Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: RE: Efficient parallel computing in Stata/MP


From   Demian Panigo <[email protected]>
To   [email protected]
Subject   Re: st: RE: Efficient parallel computing in Stata/MP
Date   Fri, 27 Sep 2013 09:58:25 -0300

Thanks Daniel,
I'm finishing a Stata "exhaustive model selection" code.....
particularly intended for analysing parameter stability across many
alternative specifications (e.g. testing all possible combination of
control variables).
Of course, because problem size grows exponentially with n (the
initial number of admissible covariates) we include some options to
efficiently cut the search universe (other than those included in
VSELECT).
Anyway, I need to notify users about the best software strategy to be
chosen (depending on user's hardware) in order to minimize computing
time.
So, before doing the "try-and-record" standar methodology (with
different Stata versions, different parallelism methodologies and
diferent types of regressions) I'm asking statalist member for "first
step advice"
Thanks a lot
Demian






2013/9/27 Daniel Feenberg <[email protected]>:
>
>
> On Fri, 27 Sep 2013, Demian Panigo wrote:
>
>> Excuse me... one more point.
>> When I say many regressions..... many is about one hundred million.
>
>
> Can I ask what distinguishes the regressions from one another? Is it
> bootstrapping? If the regressions differ from each other only in by a single
> observation, note that you create a cross-product matrix for the next
> regression regression by subtracting one observation from the last matrix
> and adding another, no need to do all the arithmetic for the other
> observations each time. Saves a lot of arithmetic if you have many
> observations, not so much if you have many variables. Stata doesn't support
> that, but there was a Fortran package years ago (f4stat I believe it was
> called), that was designed just for that.
>
> Daniel Feenberg
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/



-- 
Demian T. Panigo
Lic. en Economía, UNLP,
Master en Cs Sociales, UBA,
Doctor en Economía, EHESS-ENS (Paris)
Investigador Adjunto del CEIL-PIETTE del CONICET
Docente investigador de la UNM, de la UNLP, de la UBA, y de
Paris-Jourdan Sciences Economiques-ENS.
Miembro del Programa de Formación Popular en Economía (PROFOPE)

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index