On Thu, Oct 23, 2025 at 7:40 AM Allin Cottrell <[email protected]> wrote:
> On Thu, 23 Oct 2025, Paolo Chirico wrote: > > > I finally figured out where the problem was. > > The original filter bundle (let's call it A) contains the correct > settings, > > but if this bundle is copied (for example, B=A), B fails to copy > everything > > because the "exact" member is not copyable and the default value, 0, > remains > > in B! > > OK, so we need to ensure that our kalman functions do not mess with > the specifications initially passed in by the user via the diffuse, > sequential and dejong bundle members. We can "translate" them for > internal run-time use but should not modify them. That's now fixed, I believe, in the current gretl snapshots. Specifically, explicit settings by the user of "diffuse", "sequential" and/or "dejong" in a state-space model bundle should not be changed by execution of the kfilter() function. Allin
_______________________________________________ Gretl-users mailing list -- [email protected] To unsubscribe send an email to [email protected] Website: https://gretlml.univpm.it/postorius/lists/gretl-users.gretlml.univpm.it/
