Nash modification on toric surfaces (Q742209)

In this paper the author studies from a combinatorial point of view the Nash modification of a complex affine toric variety of dimension two (a \textit{toric surface}), i.e., the process whereby each singular point is substituted by the set of limiting positions of tangent planes at nearby regular points. The description used here was recently introduced by D. Gregoriev and P. Milman. It views the surface as the closure of the image of the monomial mapping \[ \Phi : {({\mathbb C}^{\star})}^2 \to {\mathbb C}^r, \qquad {\Phi}(x_1,x_2) \mapsto (x^{\gamma _1}, \dots, x^{\gamma _r}) \, , \] where \(\gamma _i = (\gamma_{i1},\gamma _{i2}) \in {\mathbb Z}^2\), and \(\xi =\{x_1, \dots, x_r\}\) generate the group \({\mathbb Z}^2\). The collection \(\xi = \{\gamma_i:i=1, \dots ,r\}\) is called a \textit{set of monomial exponents for the toric surface \(X\)}. If \({i_0}=({i_0},j_{0})\) is an element of \(\xi\) satisfying certain mild conditions, a description of an affine chart \(X_1\) (which is again a toric surface) of the Nash modification of \(X\), in terms a new set of monomial exponents \(\xi_{{i_0}{j_0}}\), is presented. If the semigroup of \({\mathbb Z}^2\) spanned by \(\xi_{i_{0} j_{0}}\) can be generated by two elements (whence \(X_1\) is regular) we stop. Otherwise, we repeat the process. To choose the original pair \((i_0,j_0)\) one uses a method involving a suitable linear form \(L(x,y)=ax+by\) from \({\mathbb R }^2\) to \(\mathbb R\), with \(a\) and \(b\) integers. The author proves that this process, or algorithm, always stops, i.e., eventually the set of monomial exponents involved generates a subgroup of \({\mathbb Z}^2\) that can be spanned by two elements. He also gives a bound for the number of steps necessary to achieve this partial desingularization, in terms of certain numbers associated to the initial monomial exponents and the linear form \(L\) used. In the last section he proves that his result implies that given a toric surface \(X\) as before, with function field \(K \approx {\mathbb C}(x,y)\), and a valuation \(V\) of \(K\) whose value group is not of the form \({\mathbb Z}+ \beta {\mathbb Z}\), with \(\beta\) real and irrational, then the method above yields an uniformization of \(V\). A similar result (for monomial valuations) was obtained in a more general context in [\textit{P. Gonzalez Perez} and \textit{B. Teissier}, ibid. 108, No. 1, 1--48 (2014; Zbl 1307.14071)].