Good question, except that I am not clear on what you are trying to do.
The seed at the beginning of your process, together with your number of
observations, would seem enough to characterise a particular simulation.
As I understand it, -simulate- is executed just once, and so any
initialisations are carried out just once. Whether you pass stuff as c()
or r() makes no difference to that.
In particular, anything passed as a macro, or in equivalent form, will
get evaluated just once.
You'll likely get a better answer from others who do this kind of thing
much more than I do.
Nick
Rachel
Nick,
Thanks very much for the reply. In fact, I am using Stata 10. But
suppose I want to capture the state of c(seed) not just at the end of
each run of the program, but at several points therein. It seems
that, to do this, one would have to copy c(seed) to an r-class result?
For example:
program define TrialProgram
clear
set obs 9000
capture drop x y
drawnorm x y
//Want to return c(seed) here
regress y x
drawnorm z1 z2
//Want to return the NEW state of c(seed) here
regress z1 z2
end
Rachel
On Thu, May 29, 2008 at 1:17 PM, Nick Cox <[email protected]> wrote:
> -capture drop x y- is redundant given -clear-.
>
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]] On Behalf Of Nick Cox
> Sent: 29 May 2008 18:11
> To: [email protected]
> Subject: st: RE: Returning a local macro within -simulate- command:
Type
> mismatch error?
>
> Is this Stata 10? (If you are using an earlier version, you are asked
to
> state that fact.)
>
> This seems to work for me in Stata 10:
>
> capture program drop TrialProgram
> program define TrialProgram
> clear
> set obs 9000
> capture drop x y
> drawnorm x y
> regress y x
> end
>
> simulate _b _se, seed(`c(seed)') noisily reps(30): TrialProgram
>
> Notice that you can cut out the middle macro. There is no need to copy
> from c(seed) to a r-class result, as c(seed) remains accessible.
>
> Nick
> [email protected]
>
> Rachel
>
> I'm running a simulation and would like to know the state of the
> random number generator in each loop. A simplified version of the
> program is below. I'm getting the error "Type mismatch r(109)"
> apparently as a result of the way I'm returning the local macro
> seedinit.
>
> I may be missing something obvious, but I'd appreciate any suggestions
> about how to fix this error. Thanks very much.
>
> capture program drop TrialProgram
> program define TrialProgram, rclass
> clear
> set obs 9000
> local seedinit1=c(seed)
> capture drop x y
> drawnorm x y
> regress y x
> return local seedinit "`seedinit1'"
> end
>
>
> simulate _b _se seedinit=r(seedinit), noisily reps(30): TrialProgram
*
* 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/
