Find the Optimum of a Response Surface

DESCRIPTION:

Finds the factor settings for the optimal predicted response value. Optimization can be bounded, unbounded or limited to a given interval for each factor.

USAGE:

optim.rsm.lm(rsm.object, best = "max",  bounded = T,  
     constrained, ...) 

REQUIRED ARGUMENTS:

rsm.object
the output from fitting a response surface using rsm.lm.

OPTIONAL ARGUMENTS:

best
character string giving the type of optimal value for the response, either maximum ( "max") or minimum ( "min").
bounded
logical value; if TRUE, the default, the optimization is performed within the range of the settings of the experimental factors, if FALSE the upper and lower bounds used in optimization are Inf and - Inf. If constrained is given, those bounds are used instead.
constrained
list specifying constraints in the optimization. The constraints can be a single value or two ordered values giving an interval constraint.

VALUE:

A list containing the following components:
optimum
the optimum predicted response value and the corresponding factor settings.
converged
type of convergence for the optimization algorithm.
log
The log component of the rsm.object, which specifies whether each variable is fitted on the log scale in the rsm model.
constrained
If there are constraints, returns the constraint list supplied.
outside
If the function converged, a logical vector specifying whether the optimum occured outside the range of the data used to fit the rsm.object .

NOTE:

If a variable is fitted on the log scale (for instance, if the design is generated using rsm.design, and log = T) in rsm.lm, the constraints are assumed to be specified on the natural scale. For example, if the design points on the (fitting) log scale for variable X are c(log(0.1), log(1), log(10)) , the optimization is constrained to the center point by setting constrained = list(X = 1).

SEE ALSO:

, , , , .

EXAMPLES:

ab.rsm <- rsm.lm(formula(abrsm.df), abrsm.df) 
optim(ab.rsm) 
optim(ab.rsm, constrained = list(Prime1 = 14)) 
optim(ab.rsm, constrained = list(RadDos = c(150,200),  
     Prime1 = c(-Inf, 14)))