Multivariate matrix refinable functions with arbitrary matrix dilation (Q4243608)

The author studies properties of multivariate vectors \(\Phi =(\phi_1,\ldots , \phi_r)\) of \(L^2({\mathbb R}^d)\)-functions satisfying a matrix refinement equation \[ \Phi =\sum_{\alpha \in Z^d}P_{\alpha}\Phi (M\cdot - \alpha), \] where \(M\) is a \(d\times d\) integer matrix with \(\lim_{n\rightarrow\infty} M^{-n}=0\) and \(\{P_\alpha \}\) is a finitely supported \(r\times r\) matrix sequence. Consider the corresponding mask symbol \(P(\omega)=|\det M|^{-1}\sum_{\alpha \in\;Z^d}P_{\alpha} e^{-i\alpha \omega }\) and the transition operator \(T\) associated with \(P\), acting on the space of all \(r\times r\) matrices with trigonometric polynomials as entries, \[ TC(w):=\sum_{j=0}^{|\det M|- 1}P(M^{-T}(w+2\pi\eta_j)) C(M^{-T}(w+2\pi\eta_j)) P^{\ast} (M^{-T}(w+2\pi\eta_j)) , \] where \(\eta_j\) \((j=0,\ldots , |\det M|-1)\) are distinct elements of \({\mathbb Z}^d/ M^T {\mathbb Z}^d\). Stability, orthonormality and approximation properties are characterized in terms of spectral properties of \(P\) and \(T\). The theory is illustrated with a bivariate example.