|From||David Airey <email@example.com>|
|Subject||re: Re: st: RE: Stata 9 announcement|
|Date||Tue, 5 Apr 2005 13:35:59 -0500|
I am curious what the implications are for -gllamm-. Will it be faster now? Will Stata now have built-in alternatives to much of its features? I've avoided -gllamm- partly because I heard that, while it was a great program, it could be unbearably slow. Speeding up -gllamm- seemed to be a frequent request on this list; does V. 9 address this at all?I'm just speculating and ignorant, but has "Mata" been put in place in part to deal with this? Maybe gllamm has to be re-written in Mata and plug-ins to achieve faster timing and maybe this is being left up to its authors. Stata seems to want to provide the tools to solutions in addition to many solutions. From the online pages, seems speed and Mata are in the same sentence a few times and that xtmixed is written with Mata. I guess there were 3 solutions to speed up gllamm, (1) code it into Stata (outside ado's) (2) make C plug-ins possible (3) think hard about the .ado architecture for programs like gllamm.