On chains of classical prime submodules and dimension theory of modules. (Q537069)

Let \(R\) be an associative ring with unit element 1. A proper submodule \(P\) of a left \(R\)-module \(M\) is a prime (resp. classical prime) submodule if \(\text{Ann}(N/P)=\text{Ann}(M/P)\) (resp. \(\text{Ann}(N/P)\) is a prime ideal) for all submodules \(P\subsetneqq N\subseteq M\). A submodule \(N_1\) of \(M\) is said to be strongly contained in the submodule \(N_2\), \(N_1\subset_s N_2\), if \(N_1\subset N_2\) and \(\text{Ann}(M/N_1)\subset\text{Ann}(M/N_2)\). In a previous article [J. Algebra 305, No. 2, 1128-1148 (2006; Zbl 1114.16020)], the first-named author has defined the classical Krull dimension \(\text{cl.k.dim}(M)\) as the supremum of the (ordinal) lengths of all strong chains of prime submodules of \(M\). In the current paper, the notion of classical prime dimension \(\text{cl.p.dim}(M)\) is introduced by replacing ``prime'' with ``classical prime'' in the previous definition. It is shown that \(\text{cl.p.dim}(M)\) exists if and only if every ascending strong chain of submodules of \(M\) is finite, and that in this case \(\text{cl.k.dim}(M)\) exists as well, with \(\text{cl.k.dim}(M)\leq\text{cl.p.dim}(M)\). Also, if \(R\) has classical Krull dimension \(\dim(R)\), then every \(R\)-module has \(\text{cl.p.dim}(M)\leq\dim(R)\). In Sections 3 and 4, the authors look at modules with classical prime dimensions \(0\) and \(-1\), respectively. They make the conjecture that in general \(\text{cl.p.dim}(M)=-1\) if and only if \(\text{cl.k.dim}(M)=-1\), i.e., \(M\) has a prime submodule if and only if it has a classical prime submodule, and they prove that it holds whenever \(R\) satisfies the descending chain condition for prime ideals. In the last section, it is shown (Theorem 5.3) that if \(R\) is a left bounded, prime left Goldie ring, then \(\text{cl.p.dim}(M)=\text{cl.k.dim}(M)\) for all left \(R\)-modules \(M\) if and only if \(R\) is simple Artinian. Lemma 5.2 is false, any simple Artinian ring provides a counterexample. However, Theorem 5.3 remains true; a new proof is forthcoming.