Table 1.

Comparison of average training and validation MSE losses for training and validation data on the domain boundary as given in equation (14) with σ = 0.1.

kernel	MSE loss on training data	MSE loss on validation data
RBF	0.007 62	0.0110
Matèrn $(\nu =0.1)$	$3.2\times {10}^{-8}$	0.0465
Matèrn $(\nu =1.5)$	0.005 98	0.0116
Matèrn $(\nu =4.0)$	0.005 88	0.0126
Rational Quadratic	0.007 21	0.0112