Re: st: Errors "out of range" for a simple Mata function

[Christophe Kolodziejczyk <ck.statalist@gmail.com>]

Just a minor point on the use of the token() function together with
st_data() and st_view()  (or even st_store()).
Token() creates a vector which elements contain the different words of
a string and is no longer necessary since Stata 11. It may be
necessary with st_sdata() and st_sview() with Stata 11 though. You can
write

st_view(X = .,.,varlist) instead.

See help whatsnew for more details

Christophe


2012/11/20 Nick Cox <njcoxstata@gmail.com>:
> Thanks for the correction. The word "wrong" was itself wrong. I was
> elevating my personal preference only to use views for data Ithat I
> won't change to an implication that you can't do it, which was
> incorrect.
>
> Given
>
> local macro generate
> local macro varlist
> local macro touse
>
> as discussed to date in the thread
>
> . mata : st_store(., "`generate'", "``touse''", rowsum(st_data(.,
> tokens("`varlist'"), "``touse''")))
>
> is I think a complete solution. However, I'd sympathise with anyone
> who thought that overvalues conciseness. Also, using Mata as a
> calculator in this way undermines Ly Tran's major purpose of using the
> row sum problem as a vehicle for learning Mata.
>
> Nick
>
> On Tue, Nov 20, 2012 at 2:01 AM, Christopher Baum <kit.baum@bc.edu> wrote:
>
>> Ly Tran:
>>
>> Nick Cox is correct in pointing out that you need to pass `generate' to your Mata function. But it is not necessary to use st_store() as he suggests. Your Mata code works with one small change:
>>
>> capture program drop testsum
>> program testsum, rclass
>> syntax varlist(numeric) [if] [in], GENerate(string) [FLOAT]
>> qui generate `float' `generate' = .
>> mata: tsum( "`varlist'", "`generate'")
>> end
>>
>> mata
>> void function tsum (string scalar varlist, string scalar newvar)
>> {
>> real matrix X, Z
>> st_view(X = .,., tokens(varlist))
>> st_view(Z =., ., newvar)
>> Z[.,.] = rowsum(X)
>> }
>> end
>>
>> When an entire view matrix is the LHS of a statement, you must use the syntax shown, rather than the name of the view matrix, for the statement to update the view matrix (and the Stata variable(s)) and not create a new Mata object. In the help for st_view(), read the section "Cautions when using views 2:  Assignment" to see this point illustrated. It is also mentioned in the Mata chapter of ITSP below.
>>
>> . sysuse auto,clear
>> (1978 Automobile Data)
>>
>> . testsum price mpg turn, gen(junk)
>>
>> . egen junk2=rowtotal(price mpg turn)
>>
>> . su junk*
>>
>>     Variable |       Obs        Mean    Std. Dev.       Min        Max
>> -------------+--------------------------------------------------------
>>         junk |        74    6226.203    2948.149       3354      15972
>>        junk2 |        74    6226.203    2948.149       3354      15972
>>
>>
>> One other thing: if you allow [if] and [in] in your ado-file, you should -marksample touse- and pass `touse' to Mata so that it can limit the observations being used in the Mata routine. Again, example code in ITSP.
>
> *
> *   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/



-- 
Christophe Kolodziejczyk
Research Fellow

AKF, Anvendt KommunalForskning
Danish Institute of Governmental Research
Købmagergade 22
DK-1150 København K

*
*   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/