Language Reference

NDX2SUB (dim, indices) ;

This function is supported by the IML procedure and the iml action.

The NDX2SUB function is part of the IMLMLIB library. The NDX2SUB function converts indices of a matrix into subscripts for the matrix. The arguments are as follows:

The indices of an matrix are the elements . The indices enumerate the elements in row-major order: the first p indices enumerate the first row, the next p indices enumerate the second row, and so forth. The NDX2SUB function converts indices to subscripts, which are pairs such that and .

You can use the module to display the rows and columns of elements that satisfy a certain condition. For example, the following statements locate all the even numbers in a matrix and then call the NDX2SUB function to find the subscripts of the even elements:

x = {1  2  3,
     4  5  6,
     7  8  9,
    10 11 12};
idx = loc( mod(x, 2)=0 );
dim = nrow(x) || ncol(x);
s = ndx2sub(dim, idx);
print s;

You can also use the NDX2SUB function to keep track of indices and subscripts of multidimensional arrays. Although the SAS/IML language does not support multidimensional arrays, a common technique is to store the elements of a array in a two-dimensional matrix with rows and columns. For example, you can store the contents of four arrays in a single matrix, as shown in the following program:

/* Store four 3x3 matrices in a 12x3 matrix
  (each group of three rows is a matrix) */
dim = {4 3 3};
m = j(12, 3);
p = 9; /* = prod(dim[2:ncol(dim)]) */
do i = 1 to 4;
   startNdx = 1 + (i-1)*p;
   endNdx   = i*p;
   ndx = startNdx:endNdx; /* get indices for i_th matrix */
   m[ndx] = i; /* assign or extract matrix */
   subscripts = ndx2sub(dim, ndx); /* or get subscripts */
end;
print m;

To convert from subscripts to indices, see the SUB2NDX function.