“…Probabilistic numerical methods have been developed for a series of diverse numerical tasks, including the numerical solution of linear systems, or the computation of integrals. Most notably, a series of works focused on the probabilistic quantification of approximation errors in the solution of ordinary [10,15,24,25,28,29,31,[46][47][48][51][52][53] and partial differential equations [10,12,13,15,20,35,[37][38][39]44,45]. One of the advantages of probabilistic numerical methods is that they allow "propagating uncertainty in computational pipelines" [36], where uncertainty is due to numerical discretization.…”