Quax | order derivatives of popular electronic structure methods | Machine Learning library

Quax is a Python library typically used in Artificial Intelligence, Machine Learning applications. Quax has no bugs, it has no vulnerabilities, it has build file available, it has a Permissive License and it has low support. You can download it from GitHub.

You have found Quax. The paper outlining this work was just recently published. This library supports a simple and clean API for obtaining higher-order energy derivatives of electronic structure computations such as Hartree-Fock, second-order Møller-Plesset perturbation theory (MP2), and coupled cluster with singles, doubles, and perturbative triples excitations [CCSD(T)]. Whereas most codes support only analytic gradient and occasionally Hessian computations, this code can compute analytic derivatives of arbitrary order. We use JAX for automatically differentiating electronic structure computations. The code can be easily extended to support other methods, for example using the guidance offered by the Psi4Numpy project. If you are interested in obtaining electronic energy derivatives with Quax, but are wary of and/or not familiar with the concept of automatic differentiation, we recommend this video for a brief primer. We should also note this project is mostly intended as an experimental proof-of-concept. While it can be used for research applications, users should always take steps to verify the accuracy of the results, either by checking energies and derivatives against standard electronic structure codes or by using finite differences. Generally, if the energy and gradient are correct, higher order derivatives are most likely correct to a high degree of numerical precision. Note however the caveat (described below) that systems with highly degenerate orbitals will likely be numerically unstable at high derivative orders.