a Fourier expansion of the form
X
f |k α(z) = an (r/s)e2πinz/w .
n∈Z

Definition 3. The function f is holomorphic at r/s if f |k α is holomorphic at ∞.

The suitable growth conditions at the cusps are precisely these, i.e., f is holomorphic at
the cusps.
Notations:

a) Mk (Γ0 (N ), E) is the space of modular forms of weight k, level N and character E (holo-
morphic and holomorphic at the cusps).

b) Sk (Γ0 (N ), E) is the space of modular forms of weight k, level N and character E such that
they vanish at all cusps (and are holomorphic.) Note that Sk (Γ0 (N ), E) ⊆ Mk (Γ0 (N ), E).

If f ∈ Mk (Γ0 (N ), E) and we write f (z) = ∞ n

P
n=0 an q , we formally define the L-function
attached to f to be
∞
X
L(f, s) := an n−s .
n=1

Note the absence of the term a0 . This is a natural generalization of L(χ, s) as we will see.
Next time we will show that the an have at most polynomial growth, kan k  nk/2 (in fact,
k−1
kan k << n 2 , but this result is beyond the scope of this course).