Converting each row fo data into a matrix

[1MoreThing]

Dear All,

I have a long time series of a covariance matrix stored in a sas dataset. For each observation, there are 7 variables variables like:

"date c11 c21 c31 c22 c32 c33"

where c11 - c33 are the lower triangle elements of the cov-matrix.

In order to make use of some matrix routines in IML, I would like to convert each row into a full matrix:

M= {   c11 c21 c31,

          c21 c22 c32,

          c31 c32 c33 }

Could someone provide some suggestions on the transformation please?

Thanks you!

[IanWakeling · Accepted solution]

Have a look at the SQRSYM and SQRVECH functions which are designed for doing exactly this. If you have read your six variables into the matrix x, something like:

M=sqrvech(x[i,]);

should do the trick for the ith row.