Some Siegel modular standard \(L\)-values, and Shafarevich-Tate groups (Q533820)

The Bloch-Kato conjecture gives a formula for the leading term of a motivic \(L\)-function evaluated at an integer point. According to a conjecture of Deligne, a \textit{critical point} is an algebraic multiple of a certain period (which itself is only defined up to an algebraic multiple). The Bloch-Kato conjecture is a refinement of this in that it gives a factorization of the ratio of the \(L\)-value to the period. The authors' interest in this \(L\)-function is that it allows them to find primes dividing the size of the Shafarevich-Tate group. The paper under review enunciates two conjectures in this direction: they concern Siegel modular forms and their standard \(L\)-functions. Let \(S_{k,j}(\text{Sp}(4,\mathbb{Z}))\) denote the space of vector-valued Siegel modular forms of weight \(\det^k\otimes\, \text{Sym}^j\). Let \(F\in S_{k,j}(\text{Sp}(4,\mathbb{Z}))\) and let \(L(s,F,\text{st})\) denote the standard \(L\)-function of \(F\). {Conjecture 5.3} Let \(f=\sum_{n=1}^\infty a_n(f)q^n\in S_{k'}(\text{SL}(2,\mathbb{Z})\) be a normalized Hecke eigenform. Suppose that \(k'=j+2k-2\) with \(j\geq 0\) and \(k\geq 4\) and suppose that \(F\in S_{k,j}(\text{Sp}(4,\mathbb{Z}))\) is a Hecke eigenform such that for all primes \(p\) \[ \mu_F(p)\equiv p^{k-2}+p^{j+k-1}+a_p(f) \pmod{\lambda} \] where \(\lambda | \ell\) is a prime divisor of the algebraic part of \(L(f,j+k)\). Suppose that \(\ell>2j+2k-1\) and the \(\ell\) divides neither the \(j+2\)th Bernoulli number nor the \(k'\)th Bernoulli number. Suppose also that \(j\leq k-4\). Then for any even integer \(m\) with \(0<m\leq k-2\), \(m\neq j+2\) one has \[ \text{ord}_\lambda\left(\frac{\pi^{3(m-(j+2))}L(F,j+1,\text{st})}{L(F,m,\text{st})}\right)>0. \] To give evidence for this conjecture, they prove it in the case when \(F\) is a Saito-Kurokawa lift (necessarily this is when \(j=0\)) and they provide some computational evidence as well when \(j>0\). The authors state a second, similar conjecture, as well. {Conjecture 5.4} Let \(f=\sum_{n=1}^\infty a_n(f)q^n\in S_{k'}(\text{SL}(2,\mathbb{Z}))\) be a normalized Hecke eigenform. Suppose that \(k'=j+2k-2\) with \(j\geq 0\) and \(k\geq 3\) and that \(F\in S_{k,j}(Sp(4,\mathbb{Z}))\) is Hecke eigenform such that for all \(p\) \[ \mu_F(p)\equiv p^{k-2}+p^{j+k-1}+a_p(f) \pmod{\lambda} \] where \(\lambda | \ell\) is a prime divisor (in any field \(K\) containing the coefficients of \(f\) and \(F\)) of the algebraic part of \(L(f,j+k)\). Suppose that \(\ell>2j+2k-1\) and the \(k'\)th Bernoulli number is not divisible by \(\ell\) and neither is the \(j/2\)th. Suppose that \(j/2\) is odd and that \(j\leq 2k-6\) so that \((j/2)+1\) is even with \(0<(j/2)+1\leq k-2\). Take any even \(0<m\leq k-2\) but \(m\neq (j/2)+1\). Then \[ \text{ord}_\lambda\left( \frac{\pi^{3(m-((j/2)+1)} L(F,(j/2)+1,\text{st})}{L(F,m\text{st})}\right)>0. \] Since \((j/2)+1\) is even, the scalar-valued case \(j=0\) cannot be considered and so there is nothing to prove via the Saito-Kurokawa lift. Instead, assuming Harder's conjecture, the authors provide computational evidence for this conjecture.