Björk's problem about filtered rings and graded rings (Q674649)

Let \(R\) be a filtered Noetherian ring with identity and \(M\) an \(R\)-filtered module. The associated graded ring of \(R\) is denoted by \(G(R)\) and the associated graded \(G(R)\) module of \(M\) is denoted by \(\text{gr}(M)\). \textit{J.-E. Björk} studied [Lect. Notes Math. 1404, 137-173 (1989; Zbl 0696.16006)] the relation between \(M\) being a module with a good filtration and \(\text{gr}(M)\) being a finitely generated module. If \(R\) is a ring with positive filtration and \(G(R)\) is a Noetherian ring, then \(M\) is a module with good filtration if and only if \(\text{gr}(M)\) is a finitely generated \(G(R)\) module. But when \(R\) is a Zariski filtered ring, we do not know if the conclusion is true. To be specific, Björk's problem is: Let \(R\) be a ring with a Zariskian filtration. Let \(M\) be a finitely generated \(R\) module with a separated filtration such that \(\text{gr}(M)\) is a finitely generated \(G(R)\) module. Does it follow that the filtration on \(M\) is good? In this note we solve the problem. Our result is: Theorem 1. Let \(R\) be a Zariski filtered ring and \(M\) a finitely generated \(R\) module with a separated filtration. Then \(M\) is a good filtered \(R\) module if and only if \(\text{gr}(M)\) is a finitely generated \(G(R)\) module. We get two corollaries of Theorem 1 which are about rings of differential operators and completions of filtered rings, respectively.