\(Tb\) theorem for Triebel-Lizorkin spaces over spaces of homogeneous type and their applications (Q848656)

Let \(d>0\) and \(0<\theta\leq1\). A space of homogeneous type, \(({\mathcal X}, \rho, \mu)_{d,\theta}\), is a set \({\mathcal X}\) together with a quasi-metric \(\rho\) and a nonnegative Borel measure \(\mu\) on \({\mathcal X}\) with \(\operatorname{supp}\mu={\mathcal X}\), and there exists a positive constant \(C\) such that for all \(r\in(0,\text{diam}{\mathcal X})\) and all \(x, x', y\in{\mathcal X}\), \(\mu(B(x,r))\sim r^d\) and \[ |\rho(x,y)-\rho(x',y)|\leq C\rho(x,x')^\theta[\rho(x,y)+\rho(x',y)] ^{1-\theta}. \] A complex-valued bounded function \(b\) defined on \({\mathcal X}\) is said to be a para-accretive function if there exists a positive constant \(C\) and \(\kappa\in(0,1]\) such that for all balls \(B\subset{\mathcal X}\), there is a ball \(B'\subset B\) with \(\kappa\mu(B)\leq\mu(B')\) satisfying \(\frac{1}{\mu(B)}|\int_{B'}b(x)\,d\mu(x)|\geq C>0\). For \(\eta\in(0,\theta]\), denote by \(C_0^\eta({\mathcal X})\) the set of all functions with compact support such that \(\|f\|_{C_0^\eta({\mathcal X})}\equiv\sup_{x\neq y} \frac{|f(x)-f(y)|}{|\rho(x,y)^\eta|}<\infty\). Endow \(C_0^\eta({\mathcal X})\) with the natural topology and let \((C_0^\eta({\mathcal X}))'\) be its dual space. Moreover, if \(b\) is a para-accretive function, \(M_b\) denotes the corresponding multiplication operator and \(bC_0^\eta({\mathcal X})\) denotes the image of \(C_0^\eta({\mathcal X})\) under \(M_b\) with the natural topology. Denote by \(\dot{F}_p^{\alpha,q}({\mathcal X})\) the homogeneous Triebel-Lizorkin spaces, where \(\varepsilon\in (0,\theta]\), \[ \max\left\{\frac{d}{d+\alpha},\frac{d}{d+\alpha+\varepsilon} \right\}<p<\infty, \quad \max\left\{\frac{d}{d+\alpha},\frac{d}{d+\alpha+ \varepsilon}\right\}<q\leq\infty, \] and \(|\alpha|\in(0,\varepsilon)\). Let \(b_1\) and \(b_2\) be two para-accretive functions and \(T\) be a continuous linear mapping from \(b_1C_0^\eta{\mathcal X})\) to \((b_2C_0^\eta({\mathcal X}))^{'}\) associated to a kernel \(K(x,y)\) in the sense that \[ \langle Tf,g\rangle\equiv\int_{\mathcal X}\int_{\mathcal X}g(x) b_2(x)K(x,y)b_1(y)f(y)\,d\mu(x)\,d\mu(y) \] for all \(f\) and \(g\) with disjoint supports. In this paper, the author shows that if \(T^\ast(b)=0\), \(|\langle Tf,g\rangle|\leq Cr^{d+2\eta}\|f\|_{C_0^\eta{\mathcal X})}\|g\|_{C_0^\eta({\mathcal X})},\) \(|K(x,y)|\leq C\rho(x,y)^{-d}\) and \[ |K(x,y)-K(x,y')|\leq C\rho(y,y')^\varepsilon\rho(x,y)^{-d-\varepsilon} \quad\text{for}\quad\rho(y,y')\leq\rho(x,y)/(2A), \tag \(*\) \] where \(T^{\ast}\) denotes the dual operator of \(T\) and \(A\) the quasi-metric constant of \(\rho\), then \(T\) is bounded from \(b\dot{F}_p^{\alpha,q}({\mathcal X})\) to \(b^{-1}\dot{F}_p^{\alpha,q}({\mathcal X})\) for \(-\varepsilon<\alpha<0\), \(\frac{d}{d+\alpha+ \varepsilon}<p<\infty\), \(\frac{d}{d+\alpha+\varepsilon}<q\leq\infty\). Furthermore, the author proves that when \(0<\alpha<\varepsilon\), if the assumptions \(T^{\ast}(b)=0\) and \((\ast)\) are replaced, respectively, by the assumptions \(T(b)=0\) and \[ |K(x,y)-K(x',y)|\leq C\rho(x,x')^\varepsilon\rho(x,y)^{-d-\varepsilon} \quad\text{for}\quad\rho(x,x')\leq\rho(x,y)/(2A), \] then \(T\) is also bounded from \(b\dot{F}_p^{\alpha,q}(\mathcal {X})\) to \(b^{-1}\dot{F}_p^{\alpha,q}(\mathcal {X})\) for \(\frac{d} {d+\alpha}<p<\infty\), \(\frac{d}{d+\alpha}<q\leq\infty\). As an application, new characterizations of Triebel-Lizorkin spaces with minimum regularity and cancellation conditions are given.