Isotropic and dominating mixed Lizorkin-Triebel spaces -- a comparison (Q682123)

The Sobolev space of dominating mixed smoothness \(S^t_pW(\mathbb R^d)\) is defined as the set of functions \(f\in L_p(\mathbb R^N)\) with finite norm \(\| f| S^t_pW(\mathbb R^d)\| =\sum_{| \bar\alpha|_\infty\leq t}\| D^{\bar\alpha}f| L_p(\mathbb R^d)\| \), where \(t\in\mathbb N_0\), \(1<p<\infty\), \(\bar\alpha=(\alpha_1,\dots,\alpha_d)\in\mathbb N^d_0\) and \(| \bar\alpha|_\infty=\max_{1\leq i\leq d}\alpha_i\). The space \(S^t_pW(\mathbb R^d)\) differs from the standard isotropic Sobolev space \(W^t_p(\mathbb R^d)\) by the \(\ell_\infty\)-norm taken for the multiindices in the sum instead of the standard \(\ell_1\)-norm \(| \bar\alpha|_1=\alpha_1+\dots+\alpha_d\). The spaces \(W^t_p\) and \(S^t_pW\) represent particular cases of corresponding scales of Bessel potential spaces \(H^t_p(\mathbb R^d)\) and \(S^t_pH(\mathbb R^d)\), \(t\in\mathbb R\), \(1<p<\infty\), which are the classes of distributions \(f\in \mathcal S'(\mathbb R^d)\) with the norms \(\| f| H^t_p(\mathbb R^d)\| =\| \mathcal F^{-1}[(1+| \xi| ^2)^{t/2}\mathcal F f](\cdot)| L_p(\mathbb R^d)\| \) and \(\| f| S^t_pH(\mathbb R^d)\| =\| \mathcal F^{-1}[\Pi_{i=1}^d(1+\xi_i^2)^{t/2}\mathcal F f](\cdot)| L_p(\mathbb R^d)\| \), respectively, where \(\mathcal F\) denotes the Fourier transform, and \(\xi\in\mathbb R^d\). It is known [\textit{H.-J. Schmeisser}, in: Nonlinear analysis, function spaces and applications. Vol. 8. Proceedings of the spring school, NAFSA 8, Prague, Czech Republic, May 30--June 6, 2006. Prague: Czech Academy of Sciences, Mathematical Institute. 145--204 (2007; Zbl 1289.46056)] that the chain of embeddings \(H^{td}_p(\mathbb R^d)\hookrightarrow S^t_pH(\mathbb R^d)\hookrightarrow H^t_p(\mathbb R^d)\) holds for \(t>0\). The saces \(H^t_p(\mathbb R^d)\) and \(S^t_pH(\mathbb R^d)\) can be considered as particular cases in the scales of isotropic Lizorkin-Triebel spaces \(F^t_{p,q}(\mathbb R^d)\) and Lizorkin-Triebel spaces of dominated mixed smoothness \(S^t_{p,q}F(\mathbb R^d)\): \(H^t_p=F^t_{p,2}\) and \(S^t_pH=S^t_{p,2}F\) for \(1<p<\infty\), \(t\in\mathbb R\), see [\textit{H. Triebel}, Theory of function spaces. Mathematik und ihre Anwendungen in Physik und Technik, Bd. 38. Leipzig: Akademische Verlagsgesellschaft Geest \& Portig K.-G. (1983; Zbl 0546.46028)] and [\textit{H.-J. Schmeisser} and \textit{H. Triebel}, Topics in Fourier analysis and function spaces. Chichester: John Wiley \& Sons (1987; Zbl 0661.46025)]. The aim of the paper is to describe conditions on \(t\), \(p\) and \(q\) under which the embeddings \(F^{td}_{p,q}(\mathbb R^d)\hookrightarrow S^t_{p,q}F(\mathbb R^d)\hookrightarrow F^t_{p,q}(\mathbb R^d)\) hold and to discuss the optimality of these embeddings in various aspects.