On characteristic cones, clusters and chains of infinitely near points (Q4400910)

\textit{O. Zariski} [Am. J. Math. 60, 151-204 (1938; Zbl 0018.20101)] introduced the notion of complete ideals in regular local rings of dimension 2 and later suggested that such a theory seems impossible in higher dimensions. \textit{J. Lipman} [in: Algebraic geometry and commutative algebra, Pap. Honor M. Nagata, Vol. I, 203-231 (1988; Zbl 0693.13011)] showed that some of the features of complete ideals can be generalised to higher dimensions. Counterexamples were provided by \textit{C. Huneke} [Math. Ann. 275, 617-635 (1986; Zbl 0605.13003)] and \textit{S. D. Cutkosky} [in: AMS-IMS-SIAM Summer Res. Conf. Commutative Algebra, 1992, Contemp. Math. 159, 27-39 (1994; Zbl 0811.13001)]. First, let me describe Cutkosky's approach. Let \(\pi: Z\to X\) be a birational projective morphism of normal varieties. Then, one has the characteristic cone \(\widetilde{P}(Z/X)\), which for our purpose is defined as follows: This is the convex cone generated by the Cartier divisors \(D\) on \(Z\) (in \(N^1(Z/X)\otimes {\mathbb R}\)), where \({\mathcal O}(-D)=I{\mathcal O}_Z\) for some ideal \(I\subset {\mathcal O}_X\). Then the varieties sandwiched between \(Z\) and \(X\) correspond in a one-to-one fashion, to cells of this cone. In Zariski's situation, this cone is regular polyhedral. Cutkosky gives examples where this cone is non-polyhedral and even examples where the cone is not closed. The paper under review deals with two natural questions arising from these examples. First, to find conditions under which the number of sandwiches is finite. Secondly, are there conditions under which one can deduce good properties of the characteristic cone? Certain positive and negative results are obtained. For the entire collection see [Zbl 0890.00033].