Resolution of singularities of 2-dimensional real analytic constrained differential systems (Q2166196)

The goal of the authors is the study of non-standard impasse points of implicit ODEs \[ A(x)\dot{x}=F(x) \] in the \(2\)-dimensional real case. By standard impasse points I mean transverse ones (see, e.g., R. Rabier and W. C. Rheinbolt 1994 for the real case, G. Thomas 2016 for extension to the complex case) where the solution has a branch point, the degree of which is related to the multiplicity of the point within the singular locus. However, non-standard singular points are not transverse ones, or are located at places where the singular locus is not smooth. What makes the non-standard problem more difficult is that it is necessary to desingularize the vector field near such a point, which is not feasible constructively in higher dimensions. One original parts of this paper is the use of resolution of singularities techniques. For this purpose, the authors consider the differential system as a 1-dimensional foliation on a real compact manifold with corners, where the system writes \[ \delta_i(x)\dot{x}=X_i(x) \] on a given open set \(U_i\) of its covering -- on each \(U_i\) this equation is a local equivalent system using the multiplication by the adjoint of \(A(x)\), to get a diagonal structure. Theorem \(A\) is the main result, claiming that any 2-dimensional such system is resolved globally -- through a finite number of weighted blow-ups -- to an elementary constrained system. The authors first establish a constructive method for a local resolution of singularities in Theorem \(14\). The way the blow-ups are built, with the help of a controllable Newton polygon, is detailed well in Section 4.1. The example of page 11 are not satisfactory, for the differential systems contain fake impasse points: \(xy\dot{x}=y\) and \(xy\dot{y}=x^2\) from equation (7) should be replaced with \[ x\dot{x}=1 \: \: y\dot{y}=x, \] in order to have an underlying irreducible differential ideal, and thus not introduce a discussion on an artificial collision of indeed independent systems (this is the same for the example of page 15). I guess it is possible to yield a similar example without this bias. This paper remains very pleasant and inspiring to read, for it gives a clear approach to how tackle with non-standard impasse points. The mixture of dynamical systems concepts like normal forms and modern differential geometry ones, like sheaves on compact manifolds, makes it a good starter for future researches in higher dimensions.