Installation

Requirements

Python 3.10 or later
PyTorch 2.0 or later
NumPy 1.24 or later
SciPy 1.10 or later

Basic Installation

Install qvartools in editable mode from the repository root:

pip install -e .

This installs the core dependencies (PyTorch, NumPy, SciPy).

Optional Dependencies

qvartools provides several optional dependency groups:

# PySCF — molecular integrals, RHF, CCSD
pip install -e ".[pyscf]"

# Numba — JIT-compiled acceleration for inner loops
pip install -e ".[numba]"

# GPU — CuPy for GPU-accelerated eigensolvers
pip install -e ".[gpu]"

# Quantum — Qiskit and IBM SQD addon
pip install -e ".[quantum]"

# YAML config support
pip install -e ".[configs]"

# Full (PySCF + Numba)
pip install -e ".[full]"

# Development (pytest, ruff, mypy)
pip install -e ".[dev]"

# Documentation (Sphinx + extensions)
pip install -e ".[docs]"

Using uv

uv pip install -e ".[full,dev]"

Docker (GPU)

A GPU-enabled Docker image is provided for environments with NVIDIA GPUs:

docker build -f Dockerfile.gpu -t qvartools-gpu .
docker run --gpus all --rm -it qvartools-gpu

Verifying the Installation

import qvartools
print(qvartools.__version__)
# 0.0.0

To verify PySCF support:

from qvartools.molecules import list_molecules
print(list_molecules())
# ['beh2', 'c2h4', 'ch4', 'h2', 'h2o', 'lih', 'n2', 'nh3']