optimMethods: Optimization Methods (or Algorithms) for the 'mle' Method

In kergp: Gaussian Process Laboratory

Optimization Methods (or Algorithms) for the mle Method

Description

Optimization methods (or algorithms) for the mle method.

Usage

optimMethods(optimMethod = NULL,
             optimFun = c("both", "nloptr::nloptr", "stats::optim"))

Arguments

optimMethod	A character string used to find a method in a possible approximated fashion, see Examples.
optimFun	Value of the corresponding formal argument of the mle method, or "both". In the later case the full list of algorithms will be obtained.

Value

A data frame with four character columns: optimFun, optimMethod, globLoc and derNo. The column globLoc indicate whether the method is global ("G") or local ("L"). The column derNo indicates whether the method uses derivatives (D) or not ("N") or possibly uses it ("P"). Only methods corresponding the optimFun = "stats::optim" can have the value "P" for derNo. The data frame can be zero-row if optimMethod is given and no method match.

Caution

The optimization method given in the argument optimMethod of the mle method should be compliant with the compGrad argument. Only a small number of possibilities have been tested, including the default values.

References

See The NLopt website.

See Also

mle-methods, optim, nloptr.

Examples

optimMethods()
optimMethods(optimMethod = "cobyla")
optimMethods(optimMethod = "nelder")
optimMethods(optimMethod = "BFGS")
optimMethods("CMAES")

kergp documentation built on April 4, 2025, 2:29 a.m.