Multiplicities of representations of Γ / Γ 1   in M k ( Γ 1 )   and S k ( Γ 1 )  

BinYong,Hsie {LMAM,Department of Mathematics, PeKing University, BeiJing, China 100871}

2002.9

Abstract
This paper shows that for an given irreducible representation ρ   of Γ / Γ 1   , the two functions d i m ( M k ( Γ 1 , ρ ) )   and d i m ( S k ( Γ 1 , ρ ) )   of k   are almost linear functions.

1 Main theorem

Let Γ   and Γ 1   be two Fuchsian subgroups of S L 2 ( Q )   of the first type, and assume that Γ 1   is a normal subgroup of Γ   with finite index.
For Γ 0 = Γ   or Γ 1   , let M k ( Γ 0 )   ( A k ( Γ 0 )   , S k ( Γ 0 )   ) be the space of modular forms (meromorphic modular forms, cusp forms) of Γ 0   with weight k   . Let Y = Γ \ H *   and X = Γ 1 \ H *   . Let Ω ( X )   ( Ω ( Y )   ) be the field of meromorphic functions of X   ( Y   ). Then A 0 ( Γ ) ∼ = Ω ( Y )   and A 0 ( Γ 1 ) ∼ = Ω ( X )   , in a natural way.
Γ   has a representation π k   in A k ( Γ 1 )   defined by
( π k ( γ ) f ) ( z ) = f | [ γ − 1 ] k ( z ) = ( c z + d ) − k f ( γ − 1 ( z ) ) (1),
where γ − 1 = ( a b c d ) ∈ Γ   and f ∈ A k ( Γ 1 )   . It is easy to see that π k   is a representation of Γ   and π k ( Γ 1 )   is trivial. Therefore we obtain a representation of Γ / Γ 1   . We wrote π k   for it, too. Notice that π k   stables M k ( Γ 1 )   and S k ( Γ 1 )   . We write ρ k   ( σ k   ) for the restriction of π k   to M k ( Γ 1 )   ( S k ( Γ 1 )   ).
Let R e p   ( R e p 0   ) be the set of all representations of Γ / Γ 1   ( Γ ¯ / Γ 1 ¯   ), where Γ ¯ = Γ / Γ ∩ { ± I }   and Γ 1 ¯ = Γ 1 / Γ 1 ∩ { ± I }   . In this paper, we regard a representations of Γ ¯ / Γ 1 ¯   as a representation of Γ / Γ 1   naturally without mentioned. Let R   ( R 0   ) be the regular representation of Γ / Γ 1   ( Γ ¯ / Γ 1 ¯   ). If − I   is not in Γ   , then R e p = R e p 0   , R = R 0   , and let R e p 1 = R e p   , R 1 = R   . If − I   is in Γ 1   , then R e p = R e p 0   , R = R 0   , and let R e p 1   be empty. If − I   is in Γ   but not in Γ 1   , then R e p ≠ R e p 0   , R ≠ R 0   and let R e p 1 = R e p \ R e p 0   , R 1   be the representation of Γ / Γ 1   such that R = R 0 ⊕ R 1   . Then R i = ⊕ ρ ∈ R e p i d i m ( ρ ) ρ ( i = 0 , 1 )   .
Theory of modular forms tells us that ρ k   and σ k   have decompositions
ρ k = ⊕ ρ ∈ R e p i M k ( Γ 1 , ρ ) (2),
and
σ k = ⊕ ρ ∈ R e p i S k ( Γ 1 , ρ ) (3),
where i = 0   or 1 and i ≡ k ( m o d 2 )   .
The main result of this paper is the following theorem.
Theorem Assume ρ ∈ R e p i   ( i = 0 , 1   ), then
lim k → ∞ , k ≡ i ( m o d 2 ) d i m M k ( Γ 1 , ρ ) k d i m ( ρ ) 2 , lim k → ∞ , k ≡ i ( m o d 2 ) d i m S k ( Γ 1 , ρ ) k d i m ( ρ ) 2 (4)
exist and both equal to 1 4 π ∫ Γ \ H d x d y y 2   .
We give an important application of the above theorem.
Let Γ   be Γ 0 ( N )   and Γ 1   be Γ 1 ( N )   . Notice that l ( Γ 0 ( N ) ) = N ∏ p : p r i m e , p | N ( 1 + p − 1 )   . If χ ( − I ) = 1   , we have lim k → ∞ , k e v e n d i m ( M k ( χ , N ) ) k = lim k → ∞ , k e v e n d i m ( S k ( χ , N ) ) k = N ∏ p : p r i m e , p | N ( 1 + p − 1 ) .   If χ ( − I ) = − 1   , we have lim k → ∞ , k o d d d i m ( M k ( χ , N ) ) k = lim k → ∞ , k o d d d i m ( S k ( χ , N ) ) k = N ∏ p : p r i m e , p | N ( 1 + p − 1 ) .  

2 Proof of the main result

Let X   , Y   be the same as in section 1. Write μ   for [ Γ ¯ : Γ 1 ¯ ]   . Let i = 0 , 1   and when i = 1   , we assume − I   is not in Γ 1   .
Select out a p ∈ Γ \ H   which has μ   different lifts p 1 , . . . , p μ   in X   . Select out a lift p ~ j   in H   for each p j   .
By Riemann-Roch theorem, we can find an f 1 ∈ A i ( Γ 1 )   such that f 1 ( p ~ 1 ) = 1   and f 1 ( p ~ j ) = 0   for j ∈ { 2 , . . . , μ }   . For any α ∈ Γ / Γ 1   , we can set f α = f 1 | [ α ] i   . Let W i   be the C   -linear subspace of A i ( Γ 1 )   spanned by { f α : α ∈ Γ / Γ 1 }   . It is easy to see that W i   is stable under π i ( Γ / Γ 1 )   .
We have the following lemma.
Lemma A 0 ( Γ ) ⊗ C W i ∼ = A i ( Γ 1 )   . And W i   and R i   are two isomorphic representations of Γ / Γ 1   .
  • Proof. We only need to find μ   elements in W i   which are linear independent over A 0 ( Γ )   . We can select out a subset T   of Γ / Γ 1   , which is a lift of Γ ¯ / Γ 1 ¯   . We are going to show that { f α : α ∈ T }   is linear independent over A 0 ( Γ )   . Otherwise, there are g α ∈ A 0 ( Γ ) ( α ∈ T )   , not all zero, such that ∑ α ∈ T g α f α = 0   . We may assume all g α   are holomorphic at p ~ 1 , . . . , p ~ μ   , and moreover g β ( p ~ 1 ) = 1   for some β ∈ T   . Then 0 = ∑ α ∈ T g α f α | [ β − 1 ] i .   But the right side of the above equality equals 1 at p ~ 1   . This is a contradiction.
    Therefore A 0 ( Γ ) ⊗ C W i ∼ = A i ( Γ 1 )   .
    Comparing the traces, we find that W i ∼ = R i   .
We are going to prove the main theorem.
Find an even n 0   , and a φ ∈ S n 0 ( Γ )   such that φ ⋅ f α   are all cusp forms. Let W φ   be the C   -linear subspace of S n 0 ( Γ )   spanned by { φ f α : α ∈ Γ / Γ 1 }   . Then W φ ⊗ C S k ( Γ ) ⟶ S i + n 0 + k ( Γ 1 )   and W φ ⊗ C M k ( Γ ) ⟶ M i + n 0 + k ( Γ 1 )   are both injective.
Hence for ρ ∈ R e p i   , liminf k → ∞ , 2 | k [ ρ i + n 0 + k : ρ ] i + n 0 + k ≥ liminf k → ∞ , 2 | k d i m ( M k ( Γ ) ) ⋅ [ R i : ρ ] i + n 0 + k  
= liminf k → ∞ , k ≡ i m o d 2 d i m ( M k ( Γ ) ) d i m ( ρ ) k (5).
Shimura [2] tells us that l ( Γ ) = { lim k → ∞ M k ( Γ ( N ) ) k = lim k → ∞ S k ( Γ ( N ) ) k ( − I i s n o t i n Γ ) lim k → ∞ k e v e n M k ( Γ ( N ) ) k = lim k → ∞ k e v e n S k ( Γ ( N ) ) k ( − I ∈ Γ )   exists and equals to 1 4 π ∫ Γ \ H d x d y y 2   . So l ( Γ 1 ) = l ( Γ ) ⋅ μ ,   and
lim k → ∞ k ≡ i ( m o d 2 ) d i m ( ρ k ) k = l ( Γ ) ⋅ μ (6).
Comparing (5) and (6), we find that
lim k → ∞ k ≡ i ( m o d 2 ) [ ρ k : ρ ] k = l ( Γ ) ⋅ d i m ( ρ ) (7).
Similarly, we have
lim k → ∞ k ≡ i ( m o d 2 ) [ σ k : ρ ] k = l ( Γ ) ⋅ d i m ( ρ ) (8).
We have now proved the main result of this paper.

3 A example

Let Γ   be Γ 0 ( N )   and Γ 1   be Γ 1 ( N )   . Let χ ( m o d N )   be a Dirichlet's character. Let M k ( N , χ )   be the space { f ∈ M k ( Γ 0 ( N ) ) : f | [ γ ] k = χ ( d ) f , γ = ( a b c d ) ∈ Γ 0 ( N ) }   and S k ( N , χ )   be the space M k ( N , χ ) ∩ S k ( N )   . Notice that l ( Γ 0 ( N ) ) = N ∏ p : p r i m e , p | N ( 1 + p − 1 )   .
If χ ( − I ) = 1   , we have lim k → ∞ , k e v e n d i m ( M k ( χ , N ) ) k = lim k → ∞ , k e v e n d i m ( S k ( χ , N ) ) k = N ∏ p : p r i m e , p | N ( 1 + p − 1 ) .   If χ ( − I ) = − 1   , we have lim k → ∞ , k o d d d i m ( M k ( χ , N ) ) k = lim k → ∞ , k o d d d i m ( S k ( χ , N ) ) k = N ∏ p : p r i m e , p | N ( 1 + p − 1 ) .   References

  1. J.P.Serre. Linear Representations of Finite Groups. Spring-Verlag, New York. 1977
  2. G.Shimura. Introduction to the Arithmetic Theory of Automorphic Functions. Princeton University Press, Princeton, New Jersey. 1971
  3. O.Forster. Lectures on Riemann Surfaces. Springer-Verlag, New York. 1981
  4. N.Koblitz. Introduction to Elliptic Curves and Modular Forms. Springer-Verlag, New York. 1984

0   E-mail: byhsie@math.pku.edu.cn