Totally Crawley groups (Q1100289)

A \(p\)-group \(G\) is a \(\lambda\)-Crawley group if \(p^{\lambda}G=0\) and any two groups \(H_ 1\) and \(H_ 2\) are isomorphic if \(p^{\lambda}H_ i=Z(p)\) and \(H_ i/p^{\lambda}H_ i\cong G\). A \(p\)-group \(G\) is a \(\lambda\)-Zippin group if \(p^{\lambda}G=0\) and for two groups \(H_ i\), \(i=1,2\), with \(H_ i/p^{\lambda}H_ i\cong G\) every isomorphism \(p^{\lambda}H_ 1\rightarrowtail p^{\lambda}H_ 2\) extends to an isomorphism \(H_ 1\rightarrowtail H_ 2\). The group \(G\) is totally Crawley if for each limit ordinal \(\alpha\) less than or equal to the length of \(G\) the quotient group \(G/p^{\alpha}G\) is \(\alpha\)-Crawley, and finally \(G\) is totally Zippin for each limit ordinal \(\alpha\) less than or equal to the length of \(G\) the quotient group \(G/p^{\alpha}G\) is \(\alpha\)-Zippin. Do the (weaker) Crawley properties imply the corresponding Zippin properties? \textit{Ch. Megibben} [Pac. J. Math. 107, 205-212 (1983; Zbl 0521.20035)] showed that it is undecidable in ZFC set theory whether every (\(\omega\)-)Crawley group of cardinality \(\omega_ 1\) is (\(\omega\)-)Zippin. - It is helpful here that \(\omega\)-Zippin groups are direct sums of cyclic groups, but there is no structure theorem for \(\lambda\)-Zippin groups for \(\lambda >\omega\). This makes the ``Crawley problem'' very difficult. However, it was shown by \textit{Ch. Megibben} [Proc. Am. Math. Soc. 91, 15-18 (1984; Zbl 0543.20036)] that totally Zippin groups of countable length are totally projective and thus direct sums of countable groups. This gives a certain handle on the question whether totally Crawley groups of countable length are totally Zippin. The authors prove that this question is undecidable in ZFC set theory. More precisely they prove: Theorem A. (V = L) Every totally Crawley group of countable length is a direct sum of countable groups. Theorem B. (PFA) For any countable limit ordinal \(\lambda\) there is a totally Crawley group of length \(\lambda\) and cardinality \(\omega_ 1\) which is not a direct sum of countable groups. The authors state that they were unable to prove Theorem B under the more standard assumption \(MA+\neg CH\) but had no resort to the stronger Proper Forcing Axiom. The proofs of these theorems are difficult and along the way the authors obtain a number of results of independent interest such as the following. Let \(\lambda\) denote a countable limit ordinal throughout. Corollary 4.1. (PFA) There is a \(\lambda\)-Crawley group of length \(\lambda\) which is not \(\lambda\)-Zippin. It is left open whether there is a model of standard set theory in which every \(\lambda\)-Crawley group is \(\lambda\)-Zippin for all \(\lambda\). Theorem 1.2. \((V=L)\) Every \(\lambda\)-Crawley group of cardinality \(\omega_ 1\) has quotient length \(\leq \lambda\). - Corollary 1.4. There is a \(\lambda\)-Crawley group G of cardinality \(\omega_ 1\) which is a Fuchs-5 group and has quotient length \(=\lambda\). - ``Quotient length'' is a property of filtrations of a group. Theorem 2.2. If \(G\) is a Fuchs-5-group of length \(\lambda\) such that for all countable \(p\)-groups \(C\), \(p^{\lambda}Ext(G,C)=0\) then \(G\) is \(\lambda\)-Crawley. Theorem 3.2 characterizes \(C_{\lambda}\)-groups of cardinality \(\omega_ 1\) in terms of filtrations. Huber's invariant \(\Gamma\) (G) is used at some crucial junctions. The paper is rich in results and techniques.