Why Jordan algebras are natural in statistics:quadratic regression implies Wishart distributions
Preprint
Abstract
If the space $\mathcal{Q}$ of quadratic forms in $\mathbb{R}^n$ is splitted in a direct sum $\mathcal{Q}_1\oplus...\oplus \mathcal{Q}_k$ and if $X$ and $Y$ are independent random variables of $\mathbb{R}^n$, assume that there exist a real number $a$ such that $E(X|X+Y)=a(X+Y)$ and real distinct numbers $b_1,...,b_k$ such that $E(q(X)|X+Y)=b_iq(X+Y)$ for any $q$ in $\mathcal{Q}_i.$ We prove that this happens only when $k=2$, when $\mathbb{R}^n$ can be structured in a Euclidean Jordan algebra and when $X$ and $Y$ have Wishart distributions corresponding to this structure.
Keywords
All Related Versions
This publication has 9 references indexed in Scilit:
- Laplace transforms which are negative powers of quadratic polynomialsTransactions of the American Mathematical Society, 2008
- Quadratic and inverse regressions for Wishart distributionsThe Annals of Statistics, 1998
- The Lukacs-Olkin-Rubin characterization of Wishart distributions on symmetric conesThe Annals of Statistics, 1996
- Analysis on Symmetric Cones1994
- Characterization of the Jorgensen set in generalized linear modelsTEST, 1994
- THE DAVIDSON‐KENDALL PROBLEM AND RELATED RESULTS ON THE STRUCTURE OF THE WISHART DISTRIBUTIONAustralian Journal of Statistics, 1988
- Covariance Hypotheses Which are Linear in Both the Covariance and the Inverse CovarianceThe Annals of Statistics, 1988
- Invariant Normal ModelsThe Annals of Statistics, 1975
- Statistical Analysis Based on a Certain Multivariate Complex Gaussian Distribution (An Introduction)The Annals of Mathematical Statistics, 1963