Order of execution of different Stan blocks

Modeling
1

Hi there, I have a question related to the sequence that different blocks in Stan are executed. Lets say I have the following toy example . I want to 1) get sample a, 2) apply a function on it to generate a_transformed, 3) use a_transformed as a parameter of distribution distr_2.

I wonder is this a valid workflow? It runs but I am not sure in terms of the order that model block and transformed_parameters block run, does it do the thing I expect and is the computation graph created as I expect? I mean does Stan figure out this dependency properly?

    data {

    }
    parameters {
        real<lower=0.0, upper=1.0> sensitivity;
        real<lower=0.0, upper=1.0> sensitivity_new;
        
    }
    transformed parameters {
        real sensitivity_transfomed = sensitivity * 5 + 2;
    }
    model {
      sensitivity ~ beta(5.0, 2.0);
      sensitivity_new ~ beta(sensitivity_transfomed, 3.0);
    }

Thanks for your help and answering my noob question.

2

Hi, it’s

transformed parameters{}

not transformed_parameters{}.

3

thanks for pointing this out. However my question still stands :)

4

Ah, I guess it looks ok, doesn’t it work as intended? Please provide a minimum working example :)

Richard

5

done

6

Hi

Here you have the order explained :)

2 Likes