Covariance matrix than can be inverted using the Sherman-Morrison-Woodbury
forumla.
CSMW = CovarSMW(C,B)
Create the covariance matrix CSMW = C + B*B' than can be inverted efficiently
using the Sherman-Morrison-Woodbury formula, if size(B,2) is much smaller than
size(C,1):
inv(C + B*B') = inv(C) - inv(C)*B*inv(B'*inv(C)*B + I)*B'*inv(C)
The symetric matrix C should implement the methods: size, diag, mtimes,
mldivde and the matrix B should implement the methods: size, tranpose, mtimes,
sum.
Package: divand
