0

I'm testing the Dirichlet distribution with pyStan 3.7. I'm getting a warning that alpha has two prior, I don't get why.

data {
  int<lower=1> T;  
  int<lower=1> K;  
  int y[T, K];     
}

parameters {
  simplex[K] theta[T];   
  real alpha; 
}

model {
  alpha ~ cauchy(0, 1);  
  
  for (t in 2:T) {
    theta[t] ~ dirichlet(alpha + theta[t-1]);
  }
}
Bob
  • 10,741
  • 27
  • 89
  • 143
  • Can you share the python code and the error message? It compiles fine for me. There's a bunch of other problems, like you need to make sure that the parameter vector for the dirichlet is strictly positive and you're not using y at all. – LukasNeugebauer Aug 19 '23 at 09:53

0 Answers0