On the Bures-Wasserstein distance between positive definite matrices (Q2323328): Difference between revisions

The authors define \(d( A, B)\) to be a special metric on the space of positive semidefinite matrices. This metric is of interest in quantum information theory, where is known as Bures distance, and in differential geometry, in statistics and in the theory of optimal transport, where is known as Wasserstein metric. In particular, if \(A\) and \(B\) are diagonal matrices, then \(d(A, B)\) reduces to the Hellinger distance between probability distributions and is related to the Rao-Fisher metric in information theory. The authors explore some fundamental properties of this metric from the perspective of matrix analysis. This allows them to unify several known facts, to simplify some proofs and to point out new connections. This is a beautiful and interesting paper. It contains useful and rich interdisciplinary work.