Domains of pseudo-differential operators: a case for the Triebel-Lizorkin spaces (Q2574414): Difference between revisions

A pseudo-differential operator from \(C_{0}^{\infty}(\mathbb R^n)\) into \({\mathcal D}^\prime(\mathbb R^n)\) may be viewed as a linear map of the form \[ OP(a)\;u(x)= {(2\pi)}^{-n}\int e^{ix\cdot\xi} a(x,\xi)\hat u(\xi)\,d\xi, \] where \(\hat u\) denotes the Fourier transform of \(u\). Such operators began to be systematically studied with [cf.\ \textit{J.~J.\ Kohn} and \textit{L.~Nirenberg}, Commun.\ Pure Appl.\ Math.\ 18, 269--305 (1965; Zbl 0171.35101)], and are well-defined and continuous for more general space settings (under a variety of assumptions on the Fourier symbol \(a\)). A fundamental tool in view of the continuity of \(OP(a)\) (in several space settings) turns out to be the \(S_{\rho,\delta}^s\) class introduced by \textit{L.~Hörmander} [Proc.\ Sympos.\ Pure Math.\ 10, 138--183 (1967; Zbl 0167.09603)]. The class \(S_{\rho,\delta}^s\) consists of \(C^\infty\) functions \(a\) defined by restricting the behavior of \(a\) and its derivatives, according to the estimate: \[ \left| D^\alpha_x D^\beta_\xi a(x,\xi)\right| \leq c_{\alpha\beta}(1+| \xi| )^{s- \rho| \alpha| + \delta| \beta| }, \] with \(x\in \mathbb R^n\), \(\xi\in \mathbb R^n\), and for all multi-indices \(\alpha\) and \(\beta\). The main result of the present paper shows that if \(a\in S_{1,1}^s\) (\(s\in \mathbb R\)), then \(OP(a)\) is a bounded operator in the following Triebel-Lizorkin and Besov space settings \[ OP(a) : F_{p,1}^{s}(\mathbb R^n) \to L_p(\mathbb R^n) ,\qquad p\in [1, \infty[ , \] \[ OP(a) : B_{\infty,1}^{s}(\mathbb R^n) \to L_\infty(\mathbb R^n), \] respectively. Additionally, it is exhibited that this result is optimal in the sense that for some \(a\in S_{1,1}^s\) there are corresponding unbounded operators \[ OP(a) : {\mathcal S}(\mathbb R^n) \to {\mathcal D}^\prime(\mathbb R^n) \] with \({\mathcal S}(\mathbb R^n)\) endowed with the induced topology from any of the Triebel-Lizorkin spaces \(F_{p,q}^{s}(\mathbb R^n)\) or Besov \(B_{p,q}^{s}(\mathbb R^n)\) spaces, for \(p\in [1, \infty]\) and \(q\in ]1, \infty]\) (while \({\mathcal D}^\prime(\mathbb R^n)\) has the usual topology). A short version of the present work (including that main result) had been published in [\textit{J.~Johnsen}, C.\ R.\ Math.\ Acad.\ Sci., Paris 339, No.~2, 115--118 (2004; Zbl 1045.47039)].