On the conductor formula of Bloch (Q1767044)

From MaRDI portal
scientific article
Language Label Description Also known as
English
On the conductor formula of Bloch
scientific article

    Statements

    On the conductor formula of Bloch (English)
    0 references
    0 references
    0 references
    3 March 2005
    0 references
    This \(151\)-page paper proves a vast generalisation of a formula for the Artin conductor proved by \textit{S. Bloch} [in: Algebraic geometry, Proc. Summer Res. Inst., Brunswick/Maine 1985, part 2, Proc. Symp. Pure Math. 46, No.~2, 421--450 (1987; Zbl 0654.14004)] in the case of curves and conjectured by him to hold in general. Let \(\mathfrak o\) be a discrete valuation ring, \(K\) its field of fractions, and \(F\) the residue field. Recall that for a regular proper flat \(\mathfrak o\)-scheme \(X\), the Artin conductor of \(X\) is defined as \[ \text{ Art}(X)= \chi(X_{\bar K})-\chi(X_{\bar F})+{\text{ Sw}}(X_K) \] where \(\chi\) is the \(l\)-adic Euler characteristic and \(\text{Sw}\) is the \(l\)-adic Swan conductor, \(l\) being a prime different from the charateristic of \(F\). In [loc.~cit.], Bloch defined the ``localised self-intersection class of the diagonal'' \((\Delta_X,\Delta_X)\) as an element of the group \(A_0(X_F)\) of \(0\)-cycles on the closed fibre \(X_F\) modulo rational equivalence. The ``conductor formula'' he conjectured was the equality \[ {\text{ Art}}(X)=-\deg(\Delta_X,\Delta_X). \] It has recently been observed that the formula for the Milnor number of an isolated singularity follows from this; see \textit{F. Orgogozo} [Ann. Inst. Fourier 53, No. 6, 1739--1754 (2003; Zbl 1065.14005)]. The authors prove the conductor formula under the hypothesis that the reduced closed fibre has normal crossings as a divisor on \(X\) (Theorem~6.2.3). The general case would follow from this special case if embedded resolution of singularities holds in a strong sense for the reduced closed fibre. The best earlier result required the more restrictive hypothesis that the multiplicities of the irreducible components of \(X_F\) be prime to the characteristic of \(F\) [\textit{T. Chinburg, G. Pappas, M. J. Taylor}, Math. Res. Lett. 7, No. 4, 433--446 (2000; Zbl 1097.14501)]. In fact, the authors' main result concerns a more general version -- involving correspondences -- of the conductor formula. A correspondence \(\Gamma\) on \(X_K\) gives rise to an endomorphism \(\Gamma^*\) on the \(l\)-adic étale cohomology, and hence to a Swan conductor \(\text{ Sw}(\Gamma,X_K)\). They show (Theorem~6.3.1) that \({\text{ Sw}}(\Gamma,X_K)\) is an integer independent of \(l\) and, if the reduced closed fibre has simple normal crossings as a divisor on \(X\), the equality \[ {\text{ Sw}}(\Gamma,X_K)=-\deg[[X,\Gamma]] \] holds, where \([[X,\Gamma]]\) is the ``logarithmic localised intersection product'' (Section~5.4) in a suitable graded piece of the Grothendieck group of \(X_F\). The special case \(\Gamma=\Delta_X\) (Theorem~6.2.5) is equivalent to the previous theorem. The proof makes extensive use of logarithmic geometry. The authors define a ``logarithmic self-intersection class'' \((\Delta_X,\Delta_X)_{\log}\) in \(A_0(X_F)\) by replacing \(\Omega^1_{X|{\mathfrak o}}\) in the definition of \((\Delta_X,\Delta_X)\) by the sheaf of differential \(1\)-forms with logarithmic poles and prove that \(\text{Sw}(X_K)=-\deg(\Delta_X,\Delta_X)_{\log}\) (Theorem~6.2.5). For the main theorem, de Jong's alterations are needed. The authors also provide emendations of some results from a related paper by the second author [Duke Math. J. 57, No.2, 555--578 (1988; Zbl 0687.14004)].
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references