AutoHF is an automatic-differention-based lattice model Hartree-Fock solver.

In contrast to typical self-consistent-field Hartree-Fock implementations, AutoHF leverages JAX and gradient-descend algorithms to find the Hartree-Fock solution variationally. This approach can be more robust against both saddle-points or local minima and allows flexible user-supplied generalizations beyond the standard Hartree-Fock ansatz.

Getting started

You can install autohf via pip

pip install sfqmc-autohf

For more instractions on how to install and run AutoHF, see the Documentation. Also have a look at the scripts in the examples/ folder.

Citation

If you use AutoHF in a scientific publication please consider citing the paper

@misc{autohf_arxiv,
  doi = {10.48550/ARXIV.2607.14263},
  url = {https://arxiv.org/abs/2607.14263},
  author = {Levy,  Ryan and Eskridge,  Brandon and Weber,  Lukas and Morales,  Miguel A. and Zhang,  Shiwei},
  title = {AutoHF: a general Hartree-Fock solver utilizing direct energy minimization with automatic differentiation},
  publisher = {arXiv},
  year = {2026}
}