Settings for a study of finite rank Butler groups. (Q2468541)

A (finite rank) Butler group \(G\) is the homomorphic image of a finite direct sum of subgroups of the additive rationals \(\mathbb{Q}\). That is \(G\cong X/K\), where \(X\) is (finite rank) completely decomposable and \(K\) is a pure subgroup. If the rank of \(K\) is \(n\), then \(G\) is called a \(B(n)\)-group. The \(B(0)\)-groups are thus the completely decomposable groups. The \(B(1)\)-groups have been studied extensively, particularly by these two authors. Relatively little is known about \(B(n)\)-groups for \(n>1\). In the present paper, tools used in the work on \(B(1)\)-groups are generalized and new tools are developed for the assault on \(B(n)\) for \(n>1\). A tent is a fundamental concept introduced by the authors in earlier work and generalized here. If, in the representation \(G\cong X/K\), we have \(m\) is the rank of \(X\) and \(I=\{1,2,\dots,m\}\), denote by \(P(I)\) the collection of subsets of \(I\). Then, as described in the paper, \(X\) determines a `tent', that is, a function \(t\colon P(P(I))\to T\), where \(T\) is the set of all types. The authors develop numerous other concepts, many of them combinatorial. The colorful language employed gives rise to lilting statements of results. For example: Theorem 2.12. No prime of \(G\) can pierce a lobe of its creel \(K\). A generous dollop of carefully described examples is perhaps the most helpful feature of the paper. Classification of general \(B(n)\)-groups looks to be a difficult, if not impossible task, even for \(n=2\). The best hope is to employ techniques such as those developed here to study special subclasses.