\(D\)-modules generated by rational powers of holomorphic functions (Q824249)

Let \(f\) be holomorphic function on a complex manifold \(X\) and consider the hypersurface \(Z= f^{-1}(0)\subset X\). To \(f\) is associated its Bernstein-Sato polynomial \(b_f(s)\in \mathbb{Q}[s]\) with the property \(b_f(\alpha) {\mathcal D}_Xf^{-\alpha}\subset b_f(\alpha) {\mathcal D}_Xf^{-\alpha+1}\) (\(\alpha>0\) is rational), so that the two \({\mathcal D}_X\)-modules agree when \(b_f(\alpha)\neq 0\). The question at hand is the converse: is it true that \(b_f(\alpha)= 0\) implies that \({\mathcal D}_Xf^{-\alpha}\neq {\mathcal D}_Xf^{-\alpha+1}\)? The basic assumption is that the singular subset of \(Z\) consists of isolated points in \(Z\) which we may then assume is a single point \(x\) (with maximal ideal \(\mathbf{m}_x\)), so that, when \(\alpha >1\), \[ {\mathcal D}_Xf^{-1} /{\mathcal D}_Xf^{-\alpha}= ({\mathcal D}_X/{\mathcal D}_X\mathbf{m}_x)^{n_x} \] for some integer \(n_x= n_x(\alpha)\). A main result by the author is that that the implication holds when the Milnor monodromy is semisimple, and he also gives a fairly complicated example when it does not hold. The proof is accomplished by studying the Kashiwara-Malgrange filtration and by employing the fact that the (reduced) Bernstein-Sato polynomial is the minimal polynomial of the action of \(-\partial_t t \) on the saturated Brieskorn lattice.