PyTreeNet

Official Resources

• Homepage: https://github.com/Drachier/PyTreeNet
• Documentation: arXiv:2407.13249 ("PyTreeNet: A Python Library for easy Utilisation of Tree Tensor Networks")
• Source Repository: https://github.com/Drachier/PyTreeNet
• License: EUPL v1.2

Overview

PyTreeNet is a Python library dedicated to the simulation of quantum many-body systems using Tree Tensor Networks (TTN). Developed by the "Drachier" team, it generalizes Matrix Product States (MPS) to tree-like topologies, allowing for efficient representation of systems with hierarchical entanglement structures or non-1D controlivities. It focuses on easing the implementation of complex tensor network algorithms like ground state search and time evolution.

Scientific domain: Tensor Networks, Quantum Information, Many-Body Physics Target user community: Researchers experimenting with TTN algorithms and hierarchical quantum systems

Theoretical Methods

• Tree Tensor Networks (TTN): Acyclic graph states generalizing MPS.
• Time Evolution:
  • TEBD: Time-Evolving Block Decimation generalized to trees.
  • TDVP: Time-Dependent Variational Principle for optimal manifold projection.
• Ground State Search: Variational algorithms akin to DMRG but on tree structures.

Capabilities (CRITICAL)

• Arbitrary Trees: Supports defining Hamiltonian and States on arbitrary tree graphs.
• Symbolic Input: Converts symbolic Hamiltonians into TTN operator forms (Tree MPO).
• Dynamics: Native support for real-time evolution, crucial for studying non-equilibrium quench dynamics.
• Pythonic API: high-level abstraction built on NumPy, hiding the complexity of tensor index contractions.

Key Features

Usability:

• Easy Installation: pip install pytreenet.
• Documentation: Accompanied by a detailed arXiv guide with exercises and examples.

Algorithms:

• TDVP Integration: Implements the robust TDVP algorithm, often superior to TEBD for long-time evolution and Hamiltonians with long-range terms.

Inputs & Outputs

• Input formats:
  • Python scripts defining the graph connectivity and local Hilbert spaces.
  • Symbolic Hamiltonian expressions.
• Output data types:
  • Expectation values of observables.
  • Evolved states.

Interfaces & Ecosystem

• Dependencies: NumPy, SciPy.
• Ecosystem: Can be used alongside other Python TN libraries, but functions as a standalone high-level tool.

Workflow and Usage

import pytreenet as ptn
# Define tree structure and Hamiltonian
tree = ptn.Tree(...)
H = ptn.Hamiltonian(...)
# Ground state search
psi = ptn.dmrg(H, tree)
# Time evolution
psi_t = ptn.tdvp(psi, H, dt=0.01, steps=100)

Performance Characteristics

• Efficiency: Good for systems where entanglement follows a tree hierarchy (e.g., dendrimers, effective models).
• Backend: NumPy-based (CPU), suitable for prototyping and intermediate scale problems.

Comparison with Other Codes

Verification & Sources

Primary sources:

1. GitHub Repository: https://github.com/Drachier/PyTreeNet
2. Publication: "PyTreeNet: A Python Library for easy Utilisation of Tree Tensor Networks", arXiv:2407.13249.

Verification status: ✅ VERIFIED

• Source code: OPEN (EUPL v1.2)
• State: Active research code, documented in 2024.