This repository contains the code of the experiments presented in https://arxiv.org/abs/2509.10393.

All the experiments with the setting and figures productions are contained in the folder 'experiments'. In the folder 'modules' are contained in .py file all the methods used for the experiments. The file kgd_functions.py contains the implementation of the gradient kernel, KGD, $\mathcal{J}$ (computed with KDE estimation) and some reproducing kernels (gaussian, imq, recommended kernel).