Nick Winter <[email protected]> reports,
> I accidentally stumbled on the following behavior, in which Stata seems
> to be allowing a missing value in a matrix:
> [...]
> . mat x=1e309
>
> . mat list x
>
> symmetric x[1,1]
> c1
> r1 1.#INF
>
> . di x[1,1]
> .
>
> . gen myvariable=x[1,1]
> (1141 missing values generated)
> [...]
> Does this work on other platforms? Is this supported behavior?
While I am very sympathetic to Nick's desire to store missing values in
matrices, I was not aware of the above and it should be treated as a bug --
something we at StataCorp need to fix -- and not a feature.
What is in fact being stored in the matrix is a NaN and, in the -generate-
follow on -- where missing values are expected -- Stata's NaN protection code
is detecting the NaN and converting it to a missing value. Somehow that NaN
snuck through matrix assignment, however, and unfortunately, the other matrix
routines are not up to dealing with it.
-- Bill
[email protected]
*
* 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/
