No label defined (Q4462269)

For \(f\in L^1(\mathbb R^n)\), the following notation is used: \[ \| f\| :=\int_{\mathbb R^n}| f(x)| \,dx,\quad \overline{\Delta}_hf(x)=f(x+h/2)-f(x-h/2),\quad \overline{\Delta}^N_h= \overline{\Delta}_h\overline{\Delta}_h^{N-1}. \] Now, given the Bochner-Riesz means \[ B^{\lambda}_{t}(f)(x):= (2\pi)^{-n}\int_{\mathbb R^n}\left(1-t^2| \xi| ^2\right)^{\lambda}_+\widehat{f}(\xi)e^{-ix\xi}\,dx,\quad \lambda>(n-1)/2, \] where it is used as Fourier transform \(F[f](\xi)=\widehat{f}(\xi):=\int _{\mathbb R^n}f(x)e^{-i\xi x}\,dx\), and given the K-functional \[ K\left(t,f;L^1,D(\Delta)\right):=\inf _{g\in D(\Delta)}(\| f-g\| +t\| \Delta g\| ),\quad t>0, \] \(D(\Delta )\) being the domain of the Laplacian \(\Delta \) in \(L^1(\mathbb R^n)\), it was proved by \textit{W. Trebel} [Approx. Theory, Proc. int. Colloq., Bonn 1976, Lect. Notes Math. 556, 428--438 (1976; Zbl 0358.41009)] that \[ K\left(t^2,f;L^1,D(\Delta)\right)\approx \| B^{\lambda}_{t}(f)-f\| ,\quad t>0,\;f\in L^1(\mathbb R^n), \] and also that \[ \widetilde{w}_2(t,f):=\sup_{0<\delta <t}\left\| \sum_{j=1}^{n}\overline{\Delta}_{\delta e_j}^2f\right\| \leq CK\left(t^2,f;L^1,D(\Delta)\right), \] (\(e_j\in \mathbb R^n\) is the \(j\)-th unit vector). In this paper, the authors prove the converse to be true through a straight-forward purely Fourier analytic proof, in contrast with the one stated by \textit{Z. Ditzian} [Trans. Am. Math. Soc. 326, No. 1, 407--422 (1991; Zbl 0737.46017)] with the aid of somewhat combinatorial arguments. On the other hand, also in the paper by Trebels, the following modification of the K-functional above \[ K(t,f;L^1,D((-\Delta)^{\alpha /2})):=\inf _{g\in D}(\| f-g\| +t\| F^{-1}[| \xi| ^{\alpha }\widehat{g}(\xi)]\| ),\quad \alpha >0,\;t>0, \] was characterized via the Riesz means as follows: \[ K(t^{\alpha },f;L^1,D((-\Delta)^{\alpha /2}))\approx \| R_{t}^{\lambda ,\alpha }(f)-f\| , \] where \(\lambda> (n-1)/2\), \(\alpha >0\), \(F[R_t^{\lambda,\alpha}(f)](\xi)=(1-t^{\alpha}| \xi| ^{\alpha})^{\lambda}_+ \widehat{f}(\xi)\). In connection with it, the authors also present the following Theorem. Let \(k\in L^1(\mathbb R^n)\) be such that the following conditions are satisfied: (a) \(1-\widehat{k}(\xi)\neq 0\) for \(\xi \neq 0\) ; (b) for some \(\alpha >0\) and some \(\mu_{\alpha }\in M(\mathbb R^n)\) with \(\int_{\mathbb R^n}d\mu_{\alpha}\neq 0\) there holds \[ | \xi| ^{-\alpha}(1-\widehat{k}(\xi))=[d\mu_{\alpha }]^{\widehat{\;}}(\xi)\;\forall \xi\in \mathbb R^n. \] Then we have, for all \(f\in L^1(\mathbb R^n)\), \[ K(t^{\alpha },f;L^1,D((-\Delta)^{\alpha /2}))\approx \| k_t\ast f-f\| ,\quad t>0 \] where \(k_t:=t^{-n}k(\cdot /t)\). For the entire collection see [Zbl 1013.00026].