Figure 4:

Validation of the temporal response predicted by the surrogates against the corresponding high fidelity simulation. Plots of the growth variable ϑ^g are predicted for 3 points of interest: the point in apex, middle, and the periphery of the expanded area. The test input for the surrogates is the mean of the parameters in Table 1. The output of the corresponding high fidelity simulation is shown for validation. The evaluation of the Gaussian processes (GPs) results in the prediction of a mean value and the corresponding variance, a key feature of Bayesian surrogates. The low fidelity surrogate is completely inaccurate, while the multi-fidelity and high fidelity GPs have good agreement with the true response in all three volumes.