mxFitFunctionR {OpenMx}R Documentation

Create MxFitFunctionR Object

Description

mxFitFunctionR returns an MxFitFunctionR object.

Usage

mxFitFunctionR(fitfun, ..., units="-2lnL")

Arguments

fitfun

A function that accepts two arguments.

...

The initial state information to the objective function.

units

(optional) The units of the fit statistic.

Details

The mxFitFunctionR function evaluates a user-defined R function called the 'fitfun'. mxFitFunctionR is useful in defining new mxFitFunctions, since any calculation that can be performed in R can be treated as an mxFitFunction.

The 'fitfun' argument must be a function that accepts two arguments. The first argument is the mxModel that should be evaluated, and the second argument is some persistent state information that can be stored between one iteration of optimization to the next iteration. It is valid for the function to simply ignore the second argument.

The function must return either a single numeric value, or a list of exactly two elements. If the function returns a list, the first argument must be a single numeric value and the second element will be the new persistent state information to be passed into this function at the next iteration. The single numeric value will be used by the optimizer to perform optimization.

The initial default value for the persistant state information is NA.

Throwing an exception (via stop) from inside fitfun may result in unpredictable behavior. You may want to wrap your code in tryCatch while experimenting.

Value

Returns an MxFitFunctionR object.

References

The OpenMx User's guide can be found at http://openmx.psyc.virginia.edu/documentation.

Examples


# Create and fit a model using mxFitFunctionR

library(OpenMx)

A <- mxMatrix(nrow = 2, ncol = 2, values = c(1:4), free = TRUE, name = 'A')
squared <- function(x) { x ^ 2 }

# Define the objective function in R

objFunction <- function(model, state) {
    values <- model$A$values 
    return(squared(values[1,1] - 4) + squared(values[1,2] - 3) +
        squared(values[2,1] - 2) + squared(values[2,2] - 1))
}

# Define the expectation function

fitFunction <- mxFitFunctionR(objFunction)

# Define the model

tmpModel <- mxModel(model="exampleModel", A, fitFunction)

# Fit the model and print a summary

tmpModelOut <- mxRun(tmpModel)
summary(tmpModelOut)


[Package OpenMx version 2.2.4 Index]