1 Reply Latest reply on Sep 4, 2008 3:04 PM by chipf

    Documentation error in DRANDMULTINORMAL?

    ssclift
      Leading dimention of correlation matrix

      This seems to be a minor, but confusing error in the DRANDMULTINORMAL documentation:

      — Input: DOUBLE PRECISION C(LDC,M)

      On input: variance / covariance matrix for the distribution.

      — Input: INTEGER LDC

      On input: leading dimension of C in the calling routine.
      Constraint: LDC>=N.

      Should that N not be "M"?  If I want to generate 10000 x M sets of correlated Normals, surely I don't need to make 10000 copies of the M x M correlation matrix?

       

      Thanks.

      -- Simon