Cuspidal representation

In number theory, cuspidal representations are certain representations of algebraic groups that occur discretely in L^2 spaces. The term cuspidal is derived, at a certain distance, from the cusp forms of classical modular form theory. In the contemporary formulation of automorphic representations, representations take the place of holomorphic functions; these representations may be of adelic algebraic groups.

When the group is the general linear group \operatorname{GL}_2, the cuspidal representations are directly related to cusp forms and Maass forms. For the case of cusp forms, each Hecke eigenform (newform) corresponds to a cuspidal representation.

Formulation

Let G be a reductive algebraic group over a number field K and let A denote the adeles of K. Let Z denote the centre of G and let ω be a continuous unitary character from Z(K)\Z(A)× to C×. Fix a Haar measure on G(A) and let L20(G(K)\G(A), ω) denote the Hilbert space of measurable complex-valued functions, f, on G(A) satisfying

  1. f(γg) = f(g) for all γ ∈ G(K)
  2. f(gz) = f(g)ω(z) for all z ∈ Z(A)
  3. \int_{Z(\mathbf{A})G(K)\backslash G(\mathbf{A})}|f(g)|^2\,dg < \infty
  4. \int_{U(K)\backslash U(\mathbf{A})}f(ug)\,du=0 for all unipotent radicals, U, of all proper parabolic subgroups of G(A).

This is called the space of cusp forms with central character ω on G(A). A function occurring in such a space is called a cuspidal function. This space is a unitary representation of the group G(A) where the action of g ∈ G(A) on a cuspidal function f is given by

(g\cdot f)(x)=f(xg).

The space of cusp forms with central character ω decomposes into a direct sum of Hilbert spaces

L^2_0(G(K)\backslash G(\mathbf{A}),\omega)=\hat{\bigoplus}_{(\pi,V_\pi)}m_\pi V_\pi

where the sum is over irreducible subrepresentations of L20(G(K)\G(A), ω) and mπ are positive integers (i.e. each irreducible subrepresentation occurs with finite multiplicity). A cuspidal representation of G(A) is such a subrepresentation (π, V) for some ω.

The groups for which the multiplicities mπ all equal one are said to have the multiplicity-one property.

References