Quimb

What Quimb is

Quimb is a free Python library for quantum information and many-body calculations, with particular emphasis on tensor networks. Its tensor module lets users build and manipulate tensor-network graphs, contract and optimize them, and draw their structure. It includes methods for MPS, PEPS, and MERA networks, plus quantum-circuit algorithms such as DMRG and TEBD. The core module supports quantum calculations with NumPy arrays or SciPy sparse matrices, including finding ground and excited states and carrying out time evolution. Quimb can also compute entanglement measures and stochastically estimate trace functions. Through autoray, tensor calculations can use backend array libraries including JAX and PyTorch. Documentation includes guides to quantum circuits and importing OpenQASM circuits. Quimb is distributed through PyPI and conda-forge, and its installation guide recommends Pixi. The documented core requirements include Python 3.11 or newer, NumPy, SciPy, Numba, cytoolz, tqdm, and psutil. Optional SLEPc and PETSc packages, together with MPI dependencies, support accelerated and optionally distributed partial eigensolving, SVD, and exponentiation. The project is published in the Journal of Open Source Software, and its documentation asks researchers to cite it when useful in research.

Who it suits

Quimb suits researchers and developers doing quantum information or many-body calculations, especially those working with tensor networks. It also fits users who want Python tools for quantum circuits, exact calculations, or entanglement analysis and can work with its documented Python environment.

Getting started

1. Install Python 3.11 or newer.
2. Choose PyPI or conda-forge as the distribution route.
3. Follow the installation guide, which recommends Pixi.
4. Install the documented core requirements: NumPy, SciPy, Numba, cytoolz, tqdm, and psutil.
5. Add optional SLEPc and PETSc packages with MPI dependencies if using the described acceleration options.

Verdict

Choose Quimb if you need a free Python library centered on tensor-network work or quantum information calculations. Its documented setup requires Python 3.11 or newer, and the optional acceleration route involves additional packages and MPI dependencies.