All Invariant Moments of the Wishart Distribution (Q4677097)

The authors derive general expressions for expectations from polynomials of a central Wishart matrix \(U\) and its inverse \(U^{-1}\). E.g., for a real Wishart matrix \(U\) with scale parameter \(\Sigma\), \[ \mathbf{E}(U^3)=n\Sigma(\operatorname{tr} \Sigma)^2+(n^2+n)(\Sigma\operatorname{tr}(\Sigma^2)+2\Sigma^2\operatorname{tr} \Sigma) +(n^3+3n^2+4n)\Sigma^3, \] where \(n/2\) is the shape parameter of \(U\). (E.g., \(U/n\) can be the sample covariance of \(n\) i.i.d. Gaussian vectors with covariance \(\Sigma\).) It is shown how to use the Maple software for the symbolic evaluation of such moments.