Fig. 1
From: A posteriori error estimation for model order reduction of parametric systems
The concept of multi-fidelity error estimation in a greedy process, where \(\Delta (\mu )\) represents any original error estimator, \(\Delta ^\ell (\mu )\) is a low-fidelity error estimator when we stop updating partial information of \(\Delta (\mu )\), and \(\Delta _s(\mu )\) is a surrogate of \(\Delta (\mu )\). Likewise, \(\Delta ^\ell _s(\mu )\) is a surrogate of \(\Delta ^\ell (\mu )\). tol and \(\varepsilon \) are defined in Algorithm 1, tol \(<\theta <1\) is a user-defined small value