Inverse transform in sum-to-zero constraint

[rtrangucci]

I believe the code shown here (link to manual section):

is the Stan code matching the inverse sum-to-zero transform in the next section. Stan code implementing the sum-to-zero transform is here:

vector sum_to_zero_transform(vector x) {
  int N = size(x) - 1;
  vector[N] y;
  y[N] = -x[N+1] * sqrt(1 + 1. / N);
  real sum_w = 0;
  for (n in 1:(N-1)) {
    int i = N - n;
    int i_p_1 = i + 1;
    real w = y[i_p_1] * inv_sqrt(i_p_1 * (i_p_1 + 1));
    sum_w += w;
    y[i] = (sum_w - x[i_p_1]) * sqrt(i_p_1 * i) / i;
  }
  return(y);
}