control.group(model="ar1")

[Ruby Ji]

Dear INLA-group,

I have a question about using control.group. I set the prior for ar1-rho as Gaussian with zero mean and precision of 0.2 as follows:  

rho_cp <- list(theta=list(prior='normal', param=c(0, 0.2))) 

In the model formula, I used:

form <- y ~ ... +

f(s, model=spde, group=s.group, control.group=list(model="ar1", hyper=rho_cp, initial=1, scale.model=TRUE, fixed=TRUE))

I am wondering if this setup allows me to manually set rho to 1. Does it achieve the same effect as using rw1 in the control group? When using rw1, I encountered continuous messages about "GMRFLib_factorise_sparse_matrix_TAUCS(); failing to factorize Q", which never resolved. Could you provide any insights or suggestions on this issue?

Thank you for your assistance.

Ruby

[Finn Lindgren]

Since the ar1 models is parameterised with the marginal precision, setting rho close to 1  would give a random constant, not a random walk, so it’s unlikely to be what you want.

Finn

On 15 Nov 2024, at 03:10, Ruby Ji <crysta...@gmail.com> wrote:

Dear INLA-group,

--
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, visit https://groups.google.com/d/msgid/r-inla-discussion-group/0d8ffb0a-bbdf-4416-bf95-5ab8d70db8d5n%40googlegroups.com.

[rujia bi]

I see. I will try more strict prior to see if I can solve the rw1 issue.

On Nov 14, 2024, at 11:34 PM, Finn Lindgren <finn.l...@gmail.com> wrote:

Since the ar1 models is parameterised with the marginal precision, setting rho close to 1  would give a random constant, not a random walk, so it’s unlikely to be what you want.