Anticyclotomic Iwasawa theory of abelian varieties of \(\mathrm{GL}_2\)-type at non-ordinary primes (Q6201174)

Let \(K\) be an imaginary quadratic field and let \(f\) be a weight 2 newform of level \(N_0\) with \((N_0, \mathrm{disc}(K))=1\). Write \(N_0=N^+N^-\) where primes dividing \(N^+\) (resp. \(N^-\)) are split (resp. inert) in \(K\) and assume \(N^-\) is the squarefree product of an odd number of primes. Let \(p\geqslant 5\) be a prime unramified in \(K\) and not dividing \(N_0\), and let \(K_\infty\) be the anticyclotomic \(\mathbb{Z}_p\)-extension of \(K\) with associated Iwasawa algebra \(\Lambda=\mathcal{O}_v[[T]]\) (where \(\mathcal{O}_v\) is a ring extension of \(\mathbb{Z}_p\)). The paper deals with Iwasawa theory for the (Pontrjagin dual of the) Selmer group of an abelian variety \(A_f\) of \(\mathrm{GL}_2\)-type over \(K_\infty\), and the main result is an inclusion (under several technical hypotheses) of a \(p\)-adic \(L\)-function into the characteristic ideal of the dual of the Selmer group, i.e., a divisibility relation in the Main Conjecture for this setting. The authors consider elements \({\mathcal{L}}_{f,m}\in\Lambda/(\gamma^{p^m}-1)\) (where \(\gamma\) is a topological generator of \(\mathrm{Gal}(K_\infty/K)\)), basically defined at the finite layers \(K_m\) of \(K_\infty/K\), which verify relations like \[ \pi_{m+1,m} \mathcal{L}_{f,m+1} = a_p(f)\mathcal{L}_{f,m}- \xi_{m-1} \mathcal{L}_{f,m-1}, \tag{$*$} \] where \(a_p(f)\) is the \(p\)th Fourier coefficient of \(f\), \(\pi_{m+1,m}\) is a (canonical) projection from level \(m+1\) to level \(m\) and \(\xi_{m-1}\) is a norm map going in the opposite direction. Using Sprung-type \(2\times 2\) matrices (involving \(a_p(f)\) and cyclotomic polynomials in \(\Lambda\)), they interpolate the \(\mathcal{L}_{f,m}\) to provide elements \(\mathcal{L}_f^\sharp, \mathcal{L}_f^\flat \in \Lambda= \varprojlim_m \Lambda/(\gamma^{p^m}-1)\) satisfying relations analogue to \((*)\), and define the \(p\)-adic \(L\)-functions as \[ L_p^\sharp(f,K):=\mathcal{L}_f^\sharp(\mathcal{L}_f^\sharp)^\iota \quad \text{and}\quad L_p^\flat(f,K):=\mathcal{L}_f^\flat(\mathcal{L}_f^\flat)^\iota, \] where \(\iota\) is the involution on \(\Lambda\) defined by the inversion \(\gamma\longrightarrow \gamma^{-1}\). As in \textit{M. Bertolini} and \textit{H. Darmon} [Ann. Math. (2) 162, No. 1, 1--64 (2005; Zbl 1093.11037)], such functions describe the special values \(L(f\otimes\chi,1)\) for any finite order character \(\chi\) of \(\mathrm{Gal}(K_\infty/K)\). Since these values are known to be nonzero for almost all \(\chi\), at least one of \(L_p^\sharp(f,K)\) or \(L_p^\flat(f,K)\) is nonzero (actually both when \(a_p(f)=0\), i.e., in the supersingular reduction case). For the other ingredient of a Main Conjecture the authors study primitive \(Q\)-systems of local cohomology classes for the Tate modules of abelian varieties. One of the main features of a \(Q\)-system \(\{d_m\}_{m\geqslant 0}\) is a relation similar to \((*)\): \[ \mathrm{cor}_{m+1,m} d_{m+1} = a_p(f) d_m-\mathrm{res}_{m-1,m} d_{m-1}. \] Via a \(Q\)-system and \(2\times 2\) matrices as above, they define Coleman maps (whose construction and properties are studied in detail) on the Iwasawa cohomology groups. These provide the local conditions for the definition of the Selmer groups \(\mathrm{Sel}_\bullet(K_\infty,A_{f,\infty})\) where \(\bullet\in\{\sharp,\flat\}\). Finally, the authors produce a bipartite Euler system of Heegner classes (to agree with the ``bipartite'' nature of the \(\sharp\) and \(\flat\) Selmer groups) for an abelian variety \(A_f\). An Euler system-type argument (closely related to the ones in [\textit{H. Darmon} and \textit{A. Iovita}, J. Inst. Math. Jussieu 7, No. 2, 291--325 (2008; Zbl 1146.11057)] and [\textit{R. Pollack} and \textit{T. Weston}, Compos. Math. 147, No. 5, 1353--1381 (2011; Zbl 1259.11101)]) leads to the desired inclusion \[ L_p^\bullet(f,K) \subseteq \mathrm{Char}_\Lambda(\mathrm{Sel}_\bullet(K_\infty,A_{f,\infty})^\vee) \quad\quad\text{with } \bullet\in\{\sharp,\flat\}. \] In particular, the above mentioned nontriviality of \(L_p^\bullet(f,K)\) shows that the corresponding Selmer group is \(\Lambda\)-cotorsion. All results are obtained under several (but not unusual for this type of results) technical hypotheses too long to be reported here. Anyway, it is worth mentioning that the final divisibility relation is proved in the following cases \begin{itemize} \item[i)] \(p\) splits in \(K\), \(a_p(f)\neq 0\) and \(f\) is \(p\)-isolated (i.e., not congruent to any other newform modulo primes dividing \(p\)); \item[ii)] \(p\) is inert in \(K\), \(a_p(f)=0\) (supersingular reduction) and the Hecke field of \(f\) is \(\mathbb{Q}\). \end{itemize} The two cases require quite different proofs, mainly for the construction of the \(Q\)-systems (and, consequently, the Coleman maps) and of the Heegner classes involved in the proof of the main result.