Critical points over splitting subspaces

From MaRDI portal





If \(H\) is a Hilbert space, let \(N\) be a finite-dimensional subspace \((\dim N > 0)\) of \(H\) and let \(M = N^ \perp\). Suppose that \(G: H \to \mathbb{R}\) is a continuously differentiable functional which satisfies for some \(\rho > 0\), \(\delta > 0\), \(\rho \leq B < \infty\), and \(w_ 0 \in M\setminus \{0\}\) the following conditions: \[ G(v) \leq 0,\quad v \in N;\quad G(w) \geq \rho,\quad w \in M,\;\| w\| = \delta;\;G(sw_ 0+v) \leq B,\quad s \geq 0,\quad v \in N. \tag{*} \] The author obtains the following very interesting result which represents a generalization of the well-known saddle point theorem for linking subsets of splitting subspaces: under the above assumptions there are a constant \(c\) and a sequence \(\{u_ k\} \subset H\) such that \(\rho \leq c \leq 2B\) and \(G(u_ k) \to c\), \(G'(u_ k) \to 0\). We mention that the assumption (*) is weaker than in the usual form of the saddle point theorem [see for instance \textit{P. H. Rabinowitz}, Reg. Conf. Ser. 65, vii, 100 p. (1986; Zbl 0609.58002)]. Some applications of the above result to semilinear elliptic boundary value problems are given.











This page was built for publication: Critical points over splitting subspaces

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5289701)