scipy.stats._multivariate:_cholesky_invwishart_rvs

Notes

Intended to be used as a step in the process for computing random variates of a matrix t distribution ${\mathcal{T}}_{m,n}$ by appealing to its alternative form as a matrix mixture .. math mathcal{T}_{m,n}( mathrm{df}, mathrm{M}, Sigma, Omega ) = mathcal{N}_{m,n}( mathrm{M}, mathcal{W}^{-1}_m(mathrm{df} + m - 1, Sigma), Omega ) = mathcal{N}_{m,n}( mathrm{M}, Sigma, mathcal{W}^{-1}_n(mathrm{df} - n + 1, Omega) ) where ${\mathcal{N}}_{m,n}$ is a matrix normal distribution and ${\mathcal{W}}_d^{-1}$ is an inverse Wishart distribution. Accordingly, the degrees of freedom adjustment $\mathrm{df}\to \mathrm{df}+d-1$ occurs in the scope of this function.