gpSpline: Generate a Basis Matrix for a Gaussian Process
In abnormally-distributed/cvreg: Cross Validation and Robust Estimation Utilities

The function has a usage similar to bs and ns in the splines package. However, this function utilizes smooth constructor in the package mgcv to construct a gaussian process basis matrix.

gpSpline(
  x,
  df = 5,
  knots = NULL,
  intercept = FALSE,
  fx = FALSE,
  S = NULL,
  m = c(3, -1, 1.4)
)

`x`	the predictor variable. Missing values are allowed.
`df`	degrees of freedom of the basis matrix. The default is 5. The minimum allowed is 3.
`knots`	breakpoints that define the spline. by default these are automatically selected, and not defined by the user.
`intercept`	If TRUE, an intercept is included in the basis matrix.
`fx`	If TRUE, it removes the penalization.
`S`	penalty matrix, defined internally if NULL (the default).
`m`	a numeric vector. selects the covariance function, sets the scale parameter and, if applicable, shape parameter. m[1] between 1 and 5 selects the correlation function from respectively, spherical, power exponential, and Matern with kappa = 1.5, 2.5 or 3.5. m[2] if present specifies the scale parameter, with non-positive or absent indicating that the Kammann and Wand estimate should be used. m[3] can be used to specify the shape parameter for the power exponential function. See `smooth.construct.gp.smooth.spec` for more details. the default option here is c(3,-1,1.4) which indicates the power exponential covariance function with the Kammann and Wand scale estimate, and a shape parameter of 1.4.