Steinberg homology, modular forms, and real quadratic fields (Q2663310): Difference between revisions

Let \(R\) be a commutative ring, let \(E\) be a real quadratic field, and let \(\Gamma\subset \operatorname{GL}_2(\mathbb Z)\) be a congruence subgroup. The authors study the homomorphism \(\psi_{\Gamma,E}:=-\partial\), where \(\partial: H_1(\Gamma,C)\rightarrow H_0(\Gamma,\operatorname{St}(\mathbb Q^2;R))\) is the connecting homomorphism arising from the short exact sequence \[ 0\longrightarrow \operatorname{St}(\mathbb Q^2;R)\longrightarrow \operatorname{St}(E^2;R)\longrightarrow C\longrightarrow 0, \] where \(\operatorname{St}\) denotes the Steinberg module. In the case \(R=\mathbb C\), the authors prove that the image of \(\psi_{\Gamma,E}\) can be described in terms of modular symbols. For general \(R\), they prove that \(\operatorname{Im}(\psi_{\Gamma,E})\) always lies in a certain cuspidal subspace \(H_0^{\mathrm{cusp}}(\Gamma,\operatorname{St}(\mathbb Q^2;R))\). Using work of \textit{H. W. Lenstra jun.} [Invent. Math. 42, 201--224 (1977; Zbl 0362.12012)], and assuming the Generalized Riemann Hypothesis (GRH), they prove that the cokernel of \(\psi_{\Gamma,E}\) is a finitely-generated, torsion \(R\)-module (Theorem 9.3). For specific choices of \(\Gamma\), the authors are able to prove stronger results (still assuming GRH). For example, in the case where \(\Gamma=\Gamma_1(N)\) or \(\Gamma_1(N)^{\pm}\), it is shown that \(\psi_{\Gamma,E}\) is surjective. The authors also indicate that unconditional versions of their results should follow from a suitably developed theory of so-called toral periods (at least in the case where \(R=\mathbb C\)), and they conclude by giving some numerical evidence.