Computing the weighted geometric mean of two large-scale matrices and its inverse times a vector

We investigate different approaches for computing the action of the weighted geometric mean of two large-scale positive definite matrices on a vector. We derive and analyze several algorithms, based on numerical quadrature and on the Krylov subspace, and compare them in terms of convergence speed and execution time. By exploiting an algebraic relation between the weighted geometric mean and its inverse, we show how these methods can be used to efficiently solve large linear systems whose coefficient matrix is a weighted geometric mean. According to our experiments, some of the algorithms proposed in both families are suitable choices for black-box implementations.