Multiple \(\mathbb {S}^{1}\)-orbits for the Schrödinger-Newton system. (Q2249871)

The paper is concerned with the Schrödinger equation \[ (-\varepsilon i\nabla +A)^2u + V(x)u = \frac {1}{\varepsilon ^2}\left (\frac {1}{| x| }*| u| ^2\right )u \quad \text{in } \mathbb {R}^3. \] Here, \(u\) is in a ``magnetic'' Sobolev space, \(A\: \mathbb {R}^3\to \mathbb {R}^3\) is the magnetic and \(V\: \mathbb {R}^3\to (0,\infty)\) the electric potential. In the case \(A=0\) this equation is equivalent to the Schrödinger-Newton system. It is assumed in this paper that \(V\) is invariant and \(A\) is equivariant with respect to the action of a closed subgroup \(G\) of \(\mathrm O(3)\). The solutions the authors look for, satisfy the condition \(u(gx)=\tau (g)u(x)\), where \(\tau \: G\to S^1\) is a given homomorphism of groups (so-called intertwining solutions). The main theorem gives a lower bound for the number of \(S^1\)-orbits (with respect to the action \((e^{i\theta},u)\mapsto e^{i\theta }u\)) of solutions under a certain compactness assumption on the level sets of \(V\) and for \(\varepsilon >0\) small enough. This extends a result contained in [\textit{S. Cingolani} et al., Discrete Contin. Dyn. Syst., Ser. S 6, No. 4, 891--908 (2013; Zbl 1260.35198)]. The proofs use equivariant Morse theory. The strongest conclusion concerns the case when all critical orbits are nondegenerate and gives, for each \(k\), a lower bound for the number of orbits whose Morse index is \(k\) (in terms of the rank of a mapping between certain \(k\)th equivariant cohomology groups).