Mather discrepancy as an embedding dimension in the space of arcs (Q1707958)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Mather discrepancy as an embedding dimension in the space of arcs
scientific article

    Statements

    Mather discrepancy as an embedding dimension in the space of arcs (English)
    0 references
    0 references
    0 references
    0 references
    4 April 2018
    0 references
    For an algebraic variety \(X\) over a field \(k\), its space of arcs \(X_\infty\) (introduced in [\textit{J. F. Nash jun.}, Duke Math. J. 81, No. 1, 31--38 (1995; Zbl 0880.14010)]) is intrisically associated with it, and it bears important information about the singularities of \(X\). A cylinder set is preimage of a constructible set in some jet scheme \(X_m\) by the truncation morphism \(\psi_m : X_\infty \to X_m\). The stable points of \(X_\infty\) are fat points of finite codimension which are the generic points of irreducible cylinders. Given a stable point \(P\), it is important to understand the properties of the local ring \(\mathcal{O}_{X_\infty,P}\) and its completion \(\widehat{\mathcal{O}_{X_\infty,P}}\). For example it is known that the latter is Noetherian ([\textit{A. J. Reguera}, Compos. Math. 142, No. 1, 119--130 (2006; Zbl 1118.14004)]). A main goal of this article is to compute or evaluate the dimension of the local ring. Let \(E\) be an exceptional divisor in a resolution of singularities \(\pi : Y \to X\), with \(k_E\) being the discrepancy of \(X\) with respect to \(E\). The Mather discrepancy for \(E\) ([\textit{T. de Fernex} et al., Publ. Res. Inst. Math. Sci. 44, No. 2, 425--448 (2008; Zbl 1162.14023)]) is defined as \(\hat{k}_E := \mathrm{ord}_E(\hat{K}_{Y/X})\). Suppose \(X\) is a reduced scheme of finite type over field of characteristic 0, and \(\nu\) is a divisorial valuation on an irreducible component \(X_0 \subset X\) whose center lies in \(X_{\mathrm{sing}}\). Take a resolution of singularities \(\pi : Y \to X\) such that the center of \(\nu\) on \(Y\) is a divisor \(E\). It defines a cylinder set \(N_{eE}\) of the arcs with contact \(e \geq 1\) with \(E\), and so its generic point is a stable point \(P_{eE} \in X_\infty\). In general, for any stable point \(Q\), \(\dim\mathcal{O}_{X_\infty,Q} \leq \operatorname{codim} X_\infty Z(Q)\) ([\textit{L. Ein} and \textit{M. Mustaţă}, Proc. Symp. Pure Math. 80, Pt. 2, 505--546 (2009; Zbl 1181.14019)]). In order to study \(\widehat{\mathcal{O}_{X_\infty,P_{eE}}}\), \(X_0\) is embedded in a complete intersection affine scheme \(X'\) of dimension \(\dim(X_0)\), which is an overweight deformation of an affine toric variety, associated to \(\nu\). Using this is proved that \(\operatorname{embdim}\mathcal{O}_{(X_\infty)_{\mathrm{red}},P_{eE}} = \operatorname{embdim}\widehat{\mathcal{O}_{(X_\infty),P_{eE}}} = e(\hat{k}_E + 1)\) (see also [\textit{A. J. Reguera}, J. Algebra 494, 40--76 (2018; Zbl 1388.13001)]; Am. J. Math. 131, No. 2, 313--350 (2009; Zbl 1188.14010)]). In other words, the embedding dimension of \(\mathcal{O}(X_\infty)_{\mathrm{red}},P_{eE}\) is the codimension of \(N_{eE}\) as a cylinder set. Moreover, it is described explicitly a minimal system of coordinates for \((\mathcal{O}_{(X_\infty)_{\mathrm{red}},P_{eE}})\), that is, a regular system of parameters with \(e(\hat{k}_E + 1)\) coordinates. To obtain a lower bound for \(\dim \widehat{\mathcal{O}_{(X_\infty),P_{eE}}}\), \(X\) is embedded in a complete intersection scheme \(X'\) with the same Mather discrepancy, and with a divisor \(E'\) over \(X'\) which defines the same valuation \(\nu = \nu_{E'}\). The Mather-Jacobian log-discrepancy with respect to \(E\) is defined as \(a_{MG}(E,X) := \hat{k}_E -\nu(Jac_X)+1\). Then the ring \(\widehat{\mathcal{O}_{(X_\infty),P_{eE}}}\) could be replaced by \(\widehat{\mathcal{O}_{(X'_\infty),P'_{eE}}}\) and this permits to obtain a lower bound \(\dim \widehat{\mathcal{O}_{(X_\infty),P_{eE}}} \geq ea_{MG}(E, X)\). In particular, when \(X\) is normal and complete intersection, \(\dim \mathcal{O}_{(X_\infty)_{\mathrm{red}},P_{eE}} \geq e(k_E+1)\) holds. If moreover \(\operatorname{codim}_{X_\infty} P_E = 1\), it follows that the discrepancy \(k_E \leq 0\).
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    space of arcs
    0 references
    divisorial valuations
    0 references
    embedding dimension
    0 references
    Mather discrepancy
    0 references
    stable point
    0 references
    0 references