Dicritical logarithmic foliations (Q2458369): Difference between revisions

Consider the germ of a (singular) holomorphic foliation \(\mathcal{F}\) of \((\mathbb{C}^2,0)\) without saddle-nodes in its desingularization, which is called a generalized curve in [\textit{C. Camacho, A. Lins Neto} and \textit{P. Sad}, J. Differ. Geom. 20, 143--174 (1984; Zbl 0576.32020)]. If \(\mathcal{F}\) is non-dicritical, the second author has shown that \(\mathcal{F}\) has a logarithmic model [Ann. Inst. Fourier 53, No. 3, 787--814 (2003; Zbl 1032.32019)]. Now, the authors prove that \(\mathcal{F}\) has a weak logarithmic model even in the dicritical case. This model is of the form \(\eta=0\) for \[ \eta=\sum_{i=1}^r\lambda_i\,\frac{df_i}{f_i}. \] To define \(\eta\), they have to choose a representative set \(\mathcal{S}\) of separatrices of \(\mathcal{F}\). They also prove that the possible equisingularity types of \(\mathcal{S}\) are bounded by a function of the Milnor number of \(\mathcal{F}\). \(\mathcal{L}\) is a weak logarithmic model for \(\mathcal{F}\) in the sense that the reduction of singularities of \(\mathcal{L}\) is longer than that of \(\mathcal{L}\), and both reductions are equal outside a ``scape set'' of non-singular points of \(\mathcal{F}\) in its dicritical components. This scape set can be chosen to be a singleton, which determines the residues \(\lambda_i\).