Variable exponent Campanato spaces (Q549422): Difference between revisions

Let \((X,d,\mu)\) be a quasimetric measure space with distance function \(d\) satisfying the quasitriangle inequality \(d(x,y)<{\mathcal D}(d(x,z)+ d(z,y))\), where \({\mathcal D}\geq 1\), \(x,y,z\in X\), with ball \(B(x,r)= \{y\in X: d(x,y)< r\}\), \(x\in X\), \(r> 0\). If \(p\) is a measurable function on \(X\) with \(p_-=\text{ess\,inf}\{p(x)\}\), \(p_+= \text{ess\,sup}\{p(x)\}\), such that \(1\leq p_-\leq p(.)\leq p_+< \infty\), let \(I^{p(.)}(f/\lambda)\) and \(\| f\|_{p(.)}\) be defined for \(\lambda> 0\) and measurable functions \(f\) by \[ I^{p(.)}(f/\lambda)= \int_X|f(y)/\lambda|^{p(y)}\,d\mu(y), \] \(\| f\|_{p(.)}= \text{inf}\{\lambda> 0: I^{p(.)} (f/\lambda)\leq 1\}\). Then the variable exponent space \(L^{p(.)}(X)\) is defined to be \(\{f: I^{p(.)}(f/\lambda)< \infty\) for some \(\lambda> 0\}\). If \(\lambda(x)\), \(0\leq\lambda(x)\leq 1\), \(x\in X\), is a \(\mu\)-measurable function, then the variable exponent Morrey space \(L^{p(.),\lambda(.)}(X)\) is defined by \(L^{p(.),\lambda(.)}(X)= \{f: I^{p(.),\lambda(.)}(f)< \infty\}\), where \[ I^{p(.),\lambda(.)}(f)= \sup\Biggl\{\mu(B(x, r))^{-\lambda(x)} \int_{B(x,r)} |f(y)|^{p(y)} d\mu(y): x\in X,\, r> 0\Biggr\}, \] the variable exponent Campanato space \({\mathcal L}^{p(.),\lambda(.)}(X)\) is defined by \(\{f:{\mathcal T}^{p(.),\lambda(.)} (f)<\infty\}\), where \[ {\mathcal T}^{p(.),\lambda(.)}(f)= \sup\Biggl\{\mu(B(x, r))^{-\lambda(x)} \int_{B(x,r)} |f(y)- f_{B(x,r)}|^{p(y)}\,d\mu(y): x\in X,\;r> 0\Biggr\}, \] \[ f_{B(x,r)}= \mu(B(x,r))^{-1} \int_{B(x,r)} f(y)\,d\mu(y), \] and the variable exponent Hölder space \(H^{\alpha(.)}(X)\) is defined by \(H^{\alpha(.)}(X)= \{f:|f|_{\alpha(.)}\leq C\}\) for some finite constant \(C\), where \[ |f|_{\alpha(.)}= \sup\{d(x,y)^{-\max\{\alpha(x), \alpha(y)\}}|f(x)- f(y)|: x,y\in X\}. \] The main results of this paper include the statements: (1) if \(p(.)\) satisfies a logarithmic condition of the form \((-\log d(x,y))|p(x)-p(y)|\leq C_p\), \(x,y\in X\), for some finite constant \(C_p\), \((X,d,\mu)\) is of `homogeneous type', \(\lambda(x)\) is nonnegative real-valued and \(\text{ess\,sup}\{\lambda(x)\}< 1\), then \({\mathcal L}^{p(.),\lambda(.)}(X)\) and \(L^{p(.),\lambda(.)}(X)\) are equivalent with respect to norms defined on the spaces; (2) if \(\text{inf}\{\lambda(x)\}> 1\), then \({\mathcal L}^{p(.),\lambda(.)}(X)\) and \(H^{\alpha(.)}(X)\) are equivalent if \(c_0 r^\gamma\leq \mu(B(x,r))\leq c_1 r^\gamma\), \(\alpha(x)=\gamma p(x)^{-1}(\lambda(x)- 1)\).