Wild kernels and divisibility in \(K\)-groups of global fields (Q740866)

The \(K\)-theory of fields is a rather deep invariant, which contains a wealth of arithmetic information about the field in question. From the definition of \(K_{0}\) this would not be clear at all and it took several decades to get a clear picture about how arithmetic invariants are encoded in the higher \(K\)-groups. Banaszak has contributed a whole series of articles contributing to the development of this picture for number fields. A rather early result of Moore was an exact sequence \[ K_{2}(F)\overset{H}{\longrightarrow}\coprod\mu(F_{v})\longrightarrow \mu(F)\longrightarrow0, \] where \(F\) is the number field, \(F_{v}\) runs through its Archimedean and non-Archimedean local fields and \(\mu\) denotes the groups of roots of unity. The arrow \(H\) is built from the Hilbert symbols at all places (and thus rather close to class field theory). Earlier work of Banaszak generalized this sequence to higher \(K\)-groups. The kernel of the arrow \(H\) in Moore's sequence as well as Banaszak's higher sequences are called the \textit{wild kernel}. In the ``classical case'' of \(K_{2}\) this wild kernel consists of those symbols which are impossible to detect by any Hilbert symbol, which makes them particularly mysterious. Classical work of Tate developed a picture that the wild kernel should not be too far from the group of divisible elements; there should at worst be a discrepancy of index \(2\). In fact, for many number fields, they are equal. In the present paper, Banaszak develops this theory further. He considers rather general global fields, and pays particular attention to the case of positive characteristic. A key slogan is to reduce the equality of wild kernels with divisible elements, under some strictly necessary hypotheses, to the (now proven) Quillen-Lichtenbaum conjecture. For number fields, this goes back to earlier work of Banaszak (odd primes) and Hutchinson and Weibel (at \(2\)). The present paper extends this to general global fields. Once these results are available, a panorama of related results becomes possible, e.g. the trivialization of certain homology classes going from \(\mathrm{GL}(\mathcal{O})\) to \(\mathrm{GL}(F)\) or related to comparing with étale \(K\)-theory. All these applications are carefully discussed in the paper.