MathOverflow will be down for maintenance for approximately 3 hours, starting Monday evening (06/24/2013) at approximately 9:00 PM Eastern time (UTC-4).

## real-valued functions on the modular surface

How does one write down $\mathbb{R}$-valued functions on the modular surface? I am considering taking an arbitrary function on the upper half plane $f:\mathbb{H} \to \mathbb{R}$ and averaging over the elements of $SL(2, \mathbb{Z})$. So, $$f_1(z) = \sum_{g \in SL(2, \mathrm{Z})} f(gz)$$ there may have to be a decay condition on $f$ so the function will converge. I am not necessarily looking for holomorphic functions, just smooth and well defined on $\mathbb{H}\backslash SL(2, \mathbb{Z})$.

-
 I must be missing something: what counts as "writing down a function"? There are going to be a plethora of non-zero smooth functions on this surface, so what other properties or identities are you looking for? – Yemon Choi Jul 2 2010 at 1:45 Take any smooth function of the j-invariant? – Qiaochu Yuan Jul 2 2010 at 1:59 In the absence of an application or additional context, it is hard to tell if you would prefer a universal answer as given by Yemon and Qiaochu, or something more structured. – S. Carnahan♦ Jul 2 2010 at 2:08 It could be the word I am looking for is "modular function". Does the j-invvariant have a Fourier series? What are the critical values? – John Mangual Jul 2 2010 at 19:32

You may have an easier time starting with a function that is periodic under translation by 1, then summing over cosets of translation in $SL_2(\mathbb{Z})$. If your initial function is well-behaved, your sum will converge (although often one introduces correction terms to get sections of a line bundle, i.e., modular forms of nonzero weight). This is a common method for constructing PoincarĂ© series, Real-analytic Eisenstein series (where $f$ is given by a power of the imaginary part), and Rademacher sums (where $f$ is exponential).