optimise.eu {bdpopt} R Documentation

## Optimise Expected Utility

### Description

Optimisation of expected utility, either directly over the results of a grid evaluation performed by eval.on.grid or by optimisation of the regression function constructed by fit.gpr or fit.loess. This is a generic function for an S3 object.

### Usage

optimise.eu(model, start, method = "L-BFGS-B",
lower = -Inf, upper = Inf, control = list())

### Arguments

 model A simulation model object returned by eval.on.grid, fit.gpr or fit.loess. Specifying any method other than "Grid" requires that the object has been obtained from fit.gpr or fit.loess. start The start value when performing the search for a maximum. Passed on to optim. method The optimisation method to be used. Must be one of "Nelder-Mead", "BFGS", "CG", "L-BFGS-B", "SANN", "Brent" or "Grid". lower A numeric, atomic vector giving the lower limits for the decision variables when performing the maximisation. upper A numeric, atomic vector giving the upper limits for the decision variables when performing the maximisation. control A list of control parameters passed on to optim.

### Details

The optimisation strategy depends on the value of method. All arguments except model are ignored if the method "Grid" is used. If "L-BFGS-B" is used, then the arguments lower and upper are passed on as specified to optim as the lower and upper limits for the optimisation of the decision variables. If any other value is provided for method, then optim will be used to maximise a function defined to be equal to the objective function when the decision variable argument x satisfies x >= lower, x <= upper and equal to -Inf otherwise. The actual lower and upper limits passed to optim in this last case are -Inf and Inf, respectively.

### Value

A list with components

 opt.arg A named vector containing the optimal values for the decision variables. opt.eu An estimate of the optimal expected utility.

### Author(s)

Sebastian Jobjörnsson jobjorns@chalmers.se

[Package bdpopt version 1.0-1 Index]