Fine, but your editing is selective. You chopped
the reference to -matselrc-
from STB-56 in 2000, which does the right
thing for names and covers matrices of arbitrary
size.
Conversely, Mata is the way forward. I infer
from this thread that Mata has no primitives
for rotating matrices. APL had those some
forty years ago....
Nick
n.j.cox@durham.ac.uk
Phil Schumm
> On May 18, 2006, at 12:06 PM, Nick Cox wrote:
> > mat b = e(b)
> > local R = rowsof(b)
> > local r = `R' - 1
> > mat b = b[1,`R'], b[1,1..`r']
>
>
> On May 18, 2006, at 12:28 PM, Alan Feiveson wrote:
> > Here is a way to do it using a permutation matrix C (see example
> > below for N=4)
>
> <snip>
>
> > . matrix A=I(4)
> >
> > . matrix c=A[1..1,.]
> >
> > . matrix B=A\c
>
> <snip>
>
> > . matrix C=B[2..5,.]
>
> <snip>
>
> > . matrix bn=b*C
>
>
> These solutions are both fine and, as I noted initially,
> resorting to
> Mata may not necessarily have been the most appropriate strategy in
> this case. The only thing I'd add is a bit of advertising I
> neglected before; the function I gave not only shifts the
> columns but
> also the column names (if any), and in fact, will work not only with
> 1 x c rowvectors but also with arbitrary r x c matrices.
*
* 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/