Just a heads up that I would like to change the semantics and calling sequence of the updatemu! methods. Alternatively I could split the updating part of
updatemu!{T<:FPVector}(r::GlmResp{T}, linPr::T)
into a separate function.
For an LmResp this method is trivial. For a GlmResp this method copies linPr into the eta member (or adds linPr and offset into eta) then updates mu and other vectors. In the generalized linear mixed models code I would like to build eta from various terms in place, then update mu and the other vectors. It would help to have the updating part split out from the copying part of the updatemu! methods.
Does anyone have any suggestions for the best approach? I think the second is preferable except that the name updatemu! would then more properly apply to the function called by updatemu!.
Just a heads up that I would like to change the semantics and calling sequence of the
updatemu!
methods. Alternatively I could split the updating part ofinto a separate function.
For an
LmResp
this method is trivial. For aGlmResp
this method copieslinPr
into theeta
member (or addslinPr
andoffset
intoeta
) then updatesmu
and other vectors. In the generalized linear mixed models code I would like to buildeta
from various terms in place, then updatemu
and the other vectors. It would help to have the updating part split out from the copying part of theupdatemu!
methods.Does anyone have any suggestions for the best approach? I think the second is preferable except that the name
updatemu!
would then more properly apply to the function called byupdatemu!
.