Pseudodifferential operators on spaces of distributions associated with non-negative self-adjoint operators (Q2397687)

The purpose of the article is to provide sufficient conditions for having the boundedness of suitable pseudodifferential operators. Accordingly, the authors consider \(L\), a positive self-adjoint operator on \(L_2(M,d\mu)\) where \((M,\rho,\mu)\) is a metric measure space such that \(\rho(\cdot,\cdot)\) and \(\mu\) are the distance function and the positive Radon measure, respectively. Let \(D(L)\) be the domain of \(L\) and \(\displaystyle\mathcal{D}=\bigcap_{m\geq0}D(L^m)\) such that \(\mathcal{D}'\) is the set of continuous linear maps on \(\mathcal{D}\), that is, the set of distributions on \(M\). The authors need to recall the definition of the classical Triebel-Lizorkin spaces, denoted by \(F_{pq}^s(L)\), and are defined as the set of \(f\in\mathcal{D}'\) such that \(\displaystyle||f||_{F^s_{pq}(L)}=\left\|\left(\sum_{j\geq 0}\left(2^{sj}\left|\varphi\left(2^{-j}\sqrt{L}\right)f(\cdot)\right|\right)^q\right)^{1/q}\right\|_{L_p}<\infty\) such that \(s\in\mathbb R\), \((p,q)\in (0,\infty)^2\), and \(\varphi\) is some smooth function with compact support in \(\mathbb R_+\). Also, they define the nonclassical Triebel-Lizorkin spaces, denoted by \(\widetilde{F^s_{pq}}(L)\), endowed with the finite norm \(||f||_{\widetilde{F^s_{pq}}(L)}=\left\|\displaystyle\left(\sum_{j\geq 0}\left(|\mathcal{B}(\cdot,2^{-j})|^{-s/d}\left|\varphi\left(2^{-j}\sqrt{L}\right)f(\cdot)\right|\right)^q\right)^{1/q}\right\|_{L_p}\) such that \(|\mathcal{B}(\cdot,2^{-j})|\) is the volume of the open ball on \(M\) and of radius \(2^{-j}\). As a consequence, the authors deduce that \(\mathcal{D}\) is dense in these (non)classical spaces. Next, for \(x\in M\), the authors define \(P(x,\sqrt{L})\), the pseudodifferential operator which they use for their principal results and it is defined as follows. First, they protract \(S^m_{\mathbb R^d}\), the Hörmander class of symbols on \(\mathbb R^d\) for \(m\in\mathbb R\), to the symbol class \(S^m(L)\) comprised by functions \(\sigma (\cdot,\cdot)\) on \(M\times (0,\infty)\) such that their derivatives with respect to the second variable satisfy a few conditions. Then, the authors provide the definition of an elementary symbol in \(S^m(L)\) and it is given by \(\sigma_l(x,\xi)=\displaystyle\sum_{j\geq 0}\gamma_{j,l}(x)\psi_{j,l}(\xi)\) where \((\gamma_{j,l}(\cdot))_{j\in\mathbb N}\) is a sequence of bounded functions on \(M\) for \(l\in\mathbb Z\) and \((\psi_{j,l}(\cdot))_{j\in\mathbb N}\) is constituted by smooth functions supported in a bounded real interval such that the derivatives of \(\psi_0\) are equal to zero. In the fourth section, the authors furnish meaningful properties of these latter especial symbols. Subsequently, \(\sigma_l(x,\xi)\) is the symbol of the pseudodifferential operator \(P_l(x,\sqrt{L})f(x)=\displaystyle\sum_{j\geq 0}\gamma_{jl}(x)\psi_{jl}\left(\sqrt{L}\right)f(x)\) for \(f\in\mathcal{D}\) and by using the fact that any symbol decomposes into a sum of elementary symbols (Lemma 4.2), the authors set up \(P(x,\sqrt{L})f(x)=\displaystyle\sum_{l\in\mathbb Z}P_l(x,\sqrt{L})f(x)\) (Definition 3.7). Consequently, the authors state their main result that \(P(x,\sqrt{L})\) is bounded from \(F_{pq}^{s+m}(L)\) to \(F^s_{pq}(L)\) (resp. from \(\widetilde{F_{pq}^{s+m}}(L)\) to \(\widetilde{F^s_{pq}}(L)\)) whenever its associated symbol belongs to \(S^m(L)\) such that \(s>0\) and \((p,q)\in (1,\infty)^2\) (Theorem 3.8 (resp. 6.5)). The proof is based on using Lemma 4.2 and the fact that each elementary symbol is bounded (Proposition 4.4). At the end, the authors state analogous boundedness results for the case of (non)classical Besov spaces (Theorem 7.1).