Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


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

st: AW: Multi-processor environment for time series regressions


From   "Martin Weiss" <martin.weiss1@gmx.de>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: AW: Multi-processor environment for time series regressions
Date   Tue, 29 Jun 2010 17:10:46 +0200

<> 


" All regressions will be performed subsequently and not in parallel. Given
STATA's architecture, the calculations are very time consuming and
multi-processing doesn't help much."






It is not Stata`s fault, I would say: The reasoning behind the slow speed of
TS commands, even under multiple processors, is on the first page of
http://www.stata.com/statamp/report.pdf:

"The other half
of the commands run slower than the median
speedup, and some commands are not sped up
at all, either because they are inherently se-
quential (time-series commands)"



HTH
Martin


-----Ursprüngliche Nachricht-----
Von: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu] Im Auftrag von Christian
Stammschulte
Gesendet: Dienstag, 29. Juni 2010 17:03
An: statalist@hsphsun2.harvard.edu
Betreff: st: Multi-processor environment for time series regressions

Dear All,

I have a panel dataset with about 4,000 equity portfolios (id) with weekly
returns for about 4 years. For each of the ids, I am running a time series
regression on Carhart's (1997) four factors. Hence I obtain coefficients
(alpha,beta,etc) from the 4,000 regressions which I post into a dta-file for
subsequent use. 

Since I would like to determine the distribution of alphas from the
regressions, I am using 1,000 boostrapping simulations, each simulation
randomly determining periods which are taken into account for additional
4,000 regressions. 

This leads me to 4,000,000 regressions where the problems comes into play.
All regressions will be performed subsequently and not in parallel. Given
STATA's architecture, the calculations are very time consuming and
multi-processing doesn't help much. The question is whether there is any way
to run those regressions in parallel/more supporting a multi-processor
environment? I was thinking about a panel/mata-setup but unfortunately
haven't found a way yet.

Any help/good ideas would be much appreciated.

Best regards,

Christian Stammschulte
House of Finance - University of Frankfurt

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


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


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index