FFT design considerations

Developers
37

Actually, ignore that. While testing your code I found other errors.

The first:

   -------------------------------------------------
     8:        vector[la+lb-1] out;
     9:    
    10:        out = inv_fft(fft(a) .* fft(b)); 
               ^
    11:        return out;
    12:    
   -------------------------------------------------

Ill-typed arguments supplied to assignment operator =: lhs has type vector and rhs has type complex_vector

Is because you’ve declared out (and the function return) to be type vector, but the inv_fft() function returns a complex_vector.

Once I make those changes, it compiles fine for me:

modcode  = "
functions {
  complex_vector conv(vector a, vector b) {
  
      int la = num_elements(a);
      int lb = num_elements(b);
  
      complex_vector[la+lb-1] out;
  
      out = inv_fft(fft(a) .* fft(b)); 
      return out;
  
  }
}
data { real y_mean; } parameters { real y; } model { y ~ normal(y_mean, 0); }
"
mod <- cmdstan_model(write_stan_file(modcode))
1 Like