A few solves that are symmetric could be done more efficiently if the inverse Cholesky class can do a solve of just one solve of the factorized triangular system. This is also useful for computed correlated standard errors when doing validation. This requires modifying ChoInv and ChoInvPivot in the linalg module to expose a solve_L method.
A few solves that are symmetric could be done more efficiently if the inverse Cholesky class can do a solve of just one solve of the factorized triangular system. This is also useful for computed correlated standard errors when doing validation. This requires modifying
ChoInv
andChoInvPivot
in thelinalg
module to expose asolve_L
method.