Meromorphic vector fields and elliptic fibrations (Q1885563)

A holomorphic vector field is said to be complete, if its flow is well-defined for all the complex values of time. The paper under review deals with semi-complete meromorphic vector fields. Roughly speaking, a holomorphic (meromorphic) vector field in a domain is said to be semi-complete, if there is no evident obstruction for it to be a restriction of a complete vector field on some ambient manifold. In particular, the integral of the time differential along any nonclosed path in a complex orbit should not vanish. The precise definition was introduced in a earlier paper by the author [Ann. Inst. Fourier 46, No. 2, 411--428 (1996; Zbl 0853.34002)] and is recalled in the paper under review. Each meromorphic vector field is tangent to a singular holomorphic foliation by analytic curves. In dimension two any local meromorphic vector field is a meromorphic multiple of a holomorphic vector field with isolated singularities; these are exactly the singularities of the corresponding foliation. In the first part of the paper the author studies germs of semi-complete meromorphic vector fields in dimension two at isolated singular points of the corresponding foliations. It is assumed that there is only a finite number of leaves accumulating to the singular point. The author gives an analytic classification of these semi-complete germs up to a nonzero functional factor. The list of normal forms of semi-complete germs splits into three cases according to the type of the corresponding holomorphic vector field. In each one of the two cases, when its linear part either vanishes, or just is a nilpotent Jordan cell, there are three infinite normal form series, each one numerated by an integer parameter. In each individual series the holomorphic vector field is the same (it is quadratic), and the meromorphic multipliers are all the possible integer powers of one and the same simple polynomial of low degree (at most 6). In the case, when the linear part has at least one nonzero eigenvalue, the singularity of the holomorphic vector field is a resonant saddle. There are again three infinite normal form series. The first series has a functional parameter given by the nonlinear terms of the saddle in the chart, where the meromorphic multiplier is a power of a monomial of the type \(x^ny^m\) (\(m\) and \(-n\) are the eigenvalues of the linear part). In each one of the two other series the meromorphic factor is again monomial of the type \(x^cy^d\), and the saddle is linear. The proof uses a desingularization by blowing ups (given by \textit{A. Seidenberg}'s theorem [Am. J. Math. 90, 248--269 (1968; Zbl 0159.33303)]) of the complex singularity of the corresponding foliation. It is shown that the pull-back to the desingularized manifold of a semi-complete meromorphic vector field is again semi-complete. Its singular points are elementary (i.e., their linearizations have at least one nonzero eigenvalue). The author shows that they are resonant saddles, and in ``most'' cases the monodromies along circuits in their separatrices are of finite order. In the proof the author uses some results of his previous papers: [Ann. Inst. Fourier 46, No. 2, 411--428 (1996; Zbl 0853.34002), \textit{E. Ghys} and \textit{J. C. Rebelo}, Ann. Inst. Fourier 47, No. 4, 1117--1174 (1997); erratum ibid. 50, No. 3, 1019--1020 (2000; Zbl 0938.32019), Bol. Soc. Mat. Mex., III. Ser. 5, No. 2, 359--395 (1999; Zbl 0948.34067)]. The second part of the paper concerns elliptic fibrations which are local holomorphic fibrations in dimension two by elliptic curves in a neighborhood of a singular fiber. These fibrations were described by Kodaira. The author retrieves Kodaira's description by using his own new method (similar to that mentioned above) involving semi-complete meromorphic vector fields. One of the advantages of the author's method is that it gives the structure of the singularities of the foliation associated to the fibration. He claims that using the same method it is possible to derive the Kodaira's complete description of all the elliptic surfaces [\textit{K. Kodaira}, Ann. Math. (2) 77, 563--626 (1963; Zbl 0118.15802)]. The author's method is geometric, simple and very powerful. It is believed that similar ideas can be applied in studying many different questions of algebraic geometry and holomorphic foliations.