Crystalline subrepresentations and Neron models (Q5929463)

Let \(K\) be a finite extension of \(\mathbb{Q}_p\), \(G\) be its absolute Galois group, \(A\) be an abelian variety defined over \(K\), \(\mathcal A\) be its Néron model (over the valuation ring of \(K\)) and \(\Gamma\) be the geometric points of the group of components of \({\mathcal A}_0\), this last one being the special fiber of \(\mathcal A\). One knows, following Grothendieck, that for all prime \(l\neq p\), \(\Gamma(l)\simeq H^1(I,T_l(A))_{\text{tors}}\), where \(I\) is the inertia subgroup of \(G\) and \(T_l(A)\) is the \(l\)-adic Tate module of \(A\). One knows also that this formula is wrong for \(l = p\). The aim of the authors is to obtain such formula when \(l = p\). To do that, they define the functors \(\text{Chrys}_h\) on the category of \(\mathbb{Q}_p\)-representation of \(G\): if \(V\) is such a representation, \(\text{Chrys}_h(V)\) is the maximal crystalline subrepresentation of \(V\) with Hodge-Tate weights in \([0,h]\). The authors prove that if \(K/\mathbb{Q}_p\) is unramified, then \(\Gamma(p) \simeq R^1\text{Chrys}(T_p(A))_{\text{tors}}\).