Time-dependent Neural Quantum Galerkin

This repository provides the accompanying code for the paper "Time-dependent Neural Galerkin Method for Quantum Dynamics" (https://arxiv.org/abs/2412.11778). The code is based on jax, flax and NetKet.

Content of the repository

• tnqg: core implementation of the time-dependent Neural Quantum Galerkin (t-NQG) method. The folder includes the following subfolders:
  • operator: files to compute the global L2 loss function.
  • models: Galerkin ansatz implemented in flax.
  • sampling: operations to perform Monte Carlo sampling.
  • utils: utility functions.
• scripts: executable scripts for reproducing the numerical results presented in the paper.
• data: data used in and generated by the study, organized into the subfolders:
  • benchmark: exact diagonalization results used for benchmarking.
  • qmc: finite-temperature Quantum Monte Carlo (QMC) data.
  • tnqg: results obtained with the t-NQG method.
  • tvmc: simulation data from time-dependent Variational Monte Carlo (t-VMC).

Installation

This package can be install directly from GitHub as:

git clone https://github.com/cqsl/tnqg.git
pip install -e ./tnqg

How to cite

If you use tnqg in your work, please consider citing it as:

@software{tnqg,
    author = {Sinibaldi, Alessandro},
    title = {tnqg package},
    url = {https://github.com/cqsl/tnqg},
    doi = {10.5281/zenodo.18233288},
    version = {0.0.1},
    year = {2026}
    }