Nonnegative solutions of a fractional sub-Laplacian differential inequality on Heisenberg group (Q5963383)

The main result focuses on giving necessary and sufficient conditions for a solution of a fractional sub-Laplacian differential inequality to be zero. More precisely, for \((p,\gamma,\alpha)\in (1,\infty)\times [0,2(p-1)(n+1)\big)\times(0,2)\), the authors consider the inequality NEWLINENEWLINE\[NEWLINE\;|g|_{\mathbb H^n}^\gamma[u(g)]^p\leq (-\Delta_{\mathbb H^n})^{\alpha/2}u(g),\leqno{(*)_{\gamma,\alpha}}NEWLINE\]NEWLINE NEWLINEsuch that \(g\) belongs to the Heisenberg group \(\mathbb H^n\), \(|\cdot|_{\mathbb H^n}\) symbolizes the \(\mathbb H^n\)-norm, and NEWLINE\[NEWLINE(-\Delta_{\mathbb H^n})^{\alpha/2}u(g):=\int_{\mathbb H^n}\left(u(gy)-u(g)-\chi_{B_1}(y)<\nabla_{\mathbb H^n}u(g),y>\right)\widetilde{R}_{-\alpha}(y)dy,NEWLINE\]NEWLINE NEWLINEwhere \(\chi_{B_1}\) is the characteristic function in the unit ball \(B_1\), \(\nabla_{\mathbb H^n}u:=\sum_{i=1}^{2n}(X_iu)X_i\) is the presentation of the horizontal gradient in the basis \((X_i)_{1\leq i\leq 2n}\) comprised by vector fields on \(\mathbb H^n\), and \(\widetilde{R}_{-\alpha}(\cdot)\) is a smooth function in \(\mathbb H^n\setminus\{0\}\) and defined by NEWLINENEWLINE\[NEWLINE\widetilde{R}_{-\alpha}(g)=\frac{-\alpha}{2\Gamma(-\alpha/2)}\int_0^\infty t^{-\frac{\alpha+2}{2}}h(t,g)dt,NEWLINE\]NEWLINE NEWLINEwhere \(h(t,\cdot)\) stands for the heat kernel associated to \( -\Delta_{\mathbb H^n}+\frac{\partial}{\partial t}\), and \(\Gamma(\cdot)\) is the gamma function.NEWLINENEWLINEThen, they show that if \(u\) satisfies \((*)_{\gamma,\alpha}\), then \(u\) vanishes in the case when \(p\leq \frac{Q+\gamma}{Q-\alpha}\) such that \(Q=2n+2\) represents the homogeneous dimension of \(\mathbb H^n\) (Theorem 1.1) and among the tools used for the proof, the authors suppose that \(u\) is a weak solution and they use several test functions. By using the properties of an explicit kernel, they show the existence of a positive solution whenever \(p>\frac{Q+\gamma}{Q-\alpha}\) (Lemma 2.1). The authors recall that the case of a zero weak solution for \((*)_{0,2}\) has been already studied. They deduce two corollaries, that the statement of Theorem 1.1 remains true, respectively, by substituting the inequality in \((*)_{\gamma,\alpha}\) with the equality (Corollary 1) and by swapping the Heisenberg group with \(\mathbb R^n\) (Corollary 2).