Observations on an elliptic problem with a jumping nonlinearity by the Conley index (Q1108472)

I study the existence of solutions for a nonlinear Dirichlet problem using a recent generalization (in some sense) of the Morse index: the notion of homotopy index given by \textit{C. C. Conley} [``Isolated invariant sets and the Morse index'' (1978; Zbl 0397.34056)]. The frame of this work is the following: I firstly consider the parabolic equation associated to the elliptic one and find an invariant set S of bilateral solutions. Then a variational characterization of the first problem, as the search of stationary points of a suitable functional, will guarantee the existence in S of a solution. This method has been employed by \textit{H. Amann} and \textit{E. Zehnder} [Ann. Sc. Norm. Super. Pisa, Cl. Sci., IV Ser. 7, 539-603 (1980; Zbl 0452.47077)] yet, who apply the theory of Conley to a finite dimensional space, after a reduction of the problem. Here, using the generalization of \textit{K. P. Rybakowski} [Trans. Am. Math. Soc. 269, 351-382 (1982; Zbl 0468.58016)] it is possible to face the problem directly, treating with the same arguments also the existence of a bilateral solution of the parabolic equation. The specific problem considered here may be written as \(\Delta u+g(u)=0\) where g has different derivatives at plus and minus infinity. This problem has been studied by \textit{B. Ruf} [Ann. Mat. Pura Appl., IV. Ser. 128, 133-151 (1981; Zbl 0475.35046)], essentially by degree methods, establishing a connection between the problem and a corresponding homogeneous equation. This connection is emphasized here by the Conley index which seems to be a sharper tool than the degree. It makes possible in fact a ``connectness result'' that seems not be reachable using the degree (computing the degree in 2.3-2.4 you get \((-1)^ k\) while the index gives the k-dimensional sphere allowing to distinguish different situation also when k maintains the same parity). The final result is a slight improvement of that by H. Amann and E. Zehnder (loc. cit.) considering also different derivatives in \(0^+\) and \(0^-\).