Reassignment if vs if-else?

Modeling
1

Is there a difference between reassigning like:

data {
  int is_fancier_model;
...
}
...
transformed parameters {
  mu = beta_0;
  if (is_fancier_model) {
     mu = beta_0 + beta_1*x;
  }
}

vs using if-else like:

data {
  int is_fancier_model;
...
}
...
transformed parameters {
  if (is_fancier_model) {
     mu = beta_0 + beta_1*x;
  } else {
     mu = beta_0;
  }
}
2

This might not be what you’re asking, but the better option is to conditionally aggregate:

transformed parameters {
  vector[N] mu = beta_0;
  if (is_fancier_model) {
     mu += beta_1 * x;
  }
}
3

Depending on your C++ optimization setting (e.g. -O2, etc) the code most likely ends up being executed identically in the end. I just did a quick test and all three styles above (when used in pure C++, but the Stan translation is fairly direct here) yield the same assembly code at O2

4

Does Stanc3 still initialise variables first with NaNif a declare-and-define is used? Or would the variable be initialised with the assigned value?

On my phone so can’t check myself

5

If you use stanc’s optimizations it should remove the NaN initialization.

This means the style which doesn’t use else would have benefits if the type of mu is a vector or matrix, since gcc or clang’s optimizer most likely wouldn’t delete the object creation.

6

Neat, thanks!