re: st: Use of matrix values in generate statements

[Daniel Feenberg <feenberg@nber.org>]

On Sun, 27 Mar 2011, Gordon Hughes wrote:

> I am sorry for being a little pernickety about the implications of this 
> discussion, but I would like to be clear about what is or is not possible 
> in this respect.
>
> The original observation was that the code:
>
> 	matrix input stdvalues (3700 6200...\3800 6350...\...
> 	gen stded = stdvalues[year-1992,filestat]
>
> does not work.  Dan found that

It was part of a larger program, and I stopped working on that approach 
when I read [U]14.9.

> 	gen int i = flpdyr-1992
> 	gen int j = filestat
> 	gen stded = stdvals[i,j]
>
> does work and Section 14.9 of my User Guide (Stata 10) says that
>
> 	gen stded = stdvals[2,3]
>
> should also work.

Note that the exact wording is "where r can c are scalar expressions, to 
obtain one element of matname as a scalar". Above "2" and "3" are 
certainly scalar expressions, but I need to use "flpdyr" and "filestat", 
for r and c which are vectors ("variables" in the language of Stata). 
Furthermore, I need a vector returned. The language in 14.9 is unambiguous 
(and incorrect) in claiming that that is not permitted. That is how I was 
led astray, and made the incorrect claim in my initial posting.

> So one inference is that -generate- can access matrix sub-elements when 
> either the sub-elements are either (a) constants, or (b) variables, but 
> not if they are defined as (c) expressions.  This would be odd given the 
> way most of Stata works.
>
> An alternative inference is that the problem lies in the variable type - 
> i.e. if variables or variable expressions are used to specify 
> sub-elements they must be of type int (or byte) but not float - and Dan's 
> second piece of code is required to force the result of the expression to 
> be of type int.
>
> Can anyone tells us what is the correct inference?

The coercion to int is optional, but speeds processing if the values are 
used repeatedly, as they are in my program.

Daniel Feenberg
NBER

> Gordon Hughes
> g.a.hughes@ed.ac.uk
>
> *
> *   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/