A note on Iwasawa \(\mu\)-invariants of elliptic curves (Q711428)

To discuss the present paper, first we need to discuss the work of \textit{R. Greenberg} and \textit{V. Vatsal} [Invent. Math. 142, No. 1, 17--63 (2000; Zbl 1032.11046)], in which they study the Selmer groups of isomorphic residual representations. To explain more precisely, let me first fix a prime \(p\). They studied the \(p\)-Selmer groups of the elliptic curves \(E_1, E_2\) defined over \(\mathbb Q\) with good ordinary reduction at \(p\). In particular, they supposed the residual representations given by the \(p\)-torsions, \(E_1[p]\) and \(E_2[p]\), are isomorphic as representations of the absolute Galois group \(G_{\mathbb Q}\). It is only natural to wonder whether \(\text{Sel}_p(E_1/ \mathbb Q)\) and \(\text{Sel}_p(E_2/ \mathbb Q)\) satisfy some kind of congruence relation. They provided a rather precise answer to this question in terms of their \(\lambda\)- and \(\mu\)-invariants. In particular, they proved that the \(\mu\)-invariant of one Selmer group is \(0\) if and only if that of the other is \(0\), and they also proved a precise relation between their \(\lambda\)-invariants in terms of the co-ranks of the local conditions at bad places away from \(p\) assuming their \(\mu\)-invariants are \(0\). Their most technologically innovative insight was the use of the non-primitive Selmer groups and adept use of various global duality results. They also proved a similar relation for the \(p\)-adic \(L\)-functions for the elliptic curves as well. The reviewer produced a similar result for the \(\pm\)-Selmer groups constructed by Kobayashi when \(E_1,E_2\) has good supersingular reduction at \(p>3\), but he has been told that Greenberg, Iovita, and Pollack had proved the result already in an unpublished paper. It is natural to ask whether their result can be generalized to the following situation: Suppose \(E_1[p^i]\) and \(E_2[p^i]\) are isomorphic. How are their \(\mu\)-invariants related? (If they are \(0\), it comes down to Greenberg and Vatsal's result.) The authors' answer to this question is that, \(\mu(E_1)\geq i\) if and only if \(\mu(E_2) \geq i\).