Re: [Gretl-devel] VARs as system special case ($sysA, $sysB, etc.)

Am 31.10.2015 um 16:10 schrieb Sven Schreiber: > Am 31.10.2015 um 11:19 schrieb Sven Schreiber: >> Am 30.10.2015 um 19:26 schrieb Allin Cottrell: > >>> >>> Anyway, what you're suggesting makes sense. It would help if you could >>> come up with some sort of hansl prototype or even reasonably specific >>> pseudo-code. Then I'd be happy to try to make C out of it. >> >> Will do that. I vaguely remember that there is some undocumented bundle >> floating around after estimating models that might contain some of the >> standard accessors. Is this correct, and if so, how is it >> called/accessible from hansl? > > Hm, attempting this turns out to be more difficult than expected -- > after VAR estimation I don't know how to capture the information about > the deterministic spec like constant, trend, and seasonals, there are no > accessors for that. (And they don't show up in $xlist.) > > In the meantime I will pass this info in an ad-hoc way. > Ok, here you go, VAR only so far (no VECM). Not really tested yet, sorry. There are two functions, working like this: # transfer the available accessors bundle bvar = packVaracc($ylist, 3, $compan,$coeff,$stderr,$sigma,$T,$xtxinv, 0, $xlist) # (where the integer after $ylist is the deterministic Johansen case; # 1: no constant, 2/3: constant, 4/5: constant and trend. # And the penultimate argument is the boolean switch seasonals yes/no. # add the new accessors putSysacc(&bvar) Also with respect to $vcv I remember some discussions about whether or not to provide a dof-correction, which ultimately led to the introduction of $xtxinv for VARs instead of $vcv. Nevertheless I have now calculated simply vcv = sigma ** xtxinv.