On units related to the arithmetic of elliptic curves with complex multiplication (Q1924908)

Let \(p\equiv 3\pmod 4\) be a prime number \(>3\) and denote the class number of \(K= \mathbb{Q} (\sqrt {-p})\) by \(h\). Let \(H\) be the Hilbert class field of \(K\). The values of a Hecke character \(\psi\) of \(K\) with conductor \((\sqrt {-p})\) generate a field \(T\) of degree \(h\) over \(K\). It can be shown that \(M= HT\) has degree \(h^2\) over \(K\). Since the ideals of \(K\) become principal in \(H\) and in \(T\), one may assign to each ideal \({\mathfrak a}\) of \(K\) a unit of \(M\) by dividing two generators coming from extending \({\mathfrak a}\) to \(T\) and to \(H\). \textit{B. Gross} [Lect. Notes Math. 776 (1980; Zbl 0433.14032)] has shown that these units (which play an important role in the arithmetic of a certain elliptic curve \(A(p)/ H\)) give rise to a 1-cocycle on \(\text{Cl} (K)\) with values in the unit group \(E_M\), and asks whether its cohomology class is trivial. \textit{A. Aiba} [Kodai Math. J. 13, No. 1, 30-38 (1990; Zbl 0716.11025)] has carried out computations in the simplest non-trivial case (that of \(p= 23\)) using Robert's machinery for elliptic units; he answers a strong form of Gross' question in the positive. Using refined information about elliptic units, the author presents an explicit formula for computing these units and shows that the strong form of the question has, in general, a negative answer. However, it is proved that this cohomology class does indeed vanish when \(h=3\), and the author carries out computations for all \(K\) with \(h=3\), as well as one example with \(h=9\).