Equations involving fractional Laplacian operator: compactness and application (Q2344851)

In this article, the authors consider a Brézis-Nirenberg type problem [\textit{H. Brézis} and \textit{L. Nirenberg}, Commun. Pure Appl. Math. 36, 437--477 (1983; Zbl 0541.35029)] for the fractional Laplacian: \[ \begin{aligned} (-\Delta)^{\alpha}u & = |u|^{2^{\ast}_{\alpha}-2-\epsilon} u + \lambda u \quad\text{in } \Omega,\\ u &= 0 \quad\text{on } \partial \Omega. \end{aligned}\tag{1} \] Here, \(\Omega\subset \mathbb{R}^{N}\) is a bounded, smooth domain, \(\epsilon \in [0,2^{\ast}_{\alpha}-2)\), \(\lambda >0\), \(0< \alpha < 1\), \(2^{\ast}_{\alpha}= 2N/(N-2\alpha)\), and \((-\Delta)^{\alpha}\) denotes either the spectral or the restricted fractional Laplacian. In this setup, the authors prove the existence of infinitely many solutions of (1) by deducing strong compactness of suitably bounded solutions of the corresponding subcritical problems with \(\epsilon>0\) as \(\epsilon\rightarrow 0\). This extends the results of \textit{G. Devillanova} and \textit{S. Solimini} [Adv. Differ. Equ. 7, No. 10, 1257--1280 (2002; Zbl 1208.35048)] from the case \(\alpha=1\) to the case of \(\alpha\in(0,1)\). Due to the criticality of the exponent, if \(\epsilon=0\), the Palais-Smale condition fails and critical point theory cannot directly yield the desired solutions. Hence, following the ideas of Devillanova and Solimini [loc.~cit.], a more detailed compactness analysis of the (lack of) bubbling behavior of the subcritical problem is carried out. To this end, the authors use the Caffarelli-Silvestre extension [\textit{L. Caffarelli} and \textit{L. Silvestre}, Commun. Partial Differ. Equations 32, No. 8, 1245--1260 (2007; Zbl 1143.26002)] of the fractional Laplacian to formulate (1) as a local problem: \[ \begin{aligned} -\text{div}(y^{1-2\alpha} \nabla v) & = 0 \quad\text{in } \mathcal{C}_{\Omega},\\ v &= 0 \quad\text{on } \partial_L \mathcal{C}_{\Omega},\\ y^{1-2\alpha}\frac{\partial v}{\partial \nu} & = |v(x,0)|^{2^{\ast}_{\alpha}-2-\epsilon} v(x,0) + \lambda v(x,0) \quad\text{on } \Omega \times \{0\}, \end{aligned}\tag{2} \] where \(\mathcal{ C}_{\Omega}= \Omega \times (0,\infty)\) and \(\partial_{L}\mathcal{ C}_{\Omega}= \partial \Omega \times (0,\infty)\). Considering sequences \(v_n\) of solutions of the subcritical problems (2) with \(\epsilon_n>0\), which are uniformly (in \(\epsilon_n>0\)) bounded in the energy space \[ H_{0,L}^{1}(\mathcal{C}_{\Omega}):= \left\{v\in L^{2}(\mathcal{C}_{\Omega}): v= 0 \text{ on } \partial_{L}\mathcal{C}_L, \int_{\mathcal{C}_{\Omega}}y^{1-2s}|\nabla v|^2 dx < \infty \right\}, \] the authors prove strong convergence in \(H_{0,L}^{1}(\mathcal{C}_{\Omega})\) for \(v_n\) as \(\epsilon_n \rightarrow 0\). Hence, they show that no bubbling behavior is possible along these sequences. Compared to the local case of \(\alpha=1\), the main challenge in the analysis of (2) is the derivation of the necessary boundary estimates.