In sci.stat.consult Rich Ulrich <[EMAIL PROTECTED]> wrote: >> > [ snip - redundant questioning ]
> Any covariance matrix computed on full data is assured to be > non-negative definite. That's the underlying theorem I was missing. I have to go try to prove it now. (I was thinking sampling fluctuation might once in a while bring up some sample matrices that violate this constraint, but I guess by analogy with sample univariate variances it can't happen. The proof isn't as easy for the matrix case) Thanks Ron . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
