Norm inequalities with power weights for Hörmander type multipliers (Q1360200)

In this paper multiplier theorems of Hörmander-Mikhlin type are studied. Let \(m(x)\) be a bounded, measurable function on \(\mathbb{R}^n\). The operator \(T_mf\) defined by the Fourier transform equation \((T_mf)^\wedge (x)=m(x) \widehat f(x)\) is called a multiplier operator with multiplier \(m\). Let \(s\) be a number greater than or equal to 1, and \(\alpha= (\alpha_1, \dots, \alpha_n)\) a multi-index of nonnegative integers \(\alpha_j\) with norm \(|\alpha |= \alpha_1+ \cdots+ \alpha_n\). We say \(m\in M(s,\lambda)\) if \[ B(m,s, \lambda)= |m|_\infty+ \sup_{R>0, |\alpha |\leq \lambda} \Bigl(R^{s |\alpha |-n} \int_{ R<|x|<2R} \bigl|D^\alpha m(x) \bigr|^sdx \Bigr)^{1\over s} <\infty, \] where \(\lambda\) is a positive integer. For the case where \(\lambda\) is not an integer, let \(l\) be the integer part of \(\lambda\) and let \(\gamma= \lambda-l\). We say \(m\in M(s,\lambda)\) if \[ B(m, s, \lambda)= B(m,s,l)+ \sup_{R>0,0 <|z|<R/2} I(R,z) <\infty, \] where \[ I(R,z) = \sup_{ |\alpha |=l} \Bigl(\bigl(R/ |z|\bigr)^{\gamma s} R^{s|\alpha|-n} \int_{R< |x|<2R} \bigl|D^\alpha m(x)-D^\alpha m(x-z)\bigr|^s dx\Bigr)^{1\over s}. \] We will denote by \(B_s\) the norm \(B(m,s,\lambda)\). If \(\lambda\) is an integer, then the multipliers belonging to \(M(2,\lambda)\) are the classical Hörmander-Mikhlin multipliers. We denote by \(S_{0,0}\) the space of Schwartz functions whose Fourier transforms have compact support not including the origin. Given a real number \(\sigma\), we define \(|f |_{p,\sigma}= (\int_{R^n} |f(x) |^p |x|^\sigma dx)^{1 \over p}\). The main result of this paper is the following theorem. Theorem A. Assume \(1\leq s\leq 2\), \(n/s <\lambda\), \(m\in M(s,\lambda)\) and \(1<p< \infty\). If \(\sigma\) is a real number that satisfies (i) \(\max (-n,-p \lambda)< \sigma< \min(p\lambda,-n +p(\lambda+ n-n/s))\) and (ii) \(0<n ({\sigma+ n\over np}- l)<1\), where \(l\) is the integer part of \((\sigma+ n)/(np)\), then for each \(f\in S_{0,0}\), \[ |T_m f|_{p, \sigma} \leq CB_s|f|_{p, \sigma}, \] where \(C\) is independent of \(m\) and \(f\). Let \(k\in \mathbb{R}\). We define the symbol class \(S^k_{1,0}\) to consist of the set of \(p\in C^\infty (\mathbb{R}^n \times \mathbb{R}^n)\) with the property that, for any compact \(A\) in \(\mathbb{R}^n\), and multi-indices \(\alpha,\beta\), there exists a constant \(C_{A,\alpha, \beta}\) such that \[ \bigl|D^\beta_x D^\alpha_\xi p(x,\xi) \bigr|\leq C_{A,\alpha,\beta}\bigl(1+|\xi|\bigr)^{k-|\alpha |} \] for all \(x\in A\) and \(\xi\in \mathbb{R}^n\). We may assume, without loss of generality, that \(p\) has compact support in the \(x\) variable. For each symbol \(p\in S^k_{1,0}\), we have an associated operator \(p(x,D)\) defined by \[ p(x,D)f= \int_{\mathbb{R}^n} p(x,\xi) \widehat f(\xi) e^{ix \cdot \xi} d\xi. \] As an application of Theorem A, the following theorem is proved. Theorem B. Let \(\sigma\) be a real number satisfying the hypotheses of Theorem A. Let \(k\leq 0\) and assume that \(1<p< \infty\). If \(p\in S^k_{1,0}\), then \(|p(\cdot, D) f|_{p, \sigma} \leq C|f |_{p, \sigma} \) for \(f\in S_{0,0}\) with \(C\) independent of \(f\).