Ramification and nearby cycles for \(\ell\)-adic sheaves on relative curves (Q888374)

This paper compares the characteristic cycles defined by Kato and the characteristic cycles defined by Abbes and Saito for monogenic extensions. As a corollary, the author reinterprets a formula due to Deligne and Kato that computes the dimension of the nearby cycle complex on a relative curve over an excellent strictly henselian trait. Let us be more precise. Let \(K\) be a complete discrete valuation field with ring of integers \(\mathcal O_K\) and residue field \(F\). Suppose that \(F\) has characteristic \(p>0\) which need not to be perfect; and \(K\) can be of either equal characteristic or mixed characteristic. Let \(G_K\) denote the absolute Galois group of \(K\) and \(P_K\) the wild inertia subgroup. Let \(\mathbb F\) be a finite field of characteristic different from \(p\), that admits a non-trivial character \(\psi_0: \mathbb F_p \to \mathbb F^\times\). Given a representation \(\rho: \mathrm{Gal}(\overline K / K) \to \mathrm{GL}_n(\mathbb F)\), using the ramification theory of Abbes and Saito [\textit{T. Saito}, J. Inst. Math. Jussieu 8, No. 4 769--829 (2009; Zbl 1177.14044); Math. Ann. 352, No. 3, 567--580 (2012; Zbl 1247.11146)], one can associate a characteristic cycle to \(\rho\): \[ \mathrm{CC}_{\psi_0}(\rho) \in (\Omega_F^1(\log) \otimes_F \overline F)^{\oplus (n-n_0)}, \] where \(n_0\) is the dimension of the subspace invariant under \(\rho(P_K)\). If we assume moreover that \(\rho\) factors through the Galois group of a finite extension \(L\) of \(K\) for which \(\mathcal O_L\) is generated by one element over \(\mathcal O_K\), Kato's theory of Swan conductor with differential values [\textit{K. Kato}, Adv. Stud. Pure Math. 12, 315--342 (1987; Zbl 0658.12016)] associates a characteristic cycle \( \mathrm{KCC}_{\psi_0(1)}(\rho) \). The main theorem of this paper is that \[ \mathrm{CC}_{\psi_0}(\rho) = \mathrm{KCC}_{\psi_0(1)}(\rho). \] This should be thought of as a refined version of the comparison result of Proposition 6.7 in [\textit{A. Abbes} and \textit{T. Saito}, Am. J. Math. 124, No. 5, 879--920 (2002; Zbl 1084.11064)]. The proof is to explicit exhibit both sides using the definition, where the work on the left hand side takes up most of the paper. The significance of this paper lies in: \(\mathrm{CC}_{\phi_0}(\rho)\) is defined without the auxiliary monogenic assumption. So it has the potential for generalization; see also [\textit{T. Saito}, J. Inst. Math. Jussieu 8, No. 4 769--829 (2009; Zbl 1177.14044)] and [\textit{A. Abbes} and \textit{T. Saito}, Tohoku Math. J. (2) 63, No. 4, 775--853 (2011; Zbl 1239.14012)].