The scaled inverse chi-squared distribution
\psiinv-\chi2(\nu)
\psi
l{W}-1(\psi,\nu)
\nu
This family of scaled inverse chi-squared distributions is linked to the inverse-chi-squared distribution and to the chi-squared distribution:
If
X\sim\psiinv-\chi2(\nu)
X/\psi\siminv-\chi2(\nu)
\psi/X\sim\chi2(\nu)
1/X\sim\psi-1\chi2(\nu)
Instead of
\psi
\tau2=\psi/\nu
\nu\tau2inv-\chi2(\nu)
Scale-inv-\chi2(\nu,\tau2)
In terms of
\tau2
If
X\simScale-inv-\chi2(\nu,\tau2)
X | |
\nu\tau2 |
\siminv-\chi2(\nu)
\nu\tau2 | |
X |
\sim\chi2(\nu)
1/X\sim
1 | |
\nu\tau2 |
\chi2(\nu)
This family of scaled inverse chi-squared distributions is a reparametrization of the inverse-gamma distribution.
Specifically, if
X\sim\psiinv-\chi2(\nu)=Scale-inv-\chi2(\nu,\tau2)
X\sim
|
\psi | |
2 |
\right)=
|
\nu\tau2 | |
2 |
\right)
(E(1/X))
(E(ln(X))
The scaled inverse chi-squared distribution also has a particular use in Bayesian statistics. Specifically, the scaled inverse chi-squared distribution can be used as a conjugate prior for the variance parameter of a normal distribution. The same prior in alternative parametrization is given by the inverse-gamma distribution.
The probability density function of the scaled inverse chi-squared distribution extends over the domain
x>0
f(x;\nu,
| |||||
\tau | ~ |
| ||||||
x1+\nu/2 |
where
\nu
\tau2
F(x;\nu,
| |||||
\tau | , |
\tau2\nu | \right) \left/\Gamma\left( | |
2x |
\nu | |
2 |
\right)\right.
=Q\left( | \nu | , |
2 |
\tau2\nu | |
2x |
\right)
where
\Gamma(a,x)
\Gamma(x)
Q(a,x)
\varphi(t;\nu,\tau2)=
2 | \left( | |||
|
-i\tau2\nut | |
2 |
| ||||
\right) |
K | ||||
|
\left(\sqrt{-2i\tau2\nut}\right),
where
K | ||||
|
(z)
The maximum likelihood estimate of
\tau2
\tau2=
n | |
n/\sum | |
i=1 |
1 | |
xi |
.
The maximum likelihood estimate of
\nu | |
2 |
ln\left( | \nu |
2 |
\right)-\psi\left(
\nu | |
2 |
\right)=
1 | |
n |
n | |
\sum | |
i=1 |
ln\left(xi\right)-ln\left(\tau2\right),
where
\psi(x)
\nu.
\bar{x}=
1 | |
n |
n | |
\sum | |
i=1 |
xi
\nu
\nu | |
2 |
=
\bar{x | |
The scaled inverse chi-squared distribution has a second important application, in the Bayesian estimation of the variance of a Normal distribution.
According to Bayes' theorem, the posterior probability distribution for quantities of interest is proportional to the product of a prior distribution for the quantities and a likelihood function:
p(\sigma2|D,I)\proptop(\sigma2|I) p(D|\sigma2)
The simplest scenario arises if the mean μ is already known; or, alternatively, if it is the conditional distribution of σ2 that is sought, for a particular assumed value of μ.
Then the likelihood term L(σ2|D) = p(D|σ2) has the familiar form
l{L}(\sigma2|D,\mu)=
1 | |
\left(\sqrt{2\pi |
\sigma\right)n} \exp\left[-
| |||||||
2\sigma2 |
\right]
Combining this with the rescaling-invariant prior p(σ2|I) = 1/σ2, which can be argued (e.g. following Jeffreys) to be the least informative possible prior for σ2 in this problem, gives a combined posterior probability
p(\sigma2|D,I,\mu)\propto
1 | |
\sigman+2 |
\exp\left[-
| |||||||
2\sigma2 |
\right]
Gelman et al remark that the re-appearance of this distribution, previously seen in a sampling context, may seem remarkable; but given the choice of prior the "result is not surprising".[1]
In particular, the choice of a rescaling-invariant prior for σ2 has the result that the probability for the ratio of σ2 / s2 has the same form (independent of the conditioning variable) when conditioned on s2 as when conditioned on σ2:
p(\tfrac{\sigma2}{s2}|s2)=p(\tfrac{\sigma2}{s2}|\sigma2)
In the sampling-theory case, conditioned on σ2, the probability distribution for (1/s2) is a scaled inverse chi-squared distribution; and so the probability distribution for σ2 conditioned on s2, given a scale-agnostic prior, is also a scaled inverse chi-squared distribution.
If more is known about the possible values of σ2, a distribution from the scaled inverse chi-squared family, such as Scale-inv-χ2(n0, s02) can be a convenient form to represent a more informative prior for σ2, as if from the result of n0 previous observations (though n0 need not necessarily be a whole number):
p(\sigma2|I\prime,\mu)\propto
1 | ||||
|
\exp\left[-
| |||||||||||||
2\sigma2 |
\right]
p(\sigma2|D,I\prime,\mu)\propto
1 | ||||
|
\exp\left[-
| |||||||||||||
2\sigma2 |
\right]
If the mean is not known, the most uninformative prior that can be taken for it is arguably the translation-invariant prior p(μ|I) ∝ const., which gives the following joint posterior distribution for μ and σ2,
\begin{align} p(\mu,\sigma2\midD,I)&\propto
1 | |
\sigman+2 |
\exp\left[-
| |||||||
2\sigma2 |
\right]\\ &=
1 | |
\sigman+2 |
\exp\left[-
| |||||||
) |
2}{2\sigma2}\right]\exp\left[-
n(\mu-\bar{x | |
) |
2}{2\sigma2}\right] \end{align}
\begin{align} p(\sigma2|D,I) \propto &
1 | |
\sigman+2 |
\exp\left[-
| |||||||
) |
2}{2\sigma2}\right]
infty | |
\int | |
-infty |
\exp\left[-
n(\mu-\bar{x | |
) |
2}{2\sigma2}\right]d\mu\\ = &
1 | |
\sigman+2 |
\exp\left[-
| |||||||
) |
2}{2\sigma2}\right] \sqrt{2\pi\sigma2/n}\\ \propto &(\sigma2)-(n+1)/2 \exp\left[-
(n-1)s2 | |
2\sigma2 |
\right] \end{align}
\scriptstyle{n-1}
\scriptstyle{s2=\sum(xi-\bar{x})2/(n-1)}
X\simScale-inv-\chi2(\nu,\tau2)
kX\simScale-inv-\chi2(\nu,k\tau2)
X\siminv-\chi2(\nu)
X\simScale-inv-\chi2(\nu,1/\nu)
X\simScale-inv-\chi2(\nu,\tau2)
X | |
\tau2\nu |
\siminv-\chi2(\nu)
X\simScale-inv-\chi2(\nu,\tau2)
X\sim
|
\nu\tau2 | |
2 |
\right)