A theorem on the closure of \(\Omega\)-pure subgroups of \(C_{\Omega}\) groups in the \(\Omega\)-topology (Q1825286)

Let A, B be abelian p-groups. When is Tor(A,B) a direct sum of countable groups (dsc's)? This old question of Nunke is the root from which the present paper grew. Various partial results in a number of papers lead to the following concepts and results. Let \(\lambda\) be a limit ordinal less than \(\Omega\), the first uncountable ordinal. The author uses the abbreviation G(\(\alpha)\) for the more standard \(p^{\alpha}(G)\). An (abelian p-)group is a \(C_{\lambda}\)-group if G/G(\(\alpha)\) is a dsc for all \(\alpha <\lambda\), and a normal \(C_{\Omega}\)-group if G is a \(C_{\Omega}\)-group and \(G(\Omega)=0\). The \(\Omega\)-topology is the topology on the \(C_{\lambda}\)-group G which has the family G(\(\alpha)\), \(\alpha <\Omega\), as a neighborhood basis of 0. A subgroup H of G is \(\alpha\)-pure if the extension \(H\rightarrowtail G\twoheadrightarrow G/H\) represents an equivalence class in Ext(G/H,H)(\(\alpha)\). Actually a number of properties equivalent to \(\alpha\)-purity are used. The author considers the following Theorem 3 the main result of the paper (in the statement of the theorem intersection must be replaced by union): If G is a \(C_{\Omega}\)-group, \(\mu\) a limit ordinal, and \(\{A_ i:\) \(i<\mu \}^ a \)smoothly ascending chain of \(\Omega\)-pure subgroups of G then \(\overline{\cup_{i<\mu}A_ i}=\cup_{i<\mu}\bar A_ i\). In particular, if \(\bar A_ i\subset A_{i+1}\) for each i, then \(\cup_{i<\mu}A_ i\) is closed in G. The applications of the main theorem are mostly to iterated torsion products. The results are dependent on set theoretic axioms which are not the usual ones. For a cardinal \(\kappa \geq \aleph_ 2\), the normal \(C_{\Omega}\)-group G is called a \(\kappa\)-Kurepa extension of its \(\Omega\)-pure subgroup A if \(| A| =\aleph_ 1\) and \(| \bar A| \geq \kappa\). A Kurepa family \({\mathcal F}\) is a collection of subsets of \(\Omega\) such that for every countable ordinal \(\beta\), \(| \{X\cap \beta:\) \(X\in {\mathcal F}\}| \leq \aleph_ 0\). Kurepa's Hypothesis is independent of \(ZFC+GCH\) and asserts that there exist Kurepa families of cardinality greater than \(\aleph_ 1\). Theorem 5: If \(\kappa \geq \aleph_ 2\) is a regular cardinal, then there exists a \(\kappa\)-Kurepa extension if and only if there exists a Kurepa family of cardinality \(\kappa\). There are a large number of results involving ``\(\Omega\)-developments'' and an invariant \(\Gamma\) G which cannot all be explained here. A sample result is Theorem 17: For a positive integer n, there are no \(\aleph_ n\)-Kurepa extensions if and only if \(Tor(G_ 1,...,G_ n)\) is a dsc for every collection \(G_ 1,...,G_ n\) of normal \(C_{\Omega}\)-groups.