cov.SE: Squared exponential covariance function.
In JimSkinner/spca: Structured PCA

Description Usage Arguments Examples

The squared exponential covariance function. This produces a semidefinite covariance matrix, and should only be used when constructing new covariance functions. E.g., the squared exponential plus independant.

1	cov.SE(X, X2, beta, D = NA, ...)

`X`	Matrix of data
`X2`	(optional) second matrix of data; if omitted, X is used.
`beta`	Hyperparameters; beta[1] is the log signal variance, beta[2] is the log length scale.

# Confirm that diagonal of covariance matrix is the specified variance, and
# the off-diagonals are less than this.
grid = matrix(1:10, ncol=1)
beta = rnorm(2)
K    = cov.SE(grid, beta=beta)
stopifnot(all(Matrix::diag(K)==exp(beta[1])))
stopifnot(all(K[upper.tri(K)]<exp(beta[1])))