Dear ox-users:
I'm trying to solve a dynamic panel probit problem maximizing a simulated likelihood function in Ox.
However, I have found a problem with the use of MaxBFGS (and the rest of optimizators) because they aren't able to retrieve the numerical derivatives although the program can evaluate the likelihood function.
I have changed the maximize.ox program (compiling it after to obtain the oxo program) with new values for SQRT_VALUES and DIFF_VALUES (i.e. the machine epsilon) but I couldn't make the derivative.
Has anybody suffered a similar problem?. Can suggest me any idea?.
Thanks for your attention and Happy New Year!
Best Regards
Román Mínguez
Economics Faculty
Univ. San Pablo-CEU (Madrid).
___________________________________
NOCC, http://nocc.sourceforge.net
|