MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

In Theorem 3.2 of 'Introduction to spectral theory of automorphic forms' by Iwaniec,the first bound is about the coefficients of automorphic forms

$$\sum_{|n|\le N}|n||c_n|^2<<(N+|s|)e^{\pi|s|}$$

whereas in the paper 'On the uniform equidistribution of long closed horocycles', by A. Strombergsson, it says that this bound needs some revision (the note after Proposition 4.5 in the paper.) Also another bound is given:

$$\sum_{|n|\le N}|c_n|^2=O((N+|s|)e^{\pi|s|})$$

The reason is that the formula on page 61, line 7, in Iwaniec's book, is not correct.

I'm not familiar with these stuff, and want to know which statement is correct, Iwaniec or Strombergsson? Or both? Also what's the best estimate towards these coefficients (maybe in various forms)?

Thanks a lot for your help, I appreciate it a lot!

share|cite|improve this question
up vote 5 down vote accepted

Strömbergsson's footnote refers to $\sum_{|n|\leq N}|c_n|^2$, not $\sum_{|n|\leq N}|n||c_n|^2$. In fact his $c_n$ is Iwaniec's $|n|^{1/2}\hat f_\mathfrak{a}(n)$. Normalization is a serious difficulty in the subject, one must be careful.

Strömbergsson points out that Iwaniec's proof is incorrect, but the result is fine. To fix the given line in Iwaniec's proof, integrate from $|s|/4$ instead of $|s|$, then the claimed bound is all right. This you can see from the discussion below (4.13) in Strömbergsson's paper: already the integral from $|s|/4$ to $|s|/2$ is large. Note that Strömbergsson talks about $s=1/2+iR$ (i.e. the case $\Re s=1/2$), but for the remaining $s$ (i.e. $1/2\leq s\leq 1$) Iwaniec's bound follows by compactness.

I don't know (from the top of my head) of any better uniform estimate than the one displayed. In special ranges or for Hecke eigenforms one has better estimates, but there are many open questions (e.g. how to give a good lower bound when $N$ is small compared to the various parameters of the form).

Added. This is a response to the OP's comment below. If $\lambda(n)$ denotes the Hecke eigenvalues of a primitive cusp form, then one can derive an asymptotic formula for $\sum_{n\leq N} |\lambda(n)|^2$ or a smooth version of it by applying standard Mellin transform techniques for the Dirichlet series $\sum_{n=1}^\infty |\lambda(n)|^2/n^s=L(s,f\otimes\tilde f)/\zeta(2s)$. The main term comes from the pole at $s=1$, while the error term depends on bounds for the Rankin-Selberg $L$-function $L(s,f\otimes\tilde f)$ to the left of $s=1$. The main term involves the residue at $s=1$, for which very good bounds are available by Iwaniec (Acta Arith. 56 (1990), 65–82) and by Hoffstein-Lockhart (Ann. of Math. 140 (1994), 161–181). For the error term one can use the standard convexity bound for $L(s,f\otimes\tilde f)$, and hopefully a subconvex bound will also be available later in the future. Rankin in his original paper (Proc. Cambridge Philos. Soc. 35 (1939), 351–372) established $\sum_{n\leq N} |\lambda(n)|^2=cN+O(N^{3/5})$, where the error term is still the best one and follows from a general result of Landau available here.

share|cite|improve this answer
Thanks a lot for your explanation.Now if I'm interested in the cusp forms and Eisenstein series on full modular group, what are the better estimates for Hecke eigenforms? Any reference or answer is appreciated! – user1832 Oct 27 '12 at 15:55
unknown (google): I added a section for you. – GH from MO Oct 27 '12 at 19:53
thank you very much! – user1832 Oct 28 '12 at 0:14

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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