st: Update to program - tuples - on SSC

[Joseph Luchman <jluchman@gmail.com>]

With many thanks to Kit Baum, an update to the program - tuples - is
available on SSC.

The update has two separate version requirements:

For the new standard computation, Stata version 10 is required

For computation with the "nomata" option, Stata version 8 is required.

Nick Cox and I have been working to speed up the algorithm to compute
all possible tuples which, as is noted in the help file of the
previous version, is naive and slow for longer lists (i.e., up to or
greater than 7).

The updated version brings the computation of all "ntuples" into Mata
using cvpermute().

Computation time savings is especially pronounced around 7
elements/variables in the list.  For example compare computation speed
using both:

tuples a b c d e f g h i j, display nomata

tuples a b c d e f g h i j, display


The "display" option helps to illustrate the speed boost gained with
the Mata approach.  Moreover, the speed boost helps to make long lists
feasible within a _reasonable_ time period.  For example, all
combinations of letters from a to t (20 elements and 1,048,575
combinations) in the English alphabet - on my Dell Optiplex 990 with
an Intel i7-2600 processor (3.4 GHz) with Stata 12.1 MP took about 1.5
minutes to complete - after this number of elements wait times begin
to grow more lengthy.

The results are presented in the same way as the previous versions
(i.e., in the same order), which can be observed by using the
"display" option.

For users with Stata versions 8 and 9, - tuples - will still work just
as it did previously, save that you will have to add the "nomata"
option.

The algorithm has only been tested up to about 23 elements.  Please
feel free to contact Nick and I about the program if issues are
encountered during use or for suggestions.


To update type:

adoupdate tuples, update



For an excellent discussion of the command please see Nick's previous
post on Statalist in 2006.

http://www.stata.com/statalist/archive/2006-12/msg00060.html


- joe

Joseph Nicholas Luchman, M.A.
----
Senior Research Associate | Fors Marsh Group
Email: jluchman@forsmarshgroup.com
forsmarshgroup.com
----
Doctoral Candidate
Industrial Organizational Psychology
George Mason University
https://www.researchgate.net/profile/Joseph_Luchman/
*
*   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/