Using STAN function inside user defined function

Asked Nov 23 '16 at 22:04

Active Nov 23 '16 at 22:14

Viewed 1,570 times

The likelihood function for my data (m) is like below:

          "dexp(m,beta) * pnorm(m,mu,sigma)".

The user defined loglikelihood function that I wrote inside my STAN code is like:

functions{
    real loglikelihood(int N,
    real mu,
    real sigma,
    real beta,
    real[] m
    ){
      real a[N];
      real b[N];
      real c[N];
      real final;
      for(i in 1:N){

         a[i]<-  exponential_log(m[i],beta);
         b[i]<-  normal_cdf_log( m[i],  mu,  sigma);
         c[i]<-  a[i]+b[i];
     }

     final<- sum(c);
     return(final);
  }
}

. I want to know if I am doing it right? Can I use the STAN functions like "exponential_log" and "normal_cdf_log" inside the user defined function?

edited Nov 23 '16 at 22:14

asked Nov 23 '16 at 22:04

Geotas

1

Yes, although you do not necessarily need a user-defined Stan function to do so. It would also work to just do `m ~ exponential(beta); target += normal_lcdf(m | mu, sigma);` in the model block. – Ben Goodrich Nov 23 '16 at 22:23
@BenGoodrich Thanks Ben.:) – Geotas Nov 23 '16 at 22:42
We've changed function names so it's `exponential_lpdf()` and `normal_lcdf()` for the ones you used. You can remove final and `c`and just use `return(sum(a) + sum(b))`. – Bob Carpenter Nov 25 '16 at 23:40
@BobCarpenter Hi Bob, thanks. Are those new functions in your newest user manual? – Geotas Nov 27 '16 at 20:32
Yes, every function available in Stan is in the manual. – Bob Carpenter Dec 04 '16 at 20:04

Using STAN function inside user defined function

0 Answers0