5
$\begingroup$

The Gamma distribution is the conjugate prior of Poisson distribution. What about the Truncated Gamma distribution? Is it still the conjugate prior of Poisson distribution?

$\endgroup$
  • $\begingroup$ Thanks for your answer. I thought it as still the conjugate prior but my lecturer told me it wouldn't be the conjugate prior. I am really confused. $\endgroup$ – Suki Hao Mar 8 at 14:53
6
$\begingroup$

A generic if rarely mentioned result about conjugate families is that they are defined in terms of an arbitrary dominating measure $\lambda$. This means that their density wrt this dominating measure is provided by the corresponding exponential family shape $$\exp\{A(\theta)\cdot S_0 -\lambda \psi(\theta)\}$$ but that the dominating measure $\lambda$ may include an arbitrary function $C(\theta)$ of $\theta$ wrt a standard dominating measure like the Lebesgue measure (or any other), in particular the indicator of a particular subset of the parameter space.

In the case of the Poisson example, with likelihood$$\theta^S\exp\{-n\theta\}$$the prior with density against the Lebesgue measure$$\pi(\theta)\propto \theta^{S_0}\exp\{-\lambda\theta\}\mathbb{I}_{(a,b)}(\theta)$$is associated with the posterior$$\pi(\theta)\propto \theta^{S+S_0}\exp\{-(n+\lambda)\theta\}\mathbb{I}_{(a,b)}(\theta)$$which is indeed of the same shape, $S_0$ being replaced with $S+S_0$ and $\lambda$ with $n+\lambda$.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.