Model certification problem for processes

Abstract

The certification of a low-fidelity model with respect to a high-fidelity model consists in finding reliable error bounds for a domain of validity. The presented conceptual framework allows the mathematical formulation of the certification problem for time-dependent processes. It is shown that instantaneous error estimates of the right-hand sides provide rough certification bounds. Applications are found in modeling under uncertainty, e. g. in the life sciences as well as in data science approaches. Finally, weakened requirements for model certification are discussed.