ON RESOLVING SINGULARITIES

DOI10.1112/S0024610701002745MaRDI QIDQ3151067zbMATH OpenWikidataFDO

Publication date 22 October 2002

Published in Journal of the London Mathematical Society (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/math/0102100, http://wrap.warwick.ac.uk/804/1/WRAP_Moody_resolving_singularities.pdf

zbMATH Keywords

Jacobian condition Nash blow-up Nash question non-singular blow-up

Mathematics Subject Classification ID

Singularities in algebraic geometry (14B05) Global theory and resolution of singularities (algebro-geometric aspects) (14E15) Modifications; resolution of singularities (complex-analytic aspects) (32S45)

Abstract: Let V be an irreducible affine algebraic variety over a field k of characteristic zero, and let (f_0,...,f_m) be a sequence of elements of the coordinate ring. There is probably no elementary condition on the f_i and their derivatives which determines whether the blowup of V along (f_0,...,f_m) is nonsingular. The result is that there indeed is such an elementary condition, involving the first and second derivatives of the

f_{i},

provided we admit certain singular blowups, all of which can be resolved by an additional Nash blowup. There is is a particular explicit sequence of ideals R=J_0, J_1, J_2,... subset R so that V_i=Bl_{J_i}V is the i'th Nash blowup of V, with J_i|J_{i+1} for all i. Applying our earlier paper, V_i is nonsingular if and only if the ideal class of J_{i+1} divides some power of the ideal class of J_i. The present paper brings things down to earth considerably: such a divisibility of ideal classes implies that for some Nge r+2 J_i^{N-r-2}J_{i+1}^{r+3}=J_i^NJ_{i+2}. Yet note that this identity in turn implies J_{i+2} is a divisor of some power of J_{i+1}. Thus although

V_{i}

may fail to be nonsingular, when the identity holds the {it next} variety V_{i+1} must be nonsingular. Thus the Nash question is equivalent to the assertion that the identity above holds for some sufficiently large i and N.

Review text

Let \(V\) be an irreducible affine algebraic variety over a field \(k\) of characteristic zero, and let \((f_0,\dots ,f_m)\) be a sequence of elements of the coordinate ring. There is probably no elementary condition on the \(f_i\) and their derivatives which determines whether the blow-up of \(V\) along \((f_0,\dots ,f_m)\) is non-singular. The result of the current paper is that there indeed is such an elementary condition, involving the first and second derivatives of the \(f_i,\) provided we admit certain singular blow-ups, all of which can be resolved by an additional Nash blow-up. This paper is the promised sequel of an earlier paper [\textit{J. A. Moody}, J. Algebra 189, No. 1, 90-100 (1997; Zbl 0891.14004)], in which the same program was carried out for individual vector fields. Indeed, the current paper generalizes the result of the mentioned paper to algebraic foliations of arbitrary codimension, and the case of codimension zero foliations corresponds to the problem of resolving the singularities of \(V.\) The results have a close connection with a question of Nash concerning resolutions. We now describe this briefly following a paper by \textit{J. Milnor} [Math. Intell. 17, No. 3, 11-17 (1995; Zbl 0846.01016)] where further references may be found. Let \(r=\dim(V).\) Suppose \(V=V_0 \subset W_0\) is an embedding in a non-singular variety over \(k.\) Then \(V_0\) lifts to a subvariety \(V_1\subset W_1=\text{Grass}_r(W_0)\) of the variety of \(r\)-planes in the tangent bundle of \(W_0\). The natural map \(\pi:V_1\to V_0\) is called the Nash blow-up of \(V_0\). It is the lowest blow-up where \(\pi^*(\Omega_{V_0/k})/\)torsion is locally free. Now we can repeat the process, giving a variety \(V_2\subset W_2=\text{Grass}_r(W_1)\) and so-on, and the question is whether eventually \(V_i\) is non-singular. There is a particular explicit sequence of ideals \(R=J_0, J_1, J_2,\dots \subset R\) so that \(V_0= \text{Bl}_{J_0}V\), \(V_1=\text{Bl}_{J_1}V\), \(V_2=\text{Bl}_{J_2}V,\dots .\) with \(J_i|J_{i+1}\) for all \(i.\) Applying our earlier paper [\textit{J. A. Moody}, Ill. J. Math. 45, No. 1, 163-165 (2001; Zbl 0989.13001)], \(V_i\) is non-singular if and only if the ideal class of \(J_{i+1}\) divides some power of the ideal class of \(J_i.\) The present paper brings things down to earth considerably: Such a divisibility of ideal classes implies that for this value of \(i\) and for some \(N\geq r+2\) \[ J_i^{N-r-2}J_{i+1}^{r+3}=J_i^NJ_{i+2}. \] Yet note that this identity in turn implies \(J_{i+2}\) is a divisor of some power of \(J_{i+1}.\) Thus although \(V_i\) may fail to be non-singular, when the identity holds the next variety \(V_{i+1}\) must be non-singular. Thus the Nash question is equivalent to the assertion that the identity above holds for some sufficiently large \(i\) and \(N.\)

Recommendations

Cited in

(15)

This page was built for publication: ON RESOLVING SINGULARITIES

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3151067)