|
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: How to save current status of Stata (eg. contents of macros, matrix and data)?
From |
Andreas Peichl <[email protected]> |
To |
[email protected] |
Subject |
st: How to save current status of Stata (eg. contents of macros, matrix and data)? |
Date |
Tue, 13 May 2008 18:56:51 +0200 |
Dear Statalisters,
(sorry for a possible double posting, but it seems to we that my first message
did not go through to the list, and now including a subject line)
I'm using Stata to call a program written in C++ to execute some simulations.
One run of this program takes about 15 min. I'm using several nested foreach
and forvalue loops to simulate different scenarios and to bootstrap the results
which I save in a matrix. The whole program runs for about a week. As this is a
quite time consuming exercise, I'm quite vulnerable to system crashes. I'm
therefore looking for a way to save the whole content of the Stata memory
(data, macros, matrix) to the hard disk which could then be restored if the
system crashes and I could restart at this point. (It would be something like
-preserve- and -restore- but not only for data).
Searching the Statalist archive, I came across this thread:
http://www.stata.com/statalist/archive/2005-01/msg00688.html
Using these programs to save the contents of the globals and locals together
with -save- and -svmat-, I was thinking of writing a "savetoHD" (and
readfromHD) command. I could then write my program in the following way (using
pseudo code):
readfromHD (x, start, end) // read in parameters from previous run, x is last
element used in loop
local start = `x' // set starting value of this loop to this
last element
forval x = `start'/`end' { // start loop
savetoHD // save parameters
simulation , ...
save results
}
(My problem is somewhat more complicated as I have nested loops but in principle
it should work this way).
Does anybody know a better way (or even a command) of how to achieve this
"preserve-restore" procedure? Any suggestions are warmly welcome.
Many thanks,
Andreas
_____________________
Dr. Andreas Peichl
Universit�t zu K�ln
Seminar f�r Finanzwissenschaft, Lehrstuhl Prof. Dr. C. Fuest
Albertus-Magnus-Platz | 50923 K�ln
Tel: +49(0)221-470-6201 | Fax: +49(0)221-470-5060
Mail: [email protected] | URL: http://www.wiso.uni-koeln.de/fuest
http://www.wiso.uni-koeln.de/ASPSamp/fuest/home/index.asp?l=15&wh=peichl.asp&ID_mitarbeiter=6
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/