Abstract.   We explain how different definitions of Hecke operators relate to each other.

Table of Contents

  1. The classical Hecke algebra
  2. Hecke algebras for t.d. groups

1. The classical Hecke algebra

  Let $ k $ be a field, and let $ E $ be a $ k $-variety. The following are equivalent:

  1. $ E $ is an abelian variety of dimension 1 over $ k $;

  2. $ E $ is a smooth, projective curve of genus one with a choice of origin;

we refer to such $ E $ as being elliptic.

  We call a discrete additive subgroup $ L $ of $ k $ a lattice. There exists an additive isomorphism of $ L $ with $ \mathbb{Z}^n $; we call $ n $ the rank of our lattice.

  Assume $ L $ is a complex lattice of rank 2, which we write

\[L = \left< w_1, w_2 \right> = \left< a_1 w_1 + a_2 w_2 \mid a_i \in \mathbb{Z} \right>\]

with $ w_2 / w_1 \in \mathcal{H} $ and $ \mathcal{H} $ the upper-half plane. The quotient $ \mathbb{C}/L $ is an abelian variety of dimension 1 over $ \mathbb{C} $; hence, an elliptic curve. Every complex elliptic curve arises in this manner, and there is a bijection between isomorphism classes of elliptic curves and lattices up to homothety. (This is not true for higher dimensional complex abelian varieties.) In the sequel, we let $ \mathcal{L} $ denote the free abelian group generated by complex lattices of rank 2 (i.e. elliptic curves).

  Elements of the modular group $ \gamma \in \text{SL}_2(\mathbb{Z}) $ act on $ \mathcal{H} $ by fractional linear transformations; meaning for $ z \in \mathcal{H} $ and

\[\gamma = \begin{bmatrix} a & b \\ c & d \end{bmatrix}, \text{ we put } \gamma(z) = \frac{az + b}{cz + d}.\]

  A holomorphic function $ f \colon \mathcal{H} \to \mathbb{C} $ is called a modular form of weight $ k $ if, for any $ \gamma \in \text{SL}_2(\mathbb{Z}) $,

\[f(\gamma(z)) = (cz+d)^k f(z),\]

and $ f(z) $ remains bounded as $ \text{im}(z) $ approaches infinity. Equivalently, we define modular forms as homogeneous functions $ F \colon \mathcal{L} \to \mathbb{C} $ of degree $ -k $; meaning, for $ \lambda \in \mathbb{C}^{\times} $,

\[F(\lambda w_1, \lambda w_2) = \lambda^{-k} F(w_1, w_2).\]

Indeed, given a modular form $ f $ of weight $ k $, we put

\[F(a, b) = b^{-k} f(a/b),\]

and given a homogeneous function $ F $ of weight $ -k $, we put

\[f(z) = F(z, 1).\]

This correspondence is bijective, and we denote our spaces of modular forms by $ f \in M_k $ and $ F \in M_k^{\mathcal{L}} $; put $ M_* = \bigoplus M_k $ and $ M_*^{\mathcal{L}} = \bigoplus M_k^{\mathcal{L}} $. We define the Hecke algebra on these spaces as follows.

  ($ \mathcal{H} $ point de vue) Put $ \Gamma = \text{SL}_2(\mathbb{Z}) $. Define $ H_k $ to be the free abelian group with basis elements double cosets $ \Gamma \alpha \Gamma $ where $ \alpha \in M_2^+(\mathbb{Z}) $, the plus meaning positive determinant. Take coset decompositions

\[x = \Gamma \alpha \Gamma = \coprod_{i=1}^A \Gamma \alpha_i, \qquad y = \Gamma \beta \Gamma = \coprod_{j=1}^B \Gamma \beta_j,\]

and $ z = \Gamma \xi \Gamma $, then set

\[x \cdot y = \sum_{z \subseteq \Gamma \alpha \Gamma \beta \Gamma} m(x \cdot y, z) z\]

(summing over all cosets $ z $ contained in $ \Gamma \alpha \Gamma \beta \Gamma $), where

\[m(x \cdot y, z) = \text{Card}\{ (i, j) \mid \Gamma \alpha_i \beta_j = \Gamma \zeta \};\]

this product is independent of our coset representatives, and it is associative and commutative; we extend it $ \mathbb{Z} $-linearly to define multiplication on $ H $. (Observe multiplication here is a type of convolution.) We call $ H $ the Hecke algebra.

  $ H $ can be considered a subalgebra of $ \text{End}(M_*) $, and hence $ M $ a left $ H $-module, by the action on $ f \in M_k $

\[\Gamma \alpha \Gamma \cdot f(z) = \det(\alpha)^{k-1} \sum_{i=1}^A j(\alpha_i, z)^{-k} f(\alpha_i(z)),\]

We define the $ n $-th Hecke operator as the coset

\[T_n = \Gamma \begin{bmatrix} n & 0 \\ 0 & 1 \end{bmatrix} \Gamma\]

$ H $ is generated by the $ T_n $. (This is the simplest matrix of determinant $ n $.)

  (Measure point de vue) Let

\[\varphi \colon \text{SL}_2(\mathbb{Z}) \backslash \text{M}_2^+(\mathbb{Z}) / \text{SL}_2(\mathbb{Z}) \to \mathbb{Z}\]

be a continuous function with compact support; assuming the left hand side is equipped with the discrete topology, this reduces to saying $ \varphi $ is supported on finitely many cosets. Write $ H^{\mathcal{M}} $ for the set of all such $ \varphi $. Let multiplication on $ H^{\mathcal{M}} $ be the convolution

\[(\varphi_1 \ast \varphi_2)(x) = \int_{\text{SL}_2(\mathbb{Z}) \backslash \text{M}_2^+(\mathbb{Z}) / \text{SL}_2(\mathbb{Z})} \varphi_1(y) \varphi_2(y^{-1} x) dy,\]

where every $ y $ is a point mass of volume 1. We call $ H^{\mathcal{M}} $ the Hecke algebra.

  Each $ \varphi \in H^{\mathcal{M}} $ acts on $ f \in M_k $ by

\[\varphi \cdot f = \int_{\text{SL}_2(\mathbb{Z}) \backslash M_2^+(\mathbb{Z})} \det(g)^{k-1} j(g, z)^{-k} f(g(z)) \varphi(g) dg\]

(with the same measure). In particular, the $ n $-th Hecke opeartor $ T_n^{\mathcal{M}} $ is the characteristic function on the coset of $ T_n $ above.

  (Lattice point de vue) We define the $ n $-th Hecke operator $ T_n^{\mathcal{L}} \in \text{End}(M_k^{\mathcal{L}}) $ by

\[T_n^{\mathcal{L}} F(L) = n^{k-1} \sum_{[L : L'] = n} F(L');\]

note that again the action depends on the weight $ -k $ of $ F $. We call the subalgebra $ H^{\mathcal{L}} $ of $ \text{End} (M^{\mathcal{L}}_{\ast}) $ generated by the $ T_n^{\mathcal{L}} $ the Hecke algebra.

Theorem 1.   $ T_n f(z) = T_n^{\mathcal{M}} = T_n^{\mathcal{L}} F(z, 1) $.

  The first equality follows by definition. Put $ L = \left< z, 1 \right> $. Any index $ n $ sublattice $ L' $ of $ L $ is given by a $ \mathbb{Z} $-linear transformation $ M \colon L \to L' $ with $ \det M = n $. Since \(\text{SL}_2(\mathbb{Z})\) is the kernel of the determinant, we have a bijection between isomorphism classes of index $ n $ sublattices and \(M_n^+(\mathbb{Z})/\text{SL}_2(\mathbb{Z})\). Writing $ \Gamma T_n \Gamma = \coprod_{i=1}^N \Gamma t_i $, it then follows immediately that

\[T_n f(z) = n^{k-1} \sum_{i=1}^N f(t_i(z)) = n^{k-1} \sum_{[L : L'] = n} F(L) = T_n^{\mathcal{L}} F(L).\]

Q.E.D.

2. Hecke algebras for t.d. groups

  Let $ G $ be a t.d. (totally disconnected) topological group, meaning every neighborhood of 1 contains a compact open subgroup; let $ K $ be any compact open subgroup of $ G $. Denote by $ H(G, K) $ be the algebra of compactly suppoerted characters $ \varphi \colon K \backslash G / K \to \mathbb{C} $, where the product is a with the convolution with respect to a Haar measure. We call $ H(G, K) $ the Hecke algebra of $ G $ with respect to $ K $.

  In §1, we considered "$ H^{\mathcal{M}} = H(M_2^+(\mathbb{Z}), \text{SL}_2(\mathbb{Z})) $", the quotation marks being because $ M_2^+(\mathbb{Z}) $ is not a group.

3. Iwahori-Hecke algebra

4. References