doc: gp_dist
Canonical GP distance ![$d^2(u,v) = Var[f(u)-f(v) \mid Xt] = \sigma_t^2(u)+\sigma_t^2(v)-2k(u,v)$](gp_dist_eq09964.png)
Syntax
D2 = gp_dist(Kuv, Ktu, Ktv, dKuu, dKvv, BayesInv) D2 = gp_dist(Kuv, Ktu, Ktv, dKuu, dKvv, BayesInv, Ht, Hu, Hv)
Arguments
- Kuv kernel matrix (nu, nv) between the points of U and V
- Ktu kernel matrix (nt, nu) between the points of Xt and U
- Ktv kernel matrix (nt, nv) between the points of Xt and V
- dKuu vector (nu, 1) of the diagonal kernel between the points of U
- dKvv vector (nv, 1) of the diagonal kernel between the points of V
- BayesInv struct array as returned by gp_inf(Ht, Ktt, Yt, noise)
- Ht matrix (nt, b) basis for the points of Xt
- Hu matrix (nu, b) basis for the points of U
- Hv matrix (nv, b) basis for the points of V
Outputs
- D2 matrix (nu, nv) of squared distance between U and V