Dual fusion frames (Q477318)

The authors study fusion frames in a Hilbert space \(\mathcal H\), i.e., sets \((\mathcal W, w)=\{(W_i,w_i)\}_{i\in I}\) where each \(W_i\) is a closed subspace of \(\mathcal H\), \(w_i>0\) and \[ \alpha \|f\|^2 \leq \sum_{i\in I} w_i^2\| \pi_{W_i}(f)\|^2 \leq \beta \|f\}^2, \] for all \(f\in \mathcal H\), where \(0< \alpha \leq \beta <\infty\) and \(\pi_{W_i}\) is the orthogonal projection onto \(W_i\). The synthesis and analysis operators are defined by \(T_{\mathcal W, w}(\{f_i\}_{i\in I})= \sum_{i\in I} w_if_i\) when \(f_i\in W_i\) and \(\{\|f_i\|\}\in \ell^2(I)\) and \(T^*_{\mathcal W, w}(f)= \{w_i\pi_{W_i}(f)\}_{i\in I}\). Now they define a dual fusion frame of \((\mathcal W, w)\) to be a fusion frame \((\mathcal V, v)\) such that there is a bounded linear operator \(Q\) such that \(T_{\mathcal V, v}QT^*_{\mathcal W, w}= I_{\mathcal H}\). Then they give a number of results and examples that show that this definition extends the commonly used notion, overcomes some technical problems with previous definitions, and that with this definition dual fusion frames in many cases are quite similar to dual frames.