THanks for the quick reply.
We are using the same filtered_func_data and only slightly changing the design model. In our tests (see attachments), after slightly changing the design model(design1.mat to design2.mat) and keeping the same filtered_func_data, the correlation between the two estimated ar coefficients is .9998. So if we are going to slightly change our design matrix 100 times, it would be a large time saver to be able to skip the coefficient estimation part of film_gls.
I hope this makes more sense.
-jordan
|