On uniserial modules that are not quasi-small. (Q2497460)

A module is `quasi-small' if whenever it is a direct summand of a direct sum \(\bigoplus_{i\in I}N_i\) it is a direct summand of a finite subsum \(\bigoplus_{i\in I'}N_i\) for some finite \(I'\subseteq I\). There has been a great deal of fairly recent progress in the decomposition theory of serial modules. For example \textit{G.~Puninski} [J. Lond. Math. Soc., II. Ser. 64, No. 2, 311-326 (2001; Zbl 1048.16003)] showed that a direct summand of a serial module need not be serial. In the other direction, one has [\textit{N.~V.~Dung} and \textit{A.~Facchini}, J.~Algebra 193, No. 1, 102-121 (1997; Zbl 0885.16008)] that a weak Krull Schmidt Theorem (referring to monogeny and epigeny classes) holds for serial modules over a ring iff every uniserial module over that ring is quasi-small. In this paper it is shown that if \(M\) is a direct sum of uniserial modules which are not quasi-small then every direct summand of \(M\) is serial; also that if \((U_i)_{i\in I}\) and \((V_j)_{j\in J}\) are uniserial modules which are not quasi-small then \(\bigoplus_iU_i\simeq\bigoplus_jV_j\) implies that there is a bijection \(\sigma\colon I\to J\) such that \(U_i\simeq V_{\sigma i}\) for all \(i\). On the way it is shown that a uniserial module \(U\) is not quasi-small iff it is countably generated and \(U_m\varsubsetneq U=U_e\) where \(U_m\) is the intersection of all submodules of \(U\) which are isomorphic to \(U\) and \(U_e\) is the sum of all kernels of epimorphisms from \(U\) to \(U\). It is also shown that for any uniserial modules \(U,V\) and any index set \(I\), if \(U^{(I)}\simeq V^{(J)}\) then \(U\simeq V\). \textit{G. Puninski} [in J.~Pure Appl.~Algebra 163, No. 3, 319-337 (2001; Zbl 1025.16005)] gave a classification of the pure-projective modules over a nearly simple chain domain. The proof used ideas from the model theory of modules and Puninski asked whether there is a proof which does not use these techniques. Such a proof is given by the author in the final section of the paper.