Characterizing a class of Warfield modules by Ulm submodules and Ulm factors (Q5943759)

All modules are \(R\)-modules for \(R\) a discrete valuation domain. There are two main results. A strengthening of a 1986 result of Hill, Megibben: Theorem 2.4. Let \(G\) be a module whose torsion submodule is a direct sum of cyclics, let \(n\in\mathbb{N}\) and \(\sigma\geq\omega \) be an ordinal. The following are equivalent: (1) \(G\) is Warfield; (2) \(p^nG\) is Warfield; (3) \(G/p^\sigma G\) is Warfield and \(p^\sigma G\) is completely decomposable. In the special case of the class \(\mathcal H\) (the class of modules with the property that the torsion submodule is a direct sum of cyclics and the quotient modulo torsion is divisible of arbitrary rank), the hypothesis on \(p^\sigma G\) cannot be dropped and an example is given of a torsion-free module of rank 2 that is not simply presented yet has a simply presented zero-th Ulm factor. Theorem 2.7. A module whose torsion submodule is a direct sum of cyclics is Warfield if and only if its zero-th Ulm factor is simply presented and its first Ulm submodule is completely decomposable. -- An example is given which shows that this result can not be extended to ordinals \(\sigma>\omega\). Some of the results are finally summarized in Corollary 2.10. The following are equivalent for a module \(G\) whose torsion submodule is a direct sum of cyclics and which is either strictly reduced or has a torsion zero-th Ulm factor: (1) \(G\) is Warfield. (2) \(G\) is simply presented. (3) \(G\) is a direct sum of modules of torsion-free rank one. (4) \(p^\sigma G\) is completely decomposable for every ordinal \(\sigma\geq\omega\) and \(G/p^\sigma G\) is Warfield. (5) \(p^nG\) is Warfield for every \(n\in\mathbb{N}\). (6) \(p^nG\) is simply presented for every \(n\in\mathbb{N}\). (7) \(p^nG\) is a direct sum of modules of torsion-free rank one for every \(n\in\mathbb{N}\).