The symmetric structure of the plus/minus Selmer groups of elliptic curves over totally real fields and the parity conjecture (Q1017397): Difference between revisions

Let \(E/\mathbb{Q}\) be an elliptic curve, \(F\) a real field and \(K/F\) a CM extension where every prime dividing the conductor of \(E\) splits (Heegner condition). Let \(Q\) be a prime of \(F\) which divides \(p\), splits in \(K/F\) and is of good supersingular reduction for \(E\). Let \(K_\infty\) be the \(\mathbb{Z}_p\)-extension arising from the ring class fields \(K[Q^n]\) and \(\Lambda\) the associated Iwasawa algebra. The paper deals with the structure of the minus part of the Selmer group \(\mathrm{Sel}_p(E/K_\infty)\) [as defined in \textit{S. Kobayashi}, Invent. Math. 152, No.~1, 1--36 (2003; Zbl 1047.11105)]. The main theorem asserts (with some additional hypotheses like, for example \([F_Q:\mathbb{Q}_p]=1\)) that \[ \mathrm{Sel}_p^-(E/K\infty)^\vee \sim \Lambda\otimes Y \otimes Y \otimes M \] where \( ^\vee\) denotes the Pontrjagin dual, \(Y\) is \(\Lambda\)-cotorsion and \(M\) is annihilated by a power of \(p\). The proof largely uses results from a previous paper of the same author [Compos. Math. 143, No.~1, 47--72 (2007; Zbl 1169.11022)] where the base field was \(F=\mathbb{Q}\) and the reader is often referred to that paper for many missing details. The key step is the definition of suitable local conditions for cohomology groups at a finite set of primes (including, among others, the ones dividing \(p\)), i.e. subgroups of some \(H^1\) which have to be cartesian and orthogonal with respect to a pairing defined between the whole cohomology groups. Once the right definition is given (and the properties of the pairing have been checked) the result follows from \textit{B. Howard} [Duke Math. J. 124, No.~1, 1--45 (2004; Zbl 1068.11071)] (to verify that the \(\Lambda\)-rank is 1 one needs the equidistribution of CM points). In the final section the main theorem and some results on the non-vanishing of \(L\)-functions are applied to prove that \[ \mathrm{corank}_{\mathbb{Z}_p} \mathrm{Sel}_p(E/K)\equiv \mathrm{ord}_{s=1}L(s,E) \pmod 2 \] (i.e. the \textit{parity conjecture} for this setting).