The torsion of \(p\)-ramified Iwasawa modules for \(\mathbb Z^2_p\)-extensions (Q1025900)

Let \(p\) be a prime number and \(k\) a finite extension of the field of rational numbers \({\mathbb Q}\). Let \(\tilde{k}\) be the composite of all \({\mathbb Z}_ p\)--extensions of \(k\). We have \(\text{Gal}(\tilde{k}/k)\cong {\mathbb Z}^ d\), with \(r_ 2 (k)+1 \leq d\leq [k:{\mathbb Q}]\) being \(d=r_ 2 (k)+1\) if \(k\) is abelian. R. Greenberg observed that there is no example of a number field with \(r_ 2(k)\neq 0\) such that \(\text{Tor}_{\Lambda_{ \tilde{k}/k}} X_{\tilde{k}}\neq 0\) where \(\text{Tor}_{\Lambda_{ \tilde{k}/k}} X_{\tilde{k}}\) is the submodule of \(X_{\tilde{k}}\), the Galois group of the maximal pro--\(p\)--abelian extension unramified outside all the \(p\)--primes, which consists of all \(\Lambda_{\tilde{k}/k}\)--torsion elements. Therefore we have: \textbf{Question:} When \(k\) is not totally real, that is, \(r_ 2(k)\neq 0\), is the torsion submodule \(\text{Tor}_{\Lambda_{ \tilde{k}/k}} X_{\tilde{k}}\) always trivial? In the paper under review, the author proves that if \(k={\mathbb Q} (\sqrt{-m})\) is an imaginary field such that \(p\) does not divide the class number \(h_ k\) of \(k\), then if \(p\neq 3\) or if \(p=3\) and \(m\not\equiv 3\bmod 9\), then \( \text{Tor}_{\Lambda_{ \tilde{k}/k}} X_{\tilde{k}}=0\).