qiskit-trev 0.1.1

GPU-accelerated tensor ring VQA simulation plugin for Qiskit, powered by PyTorch

qiskit-trev

Qiskit TREV is a GPU-accelerated quantum circuit simulation plugin for Qiskit, built on PyTorch. It provides efficient variational quantum algorithm (VQA) simulation using tensor ring (periodic Matrix Product State) representations, powered by PyTorch's GPU acceleration.

Features

• Tensor Ring Architecture: Efficient quantum state representation using periodic Matrix Product States
• PyTorch Backend: GPU acceleration via PyTorch tensors and CUDA
• Qiskit Integration: Works seamlessly as a Qiskit plugin with BackendV2 interface
• Multiple Measurement Methods:
  • Full Contraction
  • Perfect Sampling
  • Efficient Contraction
  • Right Suffix Contraction
• Variational Algorithm Support: Built-in parameter-shift rule gradient computation
• Hamiltonian Operations: Full support for Pauli string Hamiltonians via SparsePauliOp

Requirements

• Python 3.10+
• NVIDIA GPU with CUDA support
• PyTorch with CUDA
• Qiskit >= 1.0
• NumPy

Installation

pip install qiskit-trev

For development:

git clone https://github.com/keunjunpark/qiskit-trev.git
cd qiskit-trev
pip install -e ".[dev]"

Tutorials

See the tutorials/ directory for Jupyter notebooks:

1. Getting Started — Circuits, sampling, and the TREV backend
2. Expectation Values — Hamiltonians, estimator, and measurement methods
3. VQE Optimization — Gradient descent and CMA-ES for variational algorithms
4. Auto Batch Size — GPU memory-aware batch size tuning for parameter-shift gradients

Architecture

qiskit_trev/
├── __init__.py               # Public API
├── backend.py                # TREVBackend (Qiskit BackendV2)
├── estimator.py              # TREVEstimator (Qiskit Estimator primitive)
├── sampler.py                # TREVSampler (Qiskit Sampler primitive)
├── tensor_ring/              # Core tensor ring engine
│   ├── state.py              # Tensor ring state representation
│   ├── contraction.py        # Tensor contraction routines
│   └── gates.py              # Gate-to-tensor decomposition
├── measure/                  # Measurement strategies
│   ├── full_contraction.py
│   ├── perfect_sampling.py
│   ├── efficient_contraction.py
│   └── right_suffix.py
└── transpiler/               # TREV-specific transpiler passes
    └── passes.py

How It Differs from TREV

	TREV	qiskit-trev
Backend	PyTorch	PyTorch
Interface	Custom Circuit API	Qiskit BackendV2 / Primitives
Gradients	Parameter-shift rule	Parameter-shift rule
Ecosystem	Standalone	Qiskit plugin
Install	pip install TREV	pip install qiskit-trev

Contributing

Contributions are welcome! Please feel free to submit pull requests, report bugs, or suggest features.

pip install -e ".[dev]"
pytest

License

MIT License

Citation

If you use qiskit-trev in your research, please cite:

@software{qiskit_trev,
  title={qiskit-trev: PyTorch-based Tensor Ring VQA Simulation for Qiskit},
  author={Park, Keunjun},
  url={https://github.com/keunjunpark/qiskit-trev},
}

Acknowledgments

This project builds on TREV.