corFactor.corStruct {nlme} | R Documentation |
This method function extracts a transpose inverse square-root factor,
or a series of transpose inverse square-root factors, of the
correlation matrix, or list of correlation matrices, represented by
object
. Letting S denote a correlation
matrix, a square-root factor of S is any square
matrix L such that S = L'L. This method
extracts L^(-t).
## S3 method for class 'corStruct': corFactor(object, ...)
object |
an object inheriting from class corStruct
representing a correlation structure, which must have been
initialized (using Initialize ). |
... |
some methods for this generic require additional arguments. None are used in this method. |
If the correlation structure does not include a grouping factor, the
returned value will be a vector with a transpose inverse square-root
factor of the correlation matrix associated with object
stacked
column-wise. If the correlation structure includes a grouping factor,
the returned value will be a vector with transpose inverse
square-root factors of the correlation matrices for each group, stacked
by group and stacked column-wise within each group.
This method function is used intensively in optimization
algorithms and its value is returned as a vector for efficiency
reasons. The corMatrix
method function can be used to obtain
transpose inverse square-root factors in matrix form.
Jose Pinheiro Jose.Pinheiro@pharma.novartis.com and Douglas Bates bates@stat.wisc.edu
corMatrix.corStruct
,
recalc.corStruct
, Initialize.corStruct
cs1 <- corAR1(form = ~1 | Subject) cs1 <- Initialize(cs1, data = Orthodont) corFactor(cs1)