Frames and finite-rank integral representations of positive operator-valued measures (Q2306740)

The positive operator-valued measure (POVM) defined by a discrete or continuous frame has, by construction, an integral representation by rank-one operators. However, not every POVM has an integral representation. Given a POVM \(F:\mathcal{B}\to B(\mathcal{H})\), where \(\mathcal{B}\) is a \(\sigma\)-algebra on a set \(\Omega\) and \(B(\mathcal{H})\) is the space of bounded operators on a Hilbert space \(\mathcal{H}\), the authors gave a necessary and sufficient condition under which \(F\) has a finite rank integral representation \[ F(E)=\sum_{k=1}^m \int_E G_k(\omega) \otimes G_k (\omega)\; d\mu(\omega) \] for some weakly measurable maps \(G_k\) (\(1\leq k\leq m\)) and some positive measure \(\mu\) on \((\Omega,\mathcal{B})\). The authors gave also a similar characterization for projection-valued measures and settled negatively a problem of Ehler and Okoudjou about probability frame representation of probability POVMs [\textit{M. Ehler} and \textit{K. A. Okoudjou}, in: Finite frames. Theory and applications. Boston, MA: Birkhäuser. 415--436 (2013; Zbl 1262.42013)]. Finally, they proved that an integral representable probability POVM can be dilated to a integral representable projection-valued measure if and only if the corresponding measure is purely atomic.