On a theorem of Ono and Skinner (Q5929405)

Let \(g(z)=\sum_{n\geq 0} c(n) q^n\) be a Hecke eigenform of half-integral weight with algebraic integer coefficients. In 1998, \textit{K. Ono} and \textit{C. Skinner} [Ann. Math. (2) 147, 453-470 (1998; Zbl 0907.11017)] showed that if a certain condition holds (``\(g(z)\) is good''), then for all but finitely many rational primes \(\ell\) there exist infinitely many square-free integers \(m\) such that \(|c(m)|_\ell=1\). Here ``good'' means that there is some theta function \(h(z)\) such that \(g(z) h(z)\) is a cusp form whose new part is not of cm type. Ono and Skinner conjectured that all functions that are not linear combinations of theta functions are good. In the paper under review this conjecture is proved. The author establishes the result by showing that if \(g(z)\) is not good, then all coefficients \(c(n)\) are zero except for \(n\) of the form \(n=st^2\) with \(s\) a square-free product of primes dividing the level of \(g(z)\). \textit{M.-F. Vigneras} [Modular Functions of one Variable. VI, Proc. Int. Conf., Bonn 1976, Lect. Notes Math. 627, 79-103 (1977; Zbl 0373.10027)] showed that modular forms with this property must be in the span of the theta functions. The author also shows how the hypothesis that \(g(z) h(z)\) is a cusp form in the definition of ``good'' above can be dropped and the arguments of Ono and Skinner extended directly to accomodate the non-cuspidal case. Results related to Ono and Skinner's, using different methods, have been established by \textit{J. H. Bruinier} [Duke Math. J. 98, 595-611 (1999; Zbl 0966.11019)] and \textit{N. Jochnowitz} [Congruences between modular forms of half-integral weights and implications for class numbers and elliptic curves (unpublished manuscript)]. Another (less elementary) proof of the conjecture of Ono and Skinner has been given by \textit{J. Jimenez Urroz} and \textit{K. Ono} [Math. Res. Lett. 7, 205-212 (2000; Zbl 0958.11031)].