subset of complement rows

[Rad Schantz]

Sorry for the inelegant title -- Suppose I have a matrix of 100 rows, E(1..100). I can create a vector A=(21..30) and then extract a submatrix of rows, E[A,.]. That will comprise 9 rows. Then is there a straightforward way to extract a submatrix of rows not identified in A? It would comprise 91rows. Some command akin to E["not A",.]

[German Rodriguez]

I don't think you can do it with subscripts, but select() can help:

keep = J(rows(E), 1, 1)      
keep[A] = J(length(A), 1, 0) 
select(E, keep)

The vector -keep- has 1 for rows to keep (initially all) and 0 for rows to drop (obtained from vector A), and -select- does the work. No loops.

[Sergio Correia]

I'm not aware of any one-liner that does that, but you could use the select() function:

// First create the E and A objects:
E = J(5, 1, 1..5) + J(1, 5, 10 :* (1::5) )
A = 2 \ 3
// Note that A is a column vector (not a row vector as in your example)

//Then select the sample:
idx = J(rows(E), 1, 1)
idx[A] = J(rows(A), 1, 0)
select(E, idx)

Edit: just saw German's response; it's exactly the same so I'm guessing this trick is the accepted best practice!