mcprofile (version 0.0-9)

mcpcalcRatioBFGS: Construction of Multiple Contrast Profiles for Ratios of Parameters

Description

Only for internal use. Calculating profiles by optimization with linear constraints using optim.

Usage

mcpcalcRatioBFGS(object, CMn, CMd, control=mcprofileControl(), margin=NULL)

Arguments

object

An object of class glm or lm

CMn

A numerator contrast matrix with the number of columns equal the number of estimated parameters. Contrast coefficients should be -1 and 1, respectively. Providing row names is recommendable.

CMd

A denominator contrast matrix equivalent to the numerator matrix.

control

A list with control arguments. See mcprofileControl.

margin

Numeric vector of margins, defining the profiling range. If NULL no extra margin is considered.

Value

• An object inheriting from class mcprofileRatio with methods show, plot, confint, and test

Details

The signed root deviance is calculated and interpolated by interpolation splines. Control arguments are provided to define the profiling range: steps correspond to the number of values evaluated around the MLE, alphamax defines the maximal cutoff value by choosing a quantile of a normal distribution at 1-alphamax divided by the number of rows of the contrast matrix, if fixed.range is TRUE the deviance is evaluated at equal-sized steps in a range provided by the margin argument.

See Also

optim, profile.glm, sci.ratio, interpSpline, contrMatRatio