We work on Machine Learning techniques for Quantum Physics, in particular we are developing Neural Quantum States Variational Monte Carlo approaches. We are a team of researchers based in Paris, working between Ecole Polytechnique and College de France.
We are the main contributors behind NetKet, but we also develop additional codes that do not make their way into NetKet directly.
You can find here some of our public works
- NQXPack : A simple way to save and load flax-based NN models. It provides an easy to use solution to save and load NetKet's variational states.
- Neural Quantum States for the Body-Centered Cubic Lattice : Study of large 3D lattice Spin Models with NQS.
- Hybrid Classical-Quantum Spectra : A classical/quantum hybrid algorithm to combine Variational Monte Carlo methods with Quantum Computers to estimate Green's Functions.
- Importance Sampling for Neural Quantum States : How to converge better and faster by sampling elsewhere.
- PTVMC Systematic Study : State-of-the-art implementation of Infidelity optimization for NetKet, as well as infidelity-based dynamics, with autotuning of hyperparameters.
- ConvNext NNQS : Repository accompanying the article "Design principles of deep translationally-symmetric neural quantum states for frustrated magnets", implementing a ConvNext-based NQS in Flax.