Bayesian Analysis, Vol.14, N°3, 2019
We would like to congratulate with the Authors for this interesting development of
probabilistic numerical methods applied to the ubiquitous problem of solving linear
systems. We structured this discussion around two main points, namely the use of
Bayesian Conjugate Gradient (bcg) for Gaussian processes (gps), and the possibility
to accelerate the solution of linear systems thanks to parallelization of bcg.