Questions about SsfFit

["Bernardo Ribeiro" <ribeirob@gmail.com>]

Dear users,

I am working on a Term Structure model that uses SsfFit and the Kalman Filter algorithm for a number of models.
As I was working with a simple Vasicek model (1factor) everything went alright, but as I waked into more complex models the problems atrted arising.

When I implemented a Babbs and Nowman model (1facto also) I kept receiving Hessian not invertible messages (which I solved multiplying the data series by 10) and the model ran ok with very good results. But now I am implementing a Babbs & Nowman approach with 2 factors. And then again I got the message with the 
non invertible hessian. I tried the same approach of multiplying the data series with no success.

After I tried some different starting values I got another message (even worse I suppose) that says: 
Problem in .C("S_KalmanFil",: subroutine S_KalmanFil: Missing values in argument 7 
Use traceback() to see the call stack

Has anyone that used Kalman Filtering ans the SsfFit model already faced some of these problems? If yes, how did you solve the problem?
Are there any other ways to solve the Hessian issue?

Thanks a lot for the help,

Bernardo Ribeiro
Msc. Student COPPEAD/UFRJ