Mapping properties of operator-valued pseudo-differential operators (Q2324743)

The authors investigate pseudo-differential operators \(T^c_\sigma\) given by a symbol \(\sigma\) defined on \(\mathbb{R}^d\times\mathbb{R}^d\) with values in a von Neumann algebra \(\mathcal{M}\) by setting \(T^c_\sigma f(s)=\int_{\mathbb{R}^d}\sigma(s,\xi)\hat f(\xi)e^{2\pi\mathrm{i}s\cdot\xi}\,d\xi\), where \(f\) is defined on \(\mathbb{R}^d\) with values in the space \(L_1(\mathcal{M})+\mathcal{M}\). Note that when \(\mathcal{M}\) is noncommutative, setting \(T^r_\sigma f(s)=\int_{\mathbb{R}^d}\hat f(\xi)\sigma(s,\xi)e^{2\pi\mathrm{i}s\cdot\xi}\,d\xi\) defines a different operator: the superscripts ``\(r\)'' and ``\(c\)'' stand respectively for ``row'' and ``column''. The hypothesis on the symbol is that it belong to the \textit{Hörmander class} \(S^0_{1,\delta}\), i.e., that \[ \|D_s^\gamma D_\xi^\beta\sigma(s,\xi)\|_{\mathcal{M}}\le C_{\gamma,\beta}(1+|\xi|)^{\delta(\gamma_1+\cdots+\gamma_d)-\beta_1-\cdots-\beta_d} \] for all multi-indices \(\gamma,\beta\) of nonnegative integers. Their techniques and results differ according to whether \(0\le\delta<1\) (the case of ``regular'' symbols) or \(\delta=1\) (``forbidden'' symbols). The operators \(T^c_\sigma\) are proved to be bounded on the following three kinds of spaces: (i) the operator-valued Sobolev spaces \(H_2^\alpha(\mathbb{R}^d;L_2(\mathcal{M}))\); (ii) the operator-valued Besov spaces \(B_{p,q}^\alpha(\mathbb{R}^d;L_p(\mathcal{M}))\); (iii) the operator-valued column inhomogeneous Triebel-Lizorkin spaces \(F_{p}^{\alpha,c}(\mathbb{R}^d;\mathcal{M})\). Here, \(\alpha\) is an arbitrary real number if \(\sigma\) is regular and \(\alpha>0\) if \(\sigma\) is forbidden. Case (iii) has been introduced by the authors in [Integral Equations Oper. Theory 90, No. 6, Paper No. 65, 65~p. (2018; Zbl 06965992)]: they use decompositions of \(f\) into smooth atoms obtained there and study the action of \(T^c_\sigma\) on those. At the end of the article, they obtain the same results for tori \(\mathbb{T}^d\) and quantum tori \(\mathbb{T}_\theta^d\) instead of \(\mathbb{R}^d\). They also discuss how their article relates to [\textit{A.~González-Pérez}, \textit{M.~Junge} and \textit{J.~Parcet}, Singular integrals in quantum Euclidean spaces. Mem. Amer. Math. Soc. 272, No.~1334, Providence, RI: American Mathematical Society (2021)].