Homogeneous maps as piecewise endomorphisms (Q4030337)

This paper continues previous work by both authors [J. Aust. Math. Soc., Ser. A 50, 279-296 (1991; Zbl 0731.16029) and Quaest. Math. 14, No. 4, 419-431 (1991; Zbl 0752.16020)] and the first author [Result. Math. 18, 125-132 (1990; Zbl 0718.16037)]. Let \({\mathcal C} = \{C_ \alpha\}_{\alpha \in A}\) be a set of submodules of the unitary right \(R\)-module \(G\), with \(\bigcup_{\alpha \in A} C_ \alpha = G\) and such that for every \(s \in S = \text{End}_ RG\), and every \(C_ \alpha \in {\mathcal C}\) there exists some \(\beta \in A\) such that \(s(C_ \alpha) \subseteq C_ \beta\). Then \(\text{PE}_ R(G,{\mathcal C}) = \{f \in M_ R(G);\) for every \(\alpha \in A\), \(f| C_ \alpha\) extends to some \(s\in S\}\) is the near-ring of piecewise endomorphisms of \(G\) determined by the cover \(\mathcal C\). \(M_ R(G)\) is the centralizer near-ring of all mappings of \(G\) to \(G\) whose action commutes with the action of \(R\). Then \(\text{End}_ R G \subseteq\text{PE}_ R(G,{\mathcal C})\subseteq M_ R(G)\). The authors are mainly concerned with when one or other of these containments are equalities. The first set of results deals with possible covers. It is shown that all PE near-rings arise from closed covers, and the set of closed covers is a complete distributive lattice. Two particular covers are singled out: \(\mathcal M\) consisting of all maximal submodules, and \(\mathcal P\) consisting of all maximal cyclic submodules. After one general result attention is restricted to modules for which \(\mathcal M\) and \(\mathcal P\) are covers (which are called mc-modules). If \(G\) is a semisimple \(R\)-module then \(\text{PE}_ R(G,{\mathcal P}) = M_ R(G)\) and if \({\mathcal M} \neq {\mathcal P}\), then \(\text{PE}_ R(G,{\mathcal M}) = \text{End}_ RG\). If \(R\) is a commutative ring with identity, then the following are equivalent: 1) \(R\) is a direct sum of fields; 2) \(R\) is Noetherian and \(\text{End}_ R(G) =\text{PE}_ R(G,{\mathcal M})\) for each mc-module \(G\) with \({\mathcal M} \neq {\mathcal P}\); 3) \(R\) is Noetherian and \(\text{PE}_ R(G,{\mathcal M})\) is a ring for each mc-module \(G\) with \({\mathcal M} \neq {\mathcal P}\). Next, commutative principal ideal rings are considered. After proving a reduction result, it is shown that if \(R\) is a PIR and \(G\) is any finitely generated \(R\)-module then \(M_ R(G) =\text{PE}_ R(G,{\mathcal P})\). If \(D\) is a Noetherian UFD, then \(\text{PE}_ D(G,{\mathcal P}) = M_ D(G)\) for each finitely generated \(D\)-module if and only if \(D\) is a PID. The final set of results produces a characterization of those commutative integral domains \(D\) such that \(M_ D(D^ 2) =\text{PE}_ D(D^ 2,{\mathcal P})\). Throughout the paper, useful examples are provided to show some of the boundaries of present knowledge.