FAiR (version 0.4-15)

restrictions2model: Map free parameters to matrices

Description

It is not necessary to understand this help page if one merely wants to estimate a factor analysis model. This help page is intended for those who want to modify or extend FAiR or otherwise want some idea of how FAiR works “behind the scenes”.

This S4 generic function takes a numeric vector of free parameters and manipulates it into the matrices that are typical when estimating a factor analysis model. There is little need to call it directly. Various methods are defined for the restrictions object, corresponding to different kinds of factor analysis models; see FAiR-package.

Usage

"restrictions2model"(par, restrictions, manifest, lower, mapping_rule)

Arguments

par

A numeric vector containing values for all the free parameters to be estimated, which corresponds to the par argument for genoud and for optim.

restrictions

An object of restrictions-class.

manifest

An object of manifest-class.

lower

A small numeric scalar indicating the lower bound for positive definiteness or minimum uniqueness; see the corresponding argument to Factanal.

mapping\_rule

a logical that controls whether the mapping rule is invoked in semi-exploratory factor analysis. This argument is used to circumvent the mapping rule in some circumstances, like when the function is called indirectly by optim

Value

A list with exactly two elements named

criteria

a numeric vector

restrictions

a filled-in version of the restrictions object

Methods

The arguments that are part of the signature are restrictions and manifest. Methods are currently only defined for objects of class "manifest.basic", which are inherited by objects of class "manifest.data" and "manifest.data.mcd". There are methods for each of the classes that inherit from restrictions-class, except for "restrictions.factanal", which does not utilize the restrictions2model mechanism. There are also two arguments that are not part of the signature. The first is par, which is a numeric vector of free parameters and is conceptually similar to the par argument to optim. The second is lower, which is a small positive number that is used as a threshold for positive-definiteness of various matrices.

Details

This function is called internally by Factanal thousands of times during the course of the optimization. Let the factor analysis model in the population be $$\Sigma = \Omega(\beta\Phi\beta^\prime + \Theta)\Omega$$ and let $theta$ be a vector of all the free parameters in the factor analysis model. The restrictions2model methods are essentially a mapping from $theta$ to the free elements of $beta$, $Phi$, and $Omega$. The methods are currently defined on the restrictions argument, which is fairly skeletal at the outset, is filled in by the body of the restrictions2model method using a bunch of calls to the make_parameter-methods.

The restrictions2model method can (and does) also do some checking of whether the parameters in par are admissable in the context of a factor analysis model. For each admissability check, if par is admissable, it should receive the value of $-1.0$ in a numeric vector whose length is equal to the number of admissability checks. If par is inadmissable with respect to some check, it should receive a value greater than $-1.0$. This numeric vector should also be returned as an element of a list called "criteria".

See Also

make_parameter, restrictions-class

Examples

  showMethods("restrictions2model")