The formula for calculating the WLOP algorithm

[wanghao-98]

The formula in the paper is as above, but the code is consistent with the code when calculating the local density weights of the original point cloud and the sampled point cloud.
However, when iteratively solving the new coordinates of x, the formula is inconsistent by parts, and the source code of CGAL is consistent with this codebase;
Is there any optimization in it?
1、double iradius16 = -(double)4.0 / radius2;论文中为16

2、weight *= original_densities[index[j]]; The weights are not divided by distance
3、weight = std::exp(dist2 * iradius16) * std::pow(double(1.0) / dist ,2); // L1
if (!is_sample_densities_empty)
{
weight *= sample_densities[index[j]];
}

Weight calculations are inconsistent
I hope the big guy can guide me.