Home › Scientific Tools DB › 3. DMFT & MANY-BODY › 3.4 Tensor Networks

TeNPy

TeNPy is a comprehensive Python library for tensor network algorithms, particularly focused on one-dimensional quantum systems and DMRG (Density Matrix Renormalization Group). Developed with emphasis on accessibility and ease of use, TeN…

3. DMFT & MANY-BODY 3.4 Tensor Networks VERIFIED 1 paper
Back to Mind Map Official Website

Overview

TeNPy is a comprehensive Python library for tensor network algorithms, particularly focused on one-dimensional quantum systems and DMRG (Density Matrix Renormalization Group). Developed with emphasis on accessibility and ease of use, TeNPy provides a pure Python implementation of state-of-the-art tensor network methods with excellent documentation and a clean, modular design. The library is ideal for researchers and students learning tensor network methods while maintaining production-quality ca

Reference Papers (1)

Full Documentation

Official Resources

  • Homepage: https://tenpy.readthedocs.io/
  • Documentation: https://tenpy.readthedocs.io/en/latest/
  • Source Repository: https://github.com/tenpy/tenpy
  • License: GNU General Public License v3.0

Overview

TeNPy is a comprehensive Python library for tensor network algorithms, particularly focused on one-dimensional quantum systems and DMRG (Density Matrix Renormalization Group). Developed with emphasis on accessibility and ease of use, TeNPy provides a pure Python implementation of state-of-the-art tensor network methods with excellent documentation and a clean, modular design. The library is ideal for researchers and students learning tensor network methods while maintaining production-quality capabilities.

Scientific domain: Tensor networks, DMRG, 1D quantum systems
Target user community: Python users, researchers, educators, tensor network learners

Theoretical Methods

  • Density Matrix Renormalization Group (DMRG)
  • Matrix Product States (MPS)
  • Matrix Product Operators (MPO)
  • Time evolution (TEBD, TDVP)
  • Infinite-size systems (iDMRG)
  • Finite-temperature MPS
  • Purification methods
  • Variational algorithms
  • Entanglement measures

Capabilities (CRITICAL)

Category: Open-source tensor network library (Python)

  • DMRG (finite and infinite)
  • Ground and excited states
  • Time evolution (TEBD, TDVP)
  • Finite-temperature methods
  • Built-in models
  • Custom Hamiltonians
  • Correlation functions
  • Entanglement analysis
  • Conserved quantum numbers
  • Pure Python implementation
  • NumPy/SciPy based
  • Jupyter-friendly
  • Production quality
  • Educational value

Sources: Official documentation, GitHub, publications

Key Strengths

Python-Native:

  • Pure Python
  • NumPy/SciPy ecosystem
  • Easy to learn
  • Jupyter integration
  • Readable code

Excellent Documentation:

  • Comprehensive tutorials
  • User guide
  • Algorithm explanations
  • Example notebooks
  • API reference

Modular Design:

  • Clean architecture
  • Extensible
  • Object-oriented
  • Well-tested
  • Maintainable

Educational Value:

  • Learning tensor networks
  • Teaching tool
  • Clear implementations
  • Algorithm transparency
  • Student-friendly

Inputs & Outputs

  • Input formats:

    • Python code
    • Model definitions
    • Parameters via dictionaries
    • HDF5 loading

  • Output data types:

    • MPS objects
    • NumPy arrays
    • Observables
    • Correlation functions
    • HDF5 exports

Interfaces & Ecosystem

Python Scientific Stack:

  • NumPy arrays
  • SciPy sparse matrices
  • Matplotlib visualization
  • Jupyter notebooks
  • Pandas (analysis)

Data Management:

  • HDF5 support
  • Pickle serialization
  • Parameter management
  • Result organization

Workflow and Usage

Installation:

# Via pip
pip install physics-tenpy

# From source
git clone https://github.com/tenpy/tenpy.git
cd tenpy
pip install -e .

Basic DMRG:

import tenpy
from tenpy.networks.mps import MPS
from tenpy.models.tf_ising import TFIChain
from tenpy.algorithms import dmrg

# Define model
model_params = {
    'L': 20,
    'J': 1.0,
    'g': 1.5,
    'bc_MPS': 'finite'
}
M = TFIChain(model_params)

# Initial state
psi = MPS.from_product_state(M.lat.mps_sites(), [0]*20, bc='finite')

# DMRG
dmrg_params = {
    'trunc_params': {'chi_max': 100, 'svd_min': 1.e-10},
    'max_sweeps': 10,
    'verbose': 1
}
info = dmrg.run(psi, M, dmrg_params)

print("Energy:", info['E'])

Custom Hamiltonian:

from tenpy.models.model import CouplingMPOModel
from tenpy.networks.site import SpinHalfSite

class MyModel(CouplingMPOModel):
    def init_terms(self, model_params):
        J = model_params.get('J', 1.0)
        g = model_params.get('g', 0.5)
        
        # Nearest-neighbor
        for u in range(len(self.lat.unit_cell)):
            self.add_coupling(J, u, 'Sz', u, 'Sz', 1)
            self.add_onsite(g, u, 'Sx')

Time Evolution:

from tenpy.algorithms import tebd

# TEBD engine
tebd_params = {
    'N_steps': 10,
    'dt': 0.1,
    'order': 2,
    'trunc_params': {'chi_max': 50}
}

engine = tebd.TEBDEngine(psi, M, tebd_params)
engine.run()

# Measure observables over time
Sz = psi.expectation_value('Sz')

Infinite System:

from tenpy.algorithms import dmrg

# Infinite DMRG
model_params = {'L': 2, 'bc_MPS': 'infinite'}
M = TFIChain(model_params)

psi = MPS.from_product_state(M.lat.mps_sites(), [0,1], bc='infinite')
info = dmrg.run(psi, M, dmrg_params)

Advanced Features

Conserved Quantum Numbers:

  • U(1) charge
  • Z2 symmetries
  • Custom conserved quantities
  • Efficiency improvements
  • Automatic handling

Correlation Functions:

  • Two-point correlations
  • Multi-point functions
  • Connected correlations
  • Structure factors
  • Analysis tools

Entanglement:

  • Entanglement entropy
  • Entanglement spectrum
  • Mutual information
  • Topological invariants
  • Quantum information

Purification:

  • Finite-temperature MPS
  • Thermal states
  • Mixed states
  • Density matrices
  • Temperature evolution

Performance Characteristics

  • Speed: Python (moderate)
  • Accuracy: DMRG precision
  • System size: 100s sites (1D)
  • Purpose: Research and education
  • Typical: Desktop to small HPC

Computational Cost

  • Python overhead
  • NumPy efficiency
  • Moderate performance
  • Good for most applications
  • Not fastest but accessible

Limitations & Known Constraints

  • Performance: Python slower than C++
  • 1D focus: Primarily 1D systems
  • 2D: Limited 2D capability
  • HPC: Not optimized for supercomputers
  • System size: Moderate vs C++ codes

Comparison with Other Tensor Network Codes

  • vs ITensor: TeNPy pure Python, ITensor C++/Julia
  • vs Block: TeNPy accessible, Block specialized
  • Unique strength: Pure Python, excellent docs, educational value, Jupyter-friendly, ease of learning, clean code

Application Areas

1D Quantum Systems:

  • Spin chains
  • 1D fermions
  • Bosonic systems
  • Quantum wires
  • Topological phases

Research:

  • Tensor network methods
  • Algorithm development
  • Many-body physics
  • Quantum simulation
  • Method testing

Education:

  • Learning DMRG
  • Teaching tensor networks
  • Student projects
  • Algorithm understanding
  • Interactive demonstrations

Production:

  • 1D calculations
  • Moderate systems
  • Rapid prototyping
  • Python workflows

Best Practices

Python Usage:

  • Jupyter notebooks
  • Interactive development
  • NumPy efficiency
  • Documentation reading
  • Example following

DMRG Parameters:

  • Appropriate chi_max
  • Convergence criteria
  • Truncation errors
  • Sweeping strategy
  • Validation

Model Building:

  • Use built-in models when possible
  • Extend CouplingMPOModel
  • Test on small systems
  • Clear parameter definitions

Community and Support

  • Open-source (GPL v3)
  • GitHub repository
  • Active development
  • Issue tracking
  • User forum
  • Responsive developers
  • Growing community

Educational Resources

  • Excellent documentation
  • Tutorial notebooks
  • User guide
  • Algorithm descriptions
  • Example gallery
  • API reference
  • Physics background

Development

  • Academic project (Germany)
  • Active development
  • Community contributions
  • Regular releases
  • Feature additions
  • Bug fixes
  • Modern Python practices

Research Impact

TeNPy has become a popular choice for tensor network calculations in Python, particularly valued for teaching and learning tensor network methods while maintaining research-quality capabilities.

Verification & Sources

Primary sources:

  1. Homepage: https://tenpy.readthedocs.io/
  2. GitHub: https://github.com/tenpy/tenpy
  3. Publications: SciPost Phys. Lect. Notes 5 (2018)

Secondary sources:

  1. Tensor network literature
  2. User publications
  3. DMRG tutorials

Confidence: VERIFIED - Python tensor network library

Verification status: ✅ VERIFIED

  • Website: ACTIVE
  • GitHub: ACCESSIBLE
  • License: GPL v3 (open-source)
  • Category: Open-source tensor network library
  • Status: Actively developed
  • Community: Growing, international
  • Specialized strength: Pure Python tensor networks, DMRG implementation, excellent documentation, educational value, Jupyter-friendly, clean architecture, 1D quantum systems, accessible learning curve, production-capable, comprehensive tutorials, NumPy-based, modular design

Related Tools in 3.4 Tensor Networks

ITensor TeNPy Block DMRG++ NORG TensorNetwork Quimb TeNeS mptensor ExaTN Uni10 TNT Library TensorCircuit merapp PyTreeNet PEPS PySCF-DMRG