If a (suitably normalised) holomorphic cusp newform has q-expansion $$f(z) = \sum_n \lambda_f(n) e(nz),$$

then we know the Hecke relations for $(mn,q)=1$, $$(\star) \qquad \lambda_f(m)\lambda_f(n) = \sum_{d | (m, n)} \lambda_f\left( \frac{mn}{d^2} \right).$$

In the case of a Maass cusp form, we can write the expansion $$u(z) = \sum_{n \geqslant 1} \rho_u(n) W_s(nz),$$

where $s$ is related in an explicit way to the associated eigenvalue and $W_s$ is a Whittaker function. Similarly, the continuous part of the spectrum made of Eisenstein series at a cusp $a$ admits expansion of the form $$E_a(s,z) = \phi_a y^s + \phi_a(s) y^{1-s} + \sum_{n \geqslant 1} \phi_a(n,s)W_s(nz)$$

I am interested in the following question :

Are there analogous "Hecke relations" for the coefficients of Maass forms, $\rho_u(n)$ and $\phi_a(n,s)$?

I suppose so, but I do not have any good reference for these matters.

  • 1
    $\begingroup$ Yes, see Duke, Friedlander, and Iwaniec, The subconvexity problem for Artin $L$-functions (Inventiones 149, 2002), Section 6. In particular, the Hecke relations (or multiplicative commuting operators) are valid for any functions of period one. Since these operators commute with the Laplacian, Bob's your uncle and you can decompose into simultaneous eigenfunctions. Is this not also in one of Iwaniec's books? doi.org/10.1007/s002220200223 $\endgroup$ Commented Apr 27, 2019 at 10:37
  • 1
    $\begingroup$ See 8.5 of Iwaniec's Spectral Methods of Automorphic Forms (AMS Grad Studies in Math 53, 2002). doi.org/10.1090/gsm/053 $\endgroup$ Commented Apr 27, 2019 at 10:44
  • 4
    $\begingroup$ Just to be clear, in general the Eisenstein series attached to cusps are not eigenfunctions of the Hecke operators, so $\rho_u(n)$ will not satisfy the Hecke relations. This doesn't contradict what literature-searcher is saying, that a given Laplace eigenspace of Eisenstein series can be decomposed into Hecke eigenfunctions. $\endgroup$
    – Matt Young
    Commented Apr 27, 2019 at 14:42

1 Answer 1


Two remarks supplementing the excellent comments below your post.

  1. The relation $(\star)$ holds for every $m$ and $n$ if your restrict the sum to $(d,q)=1$. This is because $f$ is a newform (and we assume that $f$ has trivial nebentypus). The same is true for Maass cuspidal newforms.

  2. You can get a Hecke eigenbasis of Eisenstein series if you induce the basis vectors from Dirichlet characters modulo $q$. See Section 2.1.1 of this article for a quick introduction. I also recommend you to study Chapter 7 of Miyake's book, which discusses these induced Eisenstein series in detail and in classical language.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.