Two weight commutators in the Dirichlet and Neumann Laplacian settings (Q2422459): Difference between revisions

In this paper the authors establish the characterization of the weighted BMO via two weight commutator theorem in the settings of the Neumann Laplacian $\triangle_{N_{+}}$ with the $i$-th Riesz transform $R_{N_{+},i}=\frac{\partial}{\partial x_{i}}\triangle_{N_{+}}^{-\frac{1}{2}}$ on the upper half space $\mathbb{R}^n_{+}$ and the reflection Neumann Laplacian $\triangle_{N}$ with the Riesz transform $R_{N,i}=\frac{\partial}{\partial x_{i}}\triangle_{N}^{-\frac{1}{2}}$ on $\mathbb{R}^n$. We denote by $\mathrm{BMO}_{\triangle_{N_{+}},\nu}(\mathbb{R}^n_{+})$ and $\mathrm{BMO}_{\triangle_{N},\nu}(\mathbb{R}^n)$ the weighted BMO spaces associated with $\triangle_{N_{+}}$ and $\triangle_{N}$ respectively. \par Theorem 1. Suppose $1 < p< \infty$ and weights $\mu, \lambda$ in the Muckenhoupt class $A^p(\mathbb{R}^n_{+})$ and set $\nu = \mu^{\frac{1}{p}}\lambda^{-\frac{1}{p}}$. Then there are constants $0< c < C < \infty$\ such that \[ c||b||_{\mathrm{BMO}_{\triangle_{N_{+}},\nu}(\mathbb{R}^n_{+})} \leq ||[b,R_{N_{+},i}]: L^p_{\mu}(\mathbb{R}^n_{+}) \rightarrow L_{\lambda}^p(\mathbb{R}^n_{+})|| \leq C||b||_{\mathrm{BMO}_{\triangle_{N_{+}},\nu}(\mathbb{R}^n_{+})} \] in which $[b, R_{i}]$ denotes the commutator of the Riesz transform $R_{i}$ and the function $b \in \mathrm{BMO}_{\triangle_{N_{+}},\nu}$ \par This two weight commutator theorem might not be true in the setting of a general operator $L$, and in particular, the authors show that it is not true when $L$ is the Dirichlet Laplacian on $\mathbb{R}^n_{+}$. \par The authors introduce the class of weights $A^p_{\triangle_{N}}(\mathbb{R}^n)$ associated with $\triangle_{N}$, which is larger than the Muckenhoupt weighted class and contains non-doubling weights. \par Theorem 2. Suppose $1 < p< \infty$ and weights $\mu, \lambda \in A^p_{\triangle_{N}}(\mathbb{R}^n)$ and set $\nu = \mu^{\frac{1}{p}}\lambda^{-\frac{1}{p}}$. Then there are constants $0< c < C < \infty$\ such that \[ c||b||_{\mathrm{BMO}_{\triangle_{N},\nu}(\mathbb{R}^n)} \leq ||[b,R_{N,i}]: L^p_{\mu}(\mathbb{R}^n) \rightarrow L_{\lambda}^p(\mathbb{R}^n)|| \leq C||b||_{\mathrm{BMO}_{\triangle_{N},\nu}(\mathbb{R}^n)}. \] The authors introduce the weighted Hardy space $H^1_{\triangle_{N}, \nu}(\mathbb{R}^n)$ associated with $\triangle_{N}$ and prove that it is the predual of $\mathrm{BMO}_{\triangle_{N},\nu}(\mathbb{R}^n)$. This yields a weak factorization for $H^1_{\triangle_{N}, \nu}(\mathbb{R}^n)$.