We consider the problem of estimating differences in two Gaussian graphical
models (GGMs) which are known to have similar structure. The GGM structure is
encoded in its precision (inverse covariance) matrix. In many applications one
is interested in estimating the difference in two precision