PyCC

A Python-based electronic-structure package centered on coupled cluster, built on a shared Wavefunction base that also hosts Hartree-Fock, MP2, and CI. Current capabilities include:

• RHF-/UHF-/ROHF-based CCD, CC2, CCSD, CCSD(T), and CC3 energies, via either a spin-adapted (closed-shell RHF) or a spin-orbital formalism; the latter enables the open-shell (UHF/ROHF) correlated methods
• Triples drivers for various approximate triples methods
• One- and two-electron (reduced) density matrices for CC2 and CCSD (with CC3 and (T) contributions), and CC one-electron properties (e.g. dipole moments)
• RHF-EOM-CCSD excitation energies
• Linear response functions (dynamic polarizability, optical rotation) for both the spin-adapted (closed-shell) and spin-orbital references
• Real-time (RT) CC2, CCSD, and CC3 with a selection of integrators
• Local (PAO, PNO, PNO++) CCSD energies, and local RT-CC
• MP2 (MPwfn) and CISD/CID (CIwfn) energies (closed- and open-shell)
• Analytic MO-basis derivative properties for Hartree-Fock (HFwfn) and MP2 (MPwfn): energy gradient, static dipole polarizability, nuclear Hessian (force-constant matrix), atomic polar tensors (APTs / dipole derivatives, in both the length and velocity gauge), and atomic axial tensors (AATs) -- the building blocks for IR and VCD spectra. Both spin paths (spin-adapted closed-shell RHF and spin-orbital), all-electron and frozen-core; the MP2 second derivatives come via two independent routes (explicit-derivative and 2n+1)
• A uniform property interface -- pycc.dipole, pycc.gradient, pycc.polarizability, pycc.hessian, pycc.apt, pycc.aat -- that dispatches on wavefunction type and returns each property's nuclear/reference/correlation decomposition as a PropertyComponents dataclass (.total, .nuclear, .reference, .correlation, .electronic)
• GPU implementations for multiple methods
• Single- and mixed-precision arithmetic

Future plans:

• Quadratic response functions (in development)
• CC2 and CC3 excited states
• Analytic CC gradients

This repository is currently under development. To do a developmental install, download this repository and type pip install -e . in the repository directory.

This package requires the following:

• psi4
• numpy
• opt_einsum
• scipy

Optional packages:

• pytorch

Documentation

The docs/ directory holds the Sphinx documentation — a getting-started guide with runnable examples and an API reference. GitHub renders these .rst files in the browser; to build the full HTML site locally, install Sphinx (and the Read the Docs theme) and run make html from docs/.

Authors

T. Daniel Crawford, Benjamin G. Peyton, Zhe Wang, Jose Madriaga, Aparna Krishnan

Copyright

Copyright (c) 2026, T. Daniel Crawford

Acknowledgements

Project structure based on the MolSSI's Computational Molecular Science Python Cookiecutter Version 1.5.