[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
## Re: Multivariate student t: normalization in mvtrnd

**From**: |
Iñigo Urteaga |

**Subject**: |
Re: Multivariate student t: normalization in mvtrnd |

**Date**: |
Mon, 26 Nov 2012 18:03:40 -0500 |

Hi Dan,
Thank you for your help. As you suggest, V is the covariance matrix
and that is indeed why in my proposed method:
samples=Sigma' * mvtrnd( eye(n), nu, M)'; (1)
we need to use only Sigma' as given by the Cholesky decomposition
(which can be thought of as the matrix analogue of the square root of
a number).
We can obviously not use Sigma=chol(V) directly into mvtrnd, as this
matrix is not positive definite. Your suggestion of directly using the
square root over the matrix V does not work either, as the provided
sqrt(V) is again factorized by the mvtrnd function if diagonal items
are non-unit. That is, the resulting covariance relates now to:
nu/(nu-2) * sqrt(V)./(sqrt(diag(sqrt(V))*diag(sqrt(V))))
but not to nu/(nu-2) * V nor nu/(nu-2) * sqrt(V)
I am looking forward to figuring this issue out, so I truly appreciate
your suggestion.
Thank you very much,
Iñigo
On Mon, Nov 26, 2012 at 5:44 PM, Daniel J Sebald <address@hidden> wrote:
>* is typically a variable associated with standard deviation and sigma^2*
>* variance. You've defined V=Sigma * Sigma' which is more along the lines of*
>* variance, not standard deviation.*