mfx has some embedded utility programs taking derivatives, but seems cannot
be used in a stand-alone fashion. What I am trying to do is to be able to
estimate model based on some prior functional form and to get score function
or the gradient/and Hessian of that functional form with respect to the
parameter estimate. Or basically, d(f(X, B)/dB and d(d(f(X,B))/(dBdB')and
use estimates to replace B. I don't want to use testnl as it reestimate the
model. Any ado files that's available or any good algorithsm you would
suggest? Really appreciate.
Jun Xu
Department of Sociology
Indiana University at Bloomington
