attached to the object in 'gnls' even though
the data frame is sorted therein, since the call to 'Initialize' within
'gnls' will not then change them. If this explanation is true,
'Initialize.corCompSymm' should not accept an unordered 'data' formal,
or should sort the
Hi Douglas,
thanks for your answer.
My question indeed arises from a sparse matrix context: 'A' is sparse
symmetric, and 'C' must also be sparse since it would otherwise fill.
It comes from a Bayes regression with a very large number of parameters;
a loop over blocks will do the job in my sp
'update' method for
Cholesky decomposition objects seems limited to a new matrix A + m*I
with a scalar (diagonal) modification.
If this is true: are there plans to implement such up/downdates?
Best,
Yves
Yves Deville, statistical consultant, France.
_
