Re: ped function in asreml-R

Dear List

See Pag. 49 in asreml-S.pdf

"
4.4.5 Known relationship structures
giv(obj, var=FALSE, init=NA)
ped(obj, var=FALSE, init=NA)

Required arguments
obj a factor in the data frame. The name obj must also
appear as a component in
the ginverse list argument to asreml.control() to
associate an inverse relationship
matrix with the factor oj
Optional arguments
var if TRUE a variance parameter is associated with the
structure. If used in a direct
product with another variance structure then accept the
default var=FALSE
(See Section 4.5).
init a vector of length 1 giving the initial variance
parameter value. This scalar can
have an optional names attribute to set the boundary
constraint. In this case,
the name may be one of ”P”, ”U” or ”F” for positive,
unconstrained or fixed,
respectively.
Details
The giv() procedure associates a known inverse matrix with
the factor obj; the number of
rows in the inverse matrix must be length(levels(obj)) and
the order is assumed correct.
The ped() function associates an inverse relationship
matrix typically derived from a
pedigree with the factor obj. There may be more rows in
the inverse matrix than levels of
obj. More than one inverse matrix may be used in an
analysis so the ginverse argument
to asreml must be used in conjunction with giv or ped to
associate particular inverse
matrices with factors in the model."

On Thu, 12 Jul 2007 13:38:18 +0100
  iwhite_at_STAFFMAIL.ED.AC.UK wrote:
> Dear List,
>
> Does anyone know the purpose of the var=T option in
>
> asreml(Y ~ ... + ped(factor, var=T) + ... )
>
> --
> This message has been scanned for viruses and
> dangerous content by MailScanner, and is
> believed to be clean.
>

--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.