\(\wp\)-adic continuous families of Drinfeld eigenforms of finite slope (Q2227285): Difference between revisions

Let \(K=\mathbb{F}_q(t)\) and \(A=\mathbb{F}_q[t]\) (\(q\) a power of a prime \(p\)) and fix a prime \(\wp\) in \(A\). For any arithmetic subgroup \(\Gamma\subseteq\mathrm{SL}_2(A)\) (the main example being \[ \Gamma_1(\mathfrak{n}\wp^r)=\left\{\left(\begin{matrix} a & b \\ c & d\end{matrix}\right)\equiv \left(\begin{matrix} 1 & * \\ 0 & 1 \end{matrix}\right) \pmod{\mathfrak{n}\wp^r} \right\} \] for some ideal \(\mathfrak{n}\) of \(A\) prime with \(\wp\)) and any integer \(k\geqslant 2\), let \(S_k(\Gamma)\) be the space of Drinfeld cuspforms of level \(\Gamma\). The paper deals with the existence of \(\wp\)-adic families of eigenforms: an eigenform of weight \(k\) is an element of \(S_k(\Gamma)\) which is an eigenform for all the Hecke operators \(T_{\mathfrak{q}}\), and a \(\wp\)-adic family is a set of eigenforms \(\{f_{k_n}\in S_{k_n}(\Gamma)\,:\,n\in \mathbb{N}\}\) all of the same \(\wp\)-adic slope (i.e., the \(\wp\)-adic valuation of the eigenvalue at \(T_\wp\)), such that their \(T_{\mathfrak{q}}\)-eigenvalues are \(\wp\)-adically close for any prime \(\mathfrak{q}\) of \(A\). Drinfeld modular forms admit Fourier expansions but only for a class of them (the ones with \(A\)-expansion, see [\textit{A. Petrov}, J. Number Theory 133, No. 7, 2247--2266 (2013; Zbl 1286.11075)]) there is a direct and explicit relation between Fourier coefficients and \(T_{\mathfrak{q}}\)-eigenvalues, hence the problem requires an approach different from the classical characteristic zero case. Let \(a\) be a \(\wp\)-slope for an eigenform \(f_{k_1}\in S_{k_1}(\Gamma_1(\mathfrak{n}\wp^r))\) (we stick to this case here, even if the results of the paper are slightly more general) and let \(d(k_1,a)\) be the dimension of the space of eigenforms of weight \(k_1\) and \(\wp\)-slope \(a\). Exploiting Teitelbaum's interpretation of cuspforms as harmonic cocycles and some explicit formulas for the action of the Hecke operators, the author (generalizing the main result of one of his previous papers) proves \(d(k_1,a)=d(k_2,a)\) when \(k_1\) and \(k_2\) are \(p\)-adically close and \(a\) is small enough (under some technical additional hypotheses). Then, assuming \(d(k_1,a)=1\), he is able to construct a \(\wp\)-adic family basically by taking generators for the space of eigenforms of \(\wp\)-slope \(a\) in a sequence of \(p\)-adically close weights \(k_1 < k_2 < \cdots < k_n < \cdots\). The main point is that such forms verify the condition on the \(T_{\mathfrak{q}}\)-eigenvalues for any prime \(\mathfrak{q}\) of \(A\), because otherwise one could find another independent eigenform of slope \(a\), contradicting the hypothesis \(d(k_1,a)=1\).