2021-exp-ICM Jupyter notebook with Julia code to reproduce Figure 2 and 3 from the paper Gradient Descent on Infinitely Wide Neural Networks: Global Convergence & Generalization, F. Bach and L. Chizat