Matrix t-distribution

From HandWiki
Matrix t
Notation [math]\displaystyle{ {\rm T}_{n,p}(\nu,\mathbf{M},\boldsymbol\Sigma, \boldsymbol\Omega) }[/math]
Parameters

[math]\displaystyle{ \mathbf{M} }[/math] location (real [math]\displaystyle{ n\times p }[/math] matrix)
[math]\displaystyle{ \boldsymbol\Omega }[/math] scale (positive-definite real [math]\displaystyle{ p\times p }[/math] matrix)
[math]\displaystyle{ \boldsymbol\Sigma }[/math] scale (positive-definite real [math]\displaystyle{ n\times n }[/math] matrix)

[math]\displaystyle{ \nu }[/math] degrees of freedom
Support [math]\displaystyle{ \mathbf{X} \in\mathbb{R}^{n\times p} }[/math]
PDF

[math]\displaystyle{ \frac{\Gamma_p\left(\frac{\nu+n+p-1}{2}\right)}{(\pi)^\frac{np}{2} \Gamma_p\left(\frac{\nu+p-1}{2}\right)} |\boldsymbol\Omega|^{-\frac{n}{2}} |\boldsymbol\Sigma|^{-\frac{p}{2}} }[/math]

[math]\displaystyle{ \times \left|\mathbf{I}_n + \boldsymbol\Sigma^{-1}(\mathbf{X} - \mathbf{M})\boldsymbol\Omega^{-1}(\mathbf{X}-\mathbf{M})^{\rm T}\right|^{-\frac{\nu+n+p-1}{2}} }[/math]
CDF No analytic expression
Mean [math]\displaystyle{ \mathbf{M} }[/math] if [math]\displaystyle{ \nu + p - n \gt 1 }[/math], else undefined
Mode [math]\displaystyle{ \mathbf{M} }[/math]
Variance [math]\displaystyle{ \frac{\boldsymbol\Sigma \otimes \boldsymbol\Omega}{\nu-2} }[/math] if [math]\displaystyle{ \nu \gt 2 }[/math], else undefined
CF see below

In statistics, the matrix t-distribution (or matrix variate t-distribution) is the generalization of the multivariate t-distribution from vectors to matrices.[1] The matrix t-distribution shares the same relationship with the multivariate t-distribution that the matrix normal distribution shares with the multivariate normal distribution.[clarification needed] For example, the matrix t-distribution is the compound distribution that results from sampling from a matrix normal distribution having sampled the covariance matrix of the matrix normal from an inverse Wishart distribution.[citation needed][2]

In a Bayesian analysis of a multivariate linear regression model based on the matrix normal distribution, the matrix t-distribution is the posterior predictive distribution.

Definition

For a matrix t-distribution, the probability density function at the point [math]\displaystyle{ \mathbf{X} }[/math] of an [math]\displaystyle{ n\times p }[/math] space is

[math]\displaystyle{ f(\mathbf{X} ; \nu,\mathbf{M},\boldsymbol\Sigma, \boldsymbol\Omega) = K \times \left|\mathbf{I}_n + \boldsymbol\Sigma^{-1}(\mathbf{X} - \mathbf{M})\boldsymbol\Omega^{-1}(\mathbf{X}-\mathbf{M})^{\rm T}\right|^{-\frac{\nu+n+p-1}{2}}, }[/math]

where the constant of integration K is given by

[math]\displaystyle{ K = \frac{\Gamma_p\left(\frac{\nu+n+p-1}{2}\right)}{(\pi)^\frac{np}{2} \Gamma_p\left(\frac{\nu+p-1}{2}\right)} |\boldsymbol\Omega|^{-\frac{n}{2}} |\boldsymbol\Sigma|^{-\frac{p}{2}}. }[/math]

Here [math]\displaystyle{ \Gamma_p }[/math] is the multivariate gamma function.

The characteristic function and various other properties can be derived from the generalized matrix t-distribution (see below).

Generalized matrix t-distribution

Generalized matrix t
Notation [math]\displaystyle{ {\rm T}_{n,p}(\alpha,\beta,\mathbf{M},\boldsymbol\Sigma, \boldsymbol\Omega) }[/math]
Parameters

[math]\displaystyle{ \mathbf{M} }[/math] location (real [math]\displaystyle{ n\times p }[/math] matrix)
[math]\displaystyle{ \boldsymbol\Omega }[/math] scale (positive-definite real [math]\displaystyle{ p\times p }[/math] matrix)
[math]\displaystyle{ \boldsymbol\Sigma }[/math] scale (positive-definite real [math]\displaystyle{ n\times n }[/math] matrix)
[math]\displaystyle{ \alpha \gt (p-1)/2 }[/math] shape parameter

[math]\displaystyle{ \beta \gt 0 }[/math] scale parameter
Support [math]\displaystyle{ \mathbf{X} \in\mathbb{R}^{n\times p} }[/math]
PDF

[math]\displaystyle{ \frac{\Gamma_p(\alpha+n/2)}{(2\pi/\beta)^\frac{np}{2} \Gamma_p(\alpha)} |\boldsymbol\Omega|^{-\frac{n}{2}} |\boldsymbol\Sigma|^{-\frac{p}{2}} }[/math]

[math]\displaystyle{ \times \left|\mathbf{I}_n + \frac{\beta}{2}\boldsymbol\Sigma^{-1}(\mathbf{X} - \mathbf{M})\boldsymbol\Omega^{-1}(\mathbf{X}-\mathbf{M})^{\rm T}\right|^{-(\alpha+n/2)} }[/math]
CDF No analytic expression
Mean [math]\displaystyle{ \mathbf{M} }[/math]
Variance [math]\displaystyle{ \frac{2(\boldsymbol\Sigma \otimes \boldsymbol\Omega)}{\beta(2\alpha-p-1)} }[/math]
CF see below

The generalized matrix t-distribution is a generalization of the matrix t-distribution with two parameters α and β in place of ν.[3]

This reduces to the standard matrix t-distribution with [math]\displaystyle{ \beta=2, \alpha=\frac{\nu+p-1}{2}. }[/math]

The generalized matrix t-distribution is the compound distribution that results from an infinite mixture of a matrix normal distribution with an inverse multivariate gamma distribution placed over either of its covariance matrices.

Properties

If [math]\displaystyle{ \mathbf{X} \sim {\rm T}_{n,p}(\alpha,\beta,\mathbf{M},\boldsymbol\Sigma, \boldsymbol\Omega) }[/math] then[citation needed]

[math]\displaystyle{ \mathbf{X}^{\rm T} \sim {\rm T}_{p,n}(\alpha,\beta,\mathbf{M}^{\rm T},\boldsymbol\Omega, \boldsymbol\Sigma). }[/math]

The property above comes from Sylvester's determinant theorem:

[math]\displaystyle{ \det\left(\mathbf{I}_n + \frac{\beta}{2}\boldsymbol\Sigma^{-1}(\mathbf{X} - \mathbf{M})\boldsymbol\Omega^{-1}(\mathbf{X}-\mathbf{M})^{\rm T}\right) = }[/math]
[math]\displaystyle{ \det\left(\mathbf{I}_p + \frac{\beta}{2}\boldsymbol\Omega^{-1}(\mathbf{X}^{\rm T} - \mathbf{M}^{\rm T})\boldsymbol\Sigma^{-1}(\mathbf{X}^{\rm T}-\mathbf{M}^{\rm T})^{\rm T}\right) . }[/math]

If [math]\displaystyle{ \mathbf{X} \sim {\rm T}_{n,p}(\alpha,\beta,\mathbf{M},\boldsymbol\Sigma, \boldsymbol\Omega) }[/math] and [math]\displaystyle{ \mathbf{A}(n\times n) }[/math] and [math]\displaystyle{ \mathbf{B}(p\times p) }[/math] are nonsingular matrices then[citation needed]

[math]\displaystyle{ \mathbf{AXB} \sim {\rm T}_{n,p}(\alpha,\beta,\mathbf{AMB},\mathbf{A}\boldsymbol\Sigma\mathbf{A}^{\rm T}, \mathbf{B}^{\rm T}\boldsymbol\Omega\mathbf{B}) . }[/math]

The characteristic function is[3]

[math]\displaystyle{ \phi_T(\mathbf{Z}) = \frac{\exp({\rm tr}(i\mathbf{Z}'\mathbf{M}))|\boldsymbol\Omega|^\alpha}{\Gamma_p(\alpha)(2\beta)^{\alpha p}} |\mathbf{Z}'\boldsymbol\Sigma\mathbf{Z}|^\alpha B_\alpha\left(\frac{1}{2\beta}\mathbf{Z}'\boldsymbol\Sigma\mathbf{Z}\boldsymbol\Omega\right), }[/math]

where

[math]\displaystyle{ B_\delta(\mathbf{WZ}) = |\mathbf{W}|^{-\delta} \int_{\mathbf{S}\gt 0} \exp\left({\rm tr}(-\mathbf{SW}-\mathbf{S^{-1}Z})\right)|\mathbf{S}|^{-\delta-\frac12(p+1)}d\mathbf{S}, }[/math]

and where [math]\displaystyle{ B_\delta }[/math] is the type-two Bessel function of Herz[clarification needed] of a matrix argument.

See also

Notes

  1. Zhu, Shenghuo and Kai Yu and Yihong Gong (2007). "Predictive Matrix-Variate t Models." In J. C. Platt, D. Koller, Y. Singer, and S. Roweis, editors, NIPS '07: Advances in Neural Information Processing Systems 20, pages 1721–1728. MIT Press, Cambridge, MA, 2008. The notation is changed a bit in this article for consistency with the matrix normal distribution article.
  2. Gupta, Arjun K and Nagar, Daya K (1999). Matrix variate distributions. CRC Press. pp. Chapter 4. 
  3. 3.0 3.1 Iranmanesh, Anis, M. Arashi and S. M. M. Tabatabaey (2010). "On Conditional Applications of Matrix Variate Normal Distribution". Iranian Journal of Mathematical Sciences and Informatics, 5:2, pp. 33–43.

External links