An algorithm of computing \(b\)-functions (Q1355292): Difference between revisions

Let \(K\) be a field of characteristic 0. Let \(A_n(K): =K[x_1, \dots, x_n] \langle \partial_1, \dots, \partial_n \rangle\) be the Weyl algebra and let \(\widehat {\mathcal D}_n (K):= K[[x_1, \dots, x_n]] \langle\partial_1, \dots, \partial_n \rangle\), where \(\partial_i =\partial/ \partial x_i\). For a polynomial \(f(x)\in K[x_1, \dots, x_n]\), the minimal monic polynomial \(b(s)\) in \(s\) satisfying \[ P(s,x,\partial) f(x)^{s+1} =b(s)f(x)^s \] for some \(P(s,x,\partial) \in\widehat {\mathcal D}_n (K)[s]\) (resp. \(A_n(K))\) is called the \(b\)-function associated with \(f\) and denoted by \(b_f (s)\) (resp. \(\widetilde b_f(s))\). This notion was introduced by I. N. Bernstein and M. Sato [\textit{M. Sato} and \textit{T. Shintani}, Ann. Math., II. Ser. 100, 131-170 (1974; Zbl 0309.10014)], independently. The author gives an algorithm to compute the \(b\)-function \(b_f\) and the operator \(P(s,x, \partial)\) above. He uses the Gröbner bases introduced by the author for left ideals of \(A_{n+1}(K)\) related to some filtration. To compute such bases he employs the homogenization technique.