This is the density function for the Yang-Berger prior distribution for a covariance matrix or precision matrix.
This is the k x k positive-definite
covariance matrix or precision matrix for
Application: Continuous Multivariate
Density: p(theta) = 1 / |theta|^(prod (d[j] - d[j-1])), where d are increasing eigenvalues. See equation 13 in Yang and Berger (1994).
Inventor: Yang and Berger (1994)
Notation 1: p(theta) ~ YB
Yang and Berger (1994) derived a least informative prior (LIP) for a covariance matrix or precision matrix. The Yang-Berger (YB) distribution does not have any parameters. It is a reference prior for objective Bayesian inference. The Cholesky parameterization is also provided here.
The YB prior distribution results in a proper posterior. It involves an eigendecomposition of the covariance matrix or precision matrix. It is difficult to interpret a model that uses the YB prior, due to a lack of intuition regarding the relationship between eigenvalues and correlations.
Compared to Jeffreys prior for a covariance matrix, this reference prior encourages equal eigenvalues, and therefore results in a covariance matrix or precision matrix with a better shrinkage of its eigenstructure.
dyangbergerc give the density.
Yang, R. and Berger, J.O. (1994). "Estimation of a Covariance Matrix using the Reference Prior". Annals of Statistics, 2, p. 1195-1211.
1 2 3
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.