scikit-learn locally linear embedding with radius_neighbors

Question

I am trying to apply sklearn.manifold.locally_linear_embedding (LLE) to a set of irregularly sampled data. I would therefore like to find the nearest neighbors within a ball of