Figure 10.

Multi-fidelity modelling of mixed convection: (a) ${\mathbb{L}}_2$ error between the high-fidelity validation data and the computed posterior mean of the NARGP, AR1 and GP models, as the number of high-fidelity training points is increased. (b) Scatter plot of the NARGP, AR1 and GP posterior means versus the high-fidelity validation data, for a training case with n₁=200 low- and n₂=20 high-fidelity observations. (Online version in colour.)