NLMIXED observation-wise Hessian

[derrick]

With the SUBGRAD option in NLMIXED I can get observation-wise gradient
vectors.  I would also like to get observation-wise components of the
Hessian but I do not see that option.  I think that option exists in
PROC GLIMMIX but PROC GLIMMIX does not have the general(ll)
distribution option.  Does anybody know how to get observation-wise
components of Hessian matrix out of PROC NLMIXED or know of another
procedure that does and has the general(ll) distribution option?