fit.hyperparams: Fit hyperparameters for a GP using maximum likelihood...
In mattdneal/gaussianProcess: Gaussian Process Fitting and Prediction

Fit hyperparameters for a GP using maximum likelihood optimisation

fit.hyperparams(gp, sigma.n.init = NA, hyper.params.init = NA,
  prior = get.uniform.prior(), random.init.gridsize = 50,
  resample.top.inits = TRUE, resample.from = NULL, num.resamples = NULL,
  additional.optimx.runs = 1, additional.par.perc.diff = 0.1,
  random.init.scale = 1, abs.min.sigma.n = 0.001,
  abs.min.hyper.params = 0, optimx.method = "L-BFGS-B",
  max.iterations = 10000, optimx.starttests = FALSE, optimx.trace = 0,
  verbose = FALSE)

`gp`	a GaussianProcess object
`sigma.n.init`	an initial value for sigma.n, or NA for random initialisation
`hyper.params.init`	a vector of initial values for the kernel hyperparameters, or NA for random initialisation. Partial random starts can be specified by setting only some entries in the vector to be NA, or a complete random start can be specified by passing a single NA value.
`random.init.gridsize`	the number of random initialisation points to test
`resample.top.inits`	boolean indicating whether to do a second random search based on the (smoothed) distribution of the top results in the first pass
`resample.from`	number of points to build the distribution from in the second random search
`num.resamples`	number of samples for the second search
`additional.optimx.runs`	the number of additional start points to attempt optimisation from
`additional.par.perc.diff`	the percentage difference required between any two starting parameter vectors
`random.init.scale`	a scaling factor for the random generation of initialisation points
`abs.min.sigma.n`	minimum value of sigma.n
`abs.min.hyper.params`	a vector of minimum values of the hyper params
`optimx.method`	the optimx method to use for optimisation
`max.iterations`	max iterations of each optimx run
`optimx.starttests`	run optimx start tests
`optimx.trace`	optimx trace level
`verbose`	verbose output to screen