A note on computational uncertainty

摘要

Computational uncertainties inevitably arise due to equivalence representation of a continuum phenomenon by a discrete model, and the subsequent numerical solution of the model which gives rise to a loss of accuracy. In this note, we consider the connection between convergence and uncertainty and identify the functional quantification and modelling that are needed to investigate the propagation of uncertainties due to the numerical scheme, the grid and the adjacent grid points, into the uncertainty in final computed solution when an initial boundary value problem is considered.