Application-specific error control for parameter identification problems

Artikel i vetenskaplig tidskrift, 2011

An often occurring scenario in the mathematical modeling of physical phenomena is that of a two-step computation consisting of, first, identifying a relevant parameter set from experiments (such as material parameters) and, second, carrying out a subsequent simulation using these parameters. In order to ensure the quality of the results from the simulation, the different sources of errors, for example, from discretization and inexact solution, must be traced and properly reduced. In this paper, a previously developed method for goal-oriented a posteriori error estimation for identification problems is extended to accommodate the combined identification and subsequent simulation. © 2009 John Wiley & Sons, Ltd.