The Poisson Kernel for Drinfeld Modular Curves

Drinfeld modules elliptic curve L-series Mellin transform non-archimedean measure Poisson kernel formula q-expansion coefficients rigid analytic modular forms for function fields rigid analytic space space of cusp forms

Mathematics Subject Classification ID

Special values of automorphic (L)-series, periods of automorphic forms, cohomology, modular symbols (11F67) (p)-adic theory, local fields (11F85) Drinfel'd modules; higher-dimensional motives, etc. (11G09) Arithmetic theory of algebraic function fields (11R58) Arithmetic theory of polynomial rings over finite fields (11T55) Local ground fields in algebraic geometry (14G20) Formal groups, (p)-divisible groups (14L05)

Review text

This is an important and pleasing paper. It introduces a basic integral formalism into the theory of rigid analytic modular forms for function fields. The results work in complete generality (see remark below) though we will just describe them in the simplest case of \({\mathbf A}=\mathbb F_ r[T]\), \(r=p^ m\), \({\mathbf k}=\mathbb F_ r(T)\) and \({\mathbf K}={\mathbf k}_ \infty\); so \({\mathbf A}\subset {\mathbf K}\) is discrete and co-compact. Let \(\Gamma=\text{GL}_ 2({\mathbf A})\) and let \(\Omega=\overline{\mathbf K}-{\mathbf K}\) for some fixed choice of algebraic closure of \({\mathbf K}\). The space \(\Omega\) can be given the structure of a connected rigid analytic space (i.e., a non-archimedean ringed-space where one can use analytic continuation). The group \(\Gamma\) acts discretely on \(\Omega\) and the function \(q(z)=\sum_{a\in{\mathbf A}}(z-a)^{-1}=1/{\mathbf e}(z)\) is a uniformizing element at infinity. The classical definitions of ``modular form and ``cusp form carry over in a straightforward fashion to \(\{\Omega,\Gamma\}\). Let \(f(z)\) be a cusp form of weight \(k\) for \(\Gamma\). The author associates to \(f\) a measure \(\mu_ f\) (in the sense of Vishik, Amice and Vélu -- so one cannot integrate all continuous functions but rather those that are locally analytic, which suffices for most applications). These measures have symmetries which reflect the defining \(\Gamma\)-invariance properties of the cusp form. A major result of the paper is an isomorphism between the space of such measures and the space of the cusp forms of weight \(k\). One recovers the cusp form from the measure by the Poisson kernel formula: \[ f(z)=\int_{{\mathbb P}^ 1({\mathbf K})}{1\over z-x} \,d\mu(x). \] The author also gives an integral for the \(q\)-expansion coefficients of the form \(f(z)=\sum b_ jq^ j\): \(b_ j=\int_{{\mathbf K}/{\mathbf A}}{\mathbf e}(x)^{j-1}\,d\mu(x).\) (This can be proved in the following fashion which works, unlike the author's proof, in complete generality: By the remark following Lemma 13 of the paper, we can write \(f(z)=\int_{\mathbf K}{1\over z-x}\, d\mu(x).\) By the symmetry properties of \(\mu\), this equals \[ \begin{aligned} \int\limits_{{\mathbf K}/{\mathbf A}}\sum_{a\in A}{1 \over z-x+a}\,d\mu(x)&=\int {1\over{\mathbf e}(z-x)}\,d\mu(x),\\ &=\int{1\over {\mathbf e}(z)-{\mathbf e}(x)}\,d\mu(x),\end{aligned} \] as \({\mathbf e}(z)\) is additive. Now express the integrand in terms of \(q(z)={\mathbf e}(z)^{- 1}\) and expand via the geometric series. The existence and integral for \(q\)-expansions follows immediately and quite formally.) If we view \(d\mu(x)\) as `morally' \(f(x)dx\) and note that \(dq(x)=-q(x)^ 2\,dx\), the analogy with Cauchy's Theorem becomes apparent upon substituting the \(q\)-expansion for \(f\) in the integral! Let \({\mathcal U}_ 1\subset {\mathbf K}\) be the group of 1-units. Let \(s\in\mathbb Z_ p\) and \(L_ f(s)=\int_{{\mathcal U}_ 1}u^ s{d\mu(x)\over u}\). Then using the symmetry properties of \(\mu\) one can show that \(L_ f(s)=\pm L_ f(k-s)\) (where the exact sign can be found). The relationship between \(L_ f(s)\) is given by a Fourier transform [see the reviewer, J. Algebra 146, No. 1, 219--241 (1992; Zbl 0758.11030)]. These Fourier transforms arise naturally when one wants to understand the \(L\)-series of Drinfeld modules; functions naturally defined on the space \(\overline{\mathbf K}^*\times \mathbb Z_ p\). To define the F. T. one needs to pick a uniformizing power series. The power series needed to compute \(L_ f\) is the logarithm of \({\mathbf e}(z)\); for Drinfeld modules one does not quite yet know what to expect, but the logarithm does not seem to be the correct choice. It may be that there is some sort of deformation theory that allows one to ultimately relate the two type of \(L\)-series. This would allow one to use Teitelbaum's measures to study \(L\)-series of Drinfeld modules also. There is another fascinating area raised by the existence of these measures. Let \(E\) be an elliptic curve over \(\mathbf k\) with split multiplicative reduction at \(\infty\). One knows by Drinfeld that \(E\) is isogenous to a factor of the Jacobian of some moduli curve. Thus we deduce a differential form associated to \(E\to\) a cusp form \(\to\) a measure associated to \(E\to\) possibility of a characteristic \(p\) \(L\)-series for \(E\). On the other hand, E.-U. Gekeler has used this work to establish the non- existence of cusp forms of weight 1 in the case of nontrivial level structure. One takes the \(p\)-th power of such a form and then applies Teitelbaum's results to show that the corresponding measure must be trivial. While disappointing, this result is, perhaps, not surprising in the following sense: Classical theory leads to the expectation that a form of weight one ``should correspond to a representation of a finite Galois group \(G\). However, with a little thought, one sees that these representations should really be defined in characteristic 0 (else, if \(p\mid\text{Ord}(G)\), one loses control of local factors). In any case, much work needs to be done before one has exhausted the potential raised by the exciting paper being reviewed!

Recommendations

Cites work

Cited in

(30)

This page was built for publication: The Poisson Kernel for Drinfeld Modular Curves

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3979558)