Contains various optimal sampling results, studied as part of an option project at Centrale Nantes. The repo is organized as follows:
- aspect_theorique.pdf contains the theoretical aspects needed to understand optimal sampling. Also includes some detailed results
- optimal_sampling_shift.ipynb is a simple first approach to functions on intervals [a, b].
- optimal_sampling_approx_lin.ipynb applies optimal sampling theory to gradient descent by linearizing the space for each iteration
$u_t$ and creating a projection of the gradient onto this space. - optimal_sampling_SNN.ipynb applies optimal sampling to shallow neural network (SNN). (warning : not yet finished)