On the dynamics of gradients. Existence of invariant manifolds (Q1359539)

This is once more the famous gradient conjecture of René Thom discussed. The conjecture, quite old and still unsolved, goes as follows: Integral curves of grad \(f\) have tangents at their limit points \((f\) real analytic of \(n\) variables). A positive answer is given in the case where the tangent cone of \(f\) verifies additional properties (formulated in Chapter III). The problem is discussed in the language of dynamical systems (i.e., attractors, limit sets) and, consequently, ``spherical'' blow-ups are used, which are just usual blowing up with the sphere replacing the projective space and the points doubled [cf. \textit{F. Dumortier}, J. Differ. Equations 23, No. 1, 53-106 (1977; Zbl 0346.58002)]. However, in this paper we meet extraordinary clarity of formulations and notations, spherical blowing up included. A ``historical'' onset is presented (very clearly, too), and the result is globalized. It is worth noticing that the work of \textit{Hu Xing Lin} [``Sur la structure des champs de gradients de fonctions analytiques réelles'', Thèse, Paris VII (1992)] is used and given credit (this work, inspired by R. Thom himself, is very little known). A very readable and inspiring paper.