The wild kernel of exceptional number fields. (Q1421297)

Let \(p\) be a prime. Let \(F\) be a number field and let \(F_n= F(\mu_{p^n})\). Let \(\Gamma_n= \text{Gal}(F_n/F)\). Let \(C_n= \text{Cl}({\mathcal O}_{F_n}[1/p])\). \textit{F. Keune} [K-Theory 2, 625--645 (1989; Zbl 0705.19007)], building on ideas of Tate, described a natural homomorphism \(J_n: (\mu_{p^n}\otimes C_n)_{\Gamma_n}\to K_2({\mathcal O}_F)/p^n\). He proved that if \(n\) is sufficiently large and if either \(p\) is odd or if \(p= 2\) and \(F\) is non-exceptional then \(J_n\) induces an isomorphism between \((\mu_{p^n}\otimes C_n)_{\Gamma_n}\) and the \(p\)-Sylow subgroup \(WK_2(F)_p\) of the wild kernel of \(F\). In this article, the author shows that when \(p= 2\) and \(n\) is large the map \(J_n: (\mu_{p^n}\otimes C_n)_{\Gamma_n}\to WK_2(F)_2\) is neither surjective nor injective in general. He shows that the cokernel has order at most 2 and characterises exactly when the map is surjective. He provides bounds on the size of the kernel and gives an example of when the kernel of \(J_n\) is non-trivial for large \(n\).