Existence of infinitely many stationary solutions of the \(L^{2}\)-subcritical and critical NLSE on compact metric graphs (Q683494)

The interesting paper under review deals with existence of stationary solutions for the nonlinear Schrödinger equation equation on a general compact metric graph \(\mathcal{G}.\) Precisely, the author proves existence of critical points for the nonlinear Schrödinger energy functional \[ E(u,\mathcal{G})=\dfrac{1}{2}\int_{\mathcal{G}}|u'|^2\;dx -\dfrac{1}{p}\int_{\mathcal{G}}|u|^p\;dx \] for \(p\in(2,6],\) under the mass constraint \[ \|u\|^2_{L^2(\mathcal{G})}=\mu>0. \] For suitable \(\lambda\in\mathbb{R}\) such critical points solve the stationary Schrödinger equation with the focusing pure power nonlinearity \[ u''+|u|^{p-2}u=\lambda u \] on each edge of \(\mathcal{G}\) with Kirchhoff conditions at the nodes. The author proves existence of ground states, that is, global minimizers of the constrained energy. Moreover, the bound states are also studied which are critical points that do not need to be minimizers. Both the subcritical \((p\in(2,6))\) and the critical \((p=6)\) regimes are studied, showing that the situation changes significantly. In the subcritical case there exist infinitely many stationary solutions, with energy increasing to infinity, for every value of the mass and regardless the topology of \(\mathcal{G}.\) In the critical regime both ground state and bound states exist only for masses smaller than a threshold value that depends on whether the graph has at least one terminal edge (an edge of \(\mathcal{G}\) such that one of its endpoint is a vertex of degree one) or not. When \(\mathcal{G}\) has a terminal edge then the threshold value of the mass is equal to the \(L^2\)-critical mass \(\mu_{\mathbb{R}^+}=\sqrt{3}\pi/4\) on the half-line \(\mathbb{R}^+.\) In all other cases the threshold value coincides with the \(L^2\)-critical mass \(\mu_{\mathbb{R}}=\sqrt{3}\pi/2\) on the whole real line \(\mathbb{R}^+.\) The machinery employed relies on variational techniques and modified versions of the Gagliardo--Nirenberg inequality.