Defect distributions applied to differential equations with power function type coefficients (Q2193927)

The authors consider the following classes of global symbols on \(\mathbb{R}^{2d}\). The space \(M\Gamma^m_{\rho,\Lambda}\), \(m\in\mathbb{R}\), consists of all \(a\in\mathcal{C}^{\infty}(\mathbb{R}^{2d})\) such that for every \(\alpha_1,\alpha_2\in\mathbb{N}^d\) and \(\gamma_1,\gamma_2\in\{0,1\}^d\), there exists \(C>0\) such that \[ \frac{|x^{\gamma_1}\xi^{\gamma_2}\partial^{\alpha_2+\gamma_2}_{\xi} \partial^{\alpha_1+\gamma_1}_x a(x,\xi)|}{\Lambda(x,\xi)^{m-\rho|\alpha_1+\alpha_2|}} \leq C\quad \mbox{for all}\,\, x,\xi\in\mathbb{R}^d, \] while the space \(\widetilde{M\Gamma^m_{\rho,\Lambda}}\), \(m\in\mathbb{R}\), consists of all \(a\in\mathcal{C}^{\infty}(\mathbb{R}^{2d})\) such that for every \(\alpha_1,\alpha_2\in\mathbb{N}^d\) and \(\gamma\in\{0,1\}^d\), there exists \(C>0\) such that \[ \frac{|\xi^{\gamma}\partial^{\alpha_2+\gamma}_{\xi} \partial^{\alpha_1}_x a(x,\xi)|}{\Lambda(x,\xi)^{m-\rho|\alpha_2|}} \leq C\quad \mbox{for all}\,\, x,\xi\in\mathbb{R}^d. \] Here \(\Lambda\) belongs to a rather general class of weights on \(\mathbb{R}^{2d}\). They also consider the subspaces \((M\Gamma^m_{\rho,\Lambda})_0\subseteq M\Gamma^m_{\rho,\Lambda}\) and \((\widetilde{M\Gamma^m_{\rho,\Lambda}})_0\subseteq \widetilde{M\Gamma^m_{\rho,\Lambda}}\) for whose elements the above quantities vanish as \(|(x,\xi)|\rightarrow \infty\). One can naturally associate Sobolev spaces \(H^{m,p}_{\Lambda}\), \(m\in\mathbb{R}\), \(p\in (1,\infty)\), with the symbol classes \(M\Gamma^m_{\rho,\Lambda}\), \(m\in\mathbb{R}\). In the main results, the authors construct defect distributions (a generalisation of \(H\)-measures) for null sequences in the Sobolev spaces \(H^{m,p}_{\Lambda}\). Namely, given \(u_n\in H^{s,p}_{\Lambda}\), \(n\in\mathbb{N}\), and \(v_n\in H^{-s+m,q}_{\Lambda}\), \(n\in\mathbb{N}\), which converge to \(0\) in \(H^{s,p}_{\Lambda}\) and \(H^{-s+m,q}_{\Lambda}\), respectively (\(p\) and \(q\) are Hölder conjugate indexes), there exists a distribution \(\mu\in (\mathcal{S}(\mathbb{R}^d)\hat{\otimes}(M\Gamma^m_{\rho,\Lambda})_0)'\) such that (up to a subsequence) \[ \lim_{n\rightarrow \infty}\langle u_n, \overline{\overline{a}(x,D)(\varphi v_n)}\rangle = \langle \mu,\overline{\varphi}\otimes a\rangle,\quad \varphi\in\mathcal{S}(\mathbb{R}^d),\, a\in (M\Gamma^m_{\rho,\Lambda})_0; \] the same holds if one replaces \((M\Gamma^m_{\rho,\Lambda})_0\) with \((\widetilde{M\Gamma^m_{\rho,\Lambda}})_0\). At the very end, the authors apply the theory to a class of PDEs.