gptk (version 1.0)

SCGoptim: Optimise the given function using (scaled) conjugate gradients.

Description

Optimise the given function using (scaled) conjugate gradients.

Usage

options <- optimiDefaultOptions()
  newParams <- SCGoptim(x, fn, grad, options, ...)
  newParams <- CGoptim(x, fn, grad, options, ...)
  model <- modelOptimise(model, options, ...)

Arguments

model

the model to be optimised.

x

initial parameter values.

fn

objective function to minimise

grad

gradient function of the objective

options

options structure like one returned by optimiDefaultOptions. The fields are interpreted as\ option[1] : number of iterations\ option[2] : interval for the line search\ option[3] : tolerence for x to terminate the loop\ opt

...

extra arguments to pass to fn and grad

Value

• options
  an options structure
• newParams
  optimised parameter values
• model
  the optimised model.

See Also

modelObjective, modelGradient

Examples

## Not run to speed up package checks
# model <- GPLearn(..., dontOptimise=TRUE)
# options <- optimiDefaultOptions()
# model <- modelOptimise(model, options)