| TeNPy | |
|---|---|
| Name | TeNPy |
| Developer | {Max Planck Institute for the Physics of Complex Systems researchers and community contributors} |
| Released | 2015 |
| Programming language | Python |
| Operating system | Cross-platform |
| Genre | Tensor network library |
| License | BSD license |
TeNPy
TeNPy is an open-source Python library for tensor-network numerics, designed to implement and experiment with Matrix product states (MPS) and related methods in low-dimensional quantum many-body systems. It matters in the context of Quantum physics because it provides researchers and students with accessible, well-tested tools for studying strongly correlated systems, quantum phase transitions, and entanglement structure with an emphasis on reproducibility and community governance.
TeNPy (Tensor Network Python) originated from research groups focused on condensed matter and quantum information, and it consolidates algorithms for variational optimisation, time evolution, and observable calculation within a consistent framework. The project connects to foundational concepts such as Density matrix renormalization group (DMRG), Matrix product operators (MPO), and entanglement measures, enabling studies of models like the Heisenberg model, Hubbard model, and Transverse-field Ising model. TeNPy is used in academic environments including the Max Planck Institute for the Physics of Complex Systems, University of Vienna, and multiple physics and computational research groups across universities and national laboratories. The library foregrounds equitable access by being permissively licensed (BSD license) and encouraging transparent, reproducible workflows for students and researchers worldwide.
TeNPy's core abstractions implement Matrix product state and Matrix product operator representations with utilities for canonical forms, truncation, and symmetry handling. Key features include: - Native support for U(1) and Z2 symmetries enabling sector-based block-sparse tensors, which accelerates simulations of conserved-charge models like the Fermi–Hubbard model. - Tools for constructing model Hamiltonians (spin chains, fermionic chains) and mapping lattice geometries. - Variational DMRG solvers, time-evolution schemes (TEBD, time-dependent variational principle), and routines to compute static observables, correlation functions, and entanglement entropies. - IO, plotting, and checkpointing utilities for reproducibility and integration with scientific stacks such as NumPy and SciPy.
The design emphasizes modularity: tensor classes, linear algebra backends, and algorithm modules can be combined to prototype new methods or extend existing ones. By integrating with the broader Python ecosystem, TeNPy can be used alongside Jupyter Notebooks for pedagogy and GitHub for collaborative development.
TeNPy implements both standard and advanced tensor-network algorithms. The primary implemented algorithms include: - Variational Density matrix renormalization group for ground states with single- and two-site updates, support for targeting excited states, and energy variance estimation. - Time-evolution via Time-evolving block decimation (TEBD) and the Time-dependent variational principle (TDVP), including adaptive step-size and bond-dimension control. - Construction and compression of Matrix product operators using local operator libraries, Jordan–Wigner strings for fermions, and MPO-MPS contractions optimized for block-sparse symmetry tensors. - Entanglement diagnostics such as Schmidt decompositions, von Neumann entropy calculations, and entanglement spectrum extraction.
Implementations are optimized for practical use: efficient tensor contractions use NumPy arrays and benefit from BLAS/LAPACK acceleration; symmetry-aware routines reduce memory and CPU cost; and checkpointing supports long-running runs on cluster environments like SLURM-managed systems. Algorithmic choices are informed by literature including seminal works by Steven R. White (DMRG), Guifre Vidal (TEBD/entanglement), and modern developments in tensor-network theory.
TeNPy is applied across problems in condensed matter and quantum information: - Ground-state and low-energy spectroscopy of spin chains (e.g., XXZ model), ladders, and quasi-one-dimensional materials. - Strongly correlated fermions in one dimension: studies of the Hubbard model and Luttinger-liquid behaviour. - Non-equilibrium dynamics, quenches, and transport using TEBD/TDVP to study thermalization and many-body localization, connecting to topics explored at conferences such as the APS March Meeting and SC Conference presentations. - Calculation of topological invariants, symmetry-protected phases, and string-order parameters relevant to topological order and symmetry-protected topological phases (SPT). - Benchmarks for quantum simulation platforms and comparisons with tensor-network inspired algorithms used in quantum computing research.
TeNPy’s user base includes academic researchers publishing in journals like Physical Review Letters, Physical Review B, and Journal of Statistical Mechanics: Theory and Experiment.
Performance in TeNPy balances Python usability with numerical efficiency: critical inner loops rely on optimized linear algebra libraries (BLAS/LAPACK) and block-sparse routines for symmetric tensors. Scalability is primarily horizontal over parameter sweeps and ensemble runs; single-MPS runs are limited by bond dimension and available memory but can be distributed via task scheduling on HPC clusters. Reproducibility is emphasized: versioned releases, example notebooks, and explicit random-seed handling facilitate replication of published results. Community-driven benchmarks compare TeNPy with other implementations such as ITensor and custom Fortran/C++ codes to assess accuracy vs. runtime trade-offs.
TeNPy is developed openly on platforms like GitHub with contributions from researchers and students worldwide. Governance encourages inclusive contribution practices and maintains documentation, tutorials, and issue trackers. The project’s BSD license promotes reuse in academic and commercial settings while preserving attribution. TeNPy has been used in coursework, workshops, and summer schools that aim to broaden access to computational quantum physics, aligning software development with goals of equity in scientific training and global participation. Active community members include physicists from institutions such as the Max Planck Society, University of Cologne, and research groups focused on tensor networks and quantum materials.
Category:Computational physics software Category:Quantum many-body theory Category:Free scientific software