Specifying prior parameters from code?

[AntonioMartini]

Is there a way of specifying priors parameters from code/file? For example, in the code below, rather than writing dnorm(0,100) we may want to specify the prior mean and variance as a result of a calculation. Is there a way of feeding such computed prior parameters to the model?

`m3 <- map2stan(

  alist( 

    s ~ student_t( v+1, mu, sigma ),

    mu <-  k1*v1 + k2*v2 

    k1  ~ dnorm(0,100),        
    k2  ~ dnorm(0,100),        

    sigma ~ dcauchy(0,0.5),

    v ~ dexp(30)

  ), 
  data=db, 
  constraints=list(sigma="lower=0", v="lower=0"),
  chains=12, 
  cores=12,
  WAIC=FALSE )`

[rmcelreath]

This seems to work:

m <- map2stan(
    alist(
        k ~ dnorm( k_mu , k_sd )
    ),
    data=list(k_mu=c(0,0),k_sd=c(1,1)) )

It's a little awkward, since map2stan expects data to have more than one case, I had to insert two values into k_mu and k_sd in the data list. Stan did not seem to care.

In any event, the above is not tested or supported behavior. So be careful.

[AntonioMartini]

Thanks that's very helpful.