An explicit basis of modular symbols on function fields (Q305154): Difference between revisions

Let \(\mathbb F_q\) be the finite field with \(q\) elements, \(A = \mathbb F_q[T]\) and \(K = \mathbb F_q(T)\) the polynomial ring and field of rational functions in \(\mathbb F_q\), respectively, with completion \(K_{\infty} = \mathbb F_q((1/T))\). Let \(\mathfrak n\) be a non-trivial ideal of \(A\) and \(\Gamma_0(\mathfrak n)\) the Hecke congruence subgroup of \(\Gamma =\mathrm{GL}_2(A)\) with conductor \(\mathfrak n\). The group \(\Gamma\) and thus \(\Gamma_0(\mathfrak n)\) acts simplicially on \(\mathcal T\), the Bruhat-Tits tree of \(\mathrm{GL}_2(K_{\infty})\). These data are related to the Drinfeld modular curve \(X_o(\mathfrak n)\) and its arithmetic. In particular, for any commutative ring \(R\) (\(R = \mathbb Z\), \(\mathbb Z[1/q]\), \(\mathbb Q\), \(\mathbb C\), \dots) let \(\mathbf M_{\mathfrak n}(R)\) be the group of \(R\)-valued modular symbols of \(\Gamma_0(\mathfrak n)\) as defined by \textit{J. T. Teitelbaum} [Duke Math. J. 68, No. 2, 271--295 (1992; Zbl 0777.11021)]: \(\mathbf M_{\mathfrak n}(R)=H_0(\Gamma_0(\mathfrak n),M\otimes_{\mathbb Z} R)\), where \(M = \mathbb Z[\mathbb P^1(K)]^0\), the free abelian group of divisors of degree 0 on the projective line \(\mathbb P^1(K) = \mathbb P^1(A)\). As shown by Teitelbaum, \(\mathbf M_{\mathfrak n}(\mathbb Z)\) is the quotient of \(\mathbb Z[\mathbb P^1(A/\mathfrak n)]\) modulo the three families of relations {\parindent=0.7cm\begin{itemize}\item[(1.1)] \((u:v)+(-v:u) = 0\); \item[(1.2)] \((u:v)+(v:-u-v)+(-u-v:u) = 0\); \item[(1.3)] \((u':v)-(\delta_1u:\delta_2v)=0\), \end{itemize}} where \(\delta_1,\delta_2 \in \mathbb F_q^{\ast}\), \((u:v) \in \mathbb P^1(A/\mathfrak n)\). It is known that \(\mathbf M_{\mathfrak n}\) is related with the homology of \(\Gamma_0(\mathfrak n)\setminus \mathcal T\) (Prop. 4.1, [loc. cit.]) and with the group \(\mathbf H_{\mathfrak n}\) of harmonic cochains on \(\Gamma_0(\mathfrak n) \setminus \mathcal T\) (Theorem 4.2, Lemma 4.4, [loc. cit.], [the reviewer and \textit{U. Nonnengardt}, Int. J. Math. 6, No. 5, 689--708 (1995; Zbl 0858.11025)]). In particular, the torsion \(\mathbf M_{\mathfrak n}\) is trivial (cyclic of order \(q+1\)) if \(\mathfrak n\) is prime of odd (even) degree. Suppose from now on that \(\mathfrak n = \mathfrak p\) is prime of degree \(d\). The principal result of this well-written paper is Theorem 1.1, which specifies a natural basis of \(\mathbf M_{\mathfrak p}\) in case \(d\) is odd. A more detailed version, valid for arbitrary \(d\), is given in Proposition 5.13 and Theorem 5.16, which describe a direct decomposition of \(\mathbf M_{\mathfrak p}\) according to degree of generators of \(\mathbb Z[\mathbb P^1(A/\mathfrak p)]\) and bases of the respective pieces. Other important results are Theorem 1.2 (a special case of Theorem 7.10), giving the linear independence of certain Hecke images of the ``winding element'' \(\mathbf e \in \mathbf M_{\mathfrak p}(\mathbb Z)\) and Theorem 1.3, which establishes a lower bound for the number of newforms \(F \in \mathbf H_{\mathfrak n}\) whose \(L\)-series \(L(F,x)\) does not vanish at the critical point \(s=1\).