Inverse covariance matrices

Classes
class	SigmaI
	Basic inverse-covariance. More...
class	SigmaIblk
	Block-diagonal inverse-covariance. More...
class	SigmaIpex
	PEX inverse-covariance. More...

Detailed Description

Group of classes that implement inverse-covariance matrix updating. We are dealing with inverse-covarince (a.k.a. concentration) matrices because most algorithms in hierarchical models use these rather than covariance matrices. Thus, we save on matrix inversion operations by keeping track only of inverse-covariances. The save functions, however, output covariance matrices. All the internal algorithms for all the classes use the lower triangles of the underlying symmetric matrix. On saving, the matrices are symmetric and it does not matter for downstream applications. However, if a user wants to manipulate or interact with these in a sampler, addressing the lower triangle ensures stability of the algorithm.