Irregular shearlet frames: Geometry and approximation properties (Q636822)

The authors study representations of the shearlet group \(\mathbb S= \mathbb R^+\times \mathbb R\times \mathbb R^2\) given by \((\sigma(a,s,t)\psi)(x)= a^{3/4} \psi(S_sA_ax-t)\), where \(A_a=\left(\begin{smallmatrix} a & 0\\ 0 & \sqrt a\end{smallmatrix}\right)\), \(S_s=\left(\begin{smallmatrix} 1 & s\\ 0 & 1\end{smallmatrix}\right)\), and \(t\in \mathbb R^2\). For each \(\psi \in L^2(\mathbb R^2)\), they define a continuous weighted shearlet system by \({\mathcal {SH}}(\psi,w)= \{w(a,s,t)^{1/2}\sigma(a,s,t)\psi: (a,s,t)\in \mathbb S\}\) and their main concern is to obtain a deep geometric understanding of the discretization process defining the shearlet system by \({\mathcal {SH}}(\psi,\Lambda,w)= \{w(a,s,t)^{1/2}\sigma(a,s,t)\psi: (a,s,t)\in \Lambda\}\) where \(\Lambda\) is a discrete subset of \(\mathbb S\). In particular, they study irregular systems where \(\Lambda\) is not of the form \(\{(a^j,bk,cm): j,k\in \mathbb Z\), \(m\in \mathbb Z^2\}\). They introduce and analyze an upper and lower weighted shearlet density and use this to prove necessary and sufficient conditions for the shearlet system to form a frame in \(L^2(\mathbb R^2)\). Then they analyze the approximation properties of the shearlet systems, in particular the homogeneous approximation abilities of shearlet frames. Examples such as oversampled shearlet systems and co-shearlet systems are presented as well.