hyperpriors for PC prior

[Bowen He]

Hi,

I don't know whether it is highly relative to R-INLA group but I am doing some LGM simulations with known true parameters and realize some values/observations and trying to uncover the true parameters.

I want to apply the penalized complexity priors on those parameters. How can I specify the hyperpriors so that the mean of the PC prior distribution equals to the true parameters of the LGM that generates those observations?

Best,

B

[Helpdesk]

the pc-priors is only worked out for some parameters, like precision,
correlations, and so on. do you mean the fixed effects ? for these you
can use Gaussian priors, which should do well

> --
> You received this message because you are subscribed to the Google
> Groups "R-inla discussion group" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to r-inla-discussion...@googlegroups.com.
> To view this discussion on the web, visit
> https://groups.google.com/d/msgid/r-inla-discussion-group/2bd723f9-5a43-41ad-be24-8a3fdc0cd0c3n%40googlegroups.com
> .

--
Håvard Rue
he...@r-inla.org

[Bowen He]

I want to set up the PC prior for besagproper2 CAR random effect. And I have some true precision parameter value and the lambda value. How can I set up the hyperprior parameters so that my true parameters are the hyperprior's mean?

Thanks,

B