Draw n random d-dimensional vectors from a multivariate Gaussian distribution with mean mu(nxd) and covariance matrix Sigma(dxd).
If n is given then mu can be a 1-by-d vector. If it is not given mu must be n-by-d.
If the argument tol is given the eigenvalues of Sigma are checked for positivity against -100*tol. the default value of tol is
eps*norm (Sigma, "fro").