No label defined (Q4517567)

The point of study of this article is the multiplicative group of a field \(K\) denoted by \(K^*\). Of course, \(K^*\) has the property that every finitely generated finite subgroup is cyclic. \textit{P. M. Cohn} [Arch. Math. 13, 344-348 (1962; Zbl 0109.25804)] showed that any abelian group satisfying that condition can be embedded in \(K^*\) and later \textit{W. May} [Proc. Lond. Math. Soc. (3) 24, 295-306 (1972; Zbl 0232.20075)] showed how close one could get -- the problem still remains open as to precisely which groups are the multiplicative group of some field. It is shown that if \(G\) is a torsionfree divisible abelian group of infinite torsionfree rank, then \((Z/p)^* \times G\) is isomorphic to \(K^*\) for some \(K\) (necessarily with \(K\) of characteristic \(p\)). It is also show that the torsionfree rank of \(K^*\) is infinite unless \(K\) is algebraic over a finite field and also is at least the transcendence degree of \(K\) over the prime field. It is also shown that if \(k\) is a subfield of \(K\), then \(K^*/k^*\) has infinite torsionfree rank unless \(K\) is algebraic over a finite field or \(K/k\) is purely inseparable. This result was apparently presented by Reid at an AMS meeting in 1965. This last result had already been generalized by Wiegand and others. Suppose that \(K/k\) is a separable algebraic extension, \(K\) is not algebraic over a finite field and \(k_1,\ldots,k_r\) are proper intermediate fields. Consider \(K^*/k_1^* \cdots k_r^*\). If \(r \leq 2\), this has infinite torsionfree rank and finite \(n\)-torsion for all \(n\). If \(r > 2\), essentially anything can happen. See \textit{J.-L. Colliot-Thélène, R. M. Guralnick} and \textit{R. Wiegand}, J. Pure Appl. Algebra 106, 233-262 (1996; Zbl 0846.12003)] for results and examples. There are examples with \(r=2\) when this quotient is trivial if we drop the assumption that \(K/k\) is algebraic. For the entire collection see [Zbl 0944.00041].