Re: [Gretl-devel] State Space Modeling: possible mistake in User's Guide
Em 6 de março de 2018, Allin escreveu:
(...)
I think I see a cause for potential confusion. For the coefficient
matrices, subscript t indicates (possible) time dependence. For example, F
(r x r) is written as F_t in equation (33.1) to allow the the possibility
that it's time-varying (though most likely it's constant). The matrix of
exogenous variables (x), however, inherently has a leading dimension of T,
and the expression "x_t" refers, as Jack says, to the transpose of row t of
x.
I guess this could be stated when the dimension of x_t is first mentioned,
as in "x_t is a (k x 1) vector of exogenous variables, the transpose of row
t of a T x k matrix x." ?
Dear Allin and Jack,
I like Allin's idea :-). Seizing the opportunity... What do you think about
include more examples about state-space models on User's Guide? I think it
could be very useful for newcomers like me :-)
By the way, now I'm trying to replicate all Commandeur & Koopman's book
with the very important help of Gretl community (mainly Jack) and I would
love to share the results with all the users (I'm finishing chapter 5). I
think it could be available in the same way as the scripts from Greene and
Ramanathan.
In addition, we could ask Commandeur and Koopman to include our
replications as they did with R's replication on the website http://www.
ssfpack.com/CKbook.html
The main goal here is broaden Gretl's visibility :-)
Best,
Henrique
Attachments:
- attachment.html (text/html — 2.5 KB)