qmprs is a state-of-the-art package for approximately compiling quantum circuits from high-level primitives such as statevectors and operators using Quantum Matrix Product Reduced Synthesis (QMPRS). This package enables optimal state-preparation and unitary synthesis using a high fidelity approximation through Matrix Product States (MPS) and Matrix Product Operators (MPO), which enable an exponential reduction from the conventional
- python 3.10, 3.11, 3.12
- Ubuntu
Currently, due to this issue qmprs works reliably only on Ubuntu.
qmprs can be installed with the command:
pip install qmprs
Pip will handle all dependencies automatically and you will always install the latest (and well-tested) version.
To install from source:
pip install git+https://github.com/Qualition/qmprs
The notebooks are a good way for understanding how qmprs works. Depending on your preference, you may use the package as end-to-end, or use it in parts for low-level modifications.
import numpy as np
from quick.circuit import QiskitCircuit
from qmprs.synthesis.mps_encoding import Sequential
num_qubits = 10
# Generate a random state
random_state = np.random.rand(2**num_qubits) + 1j * np.random.rand(2**num_qubits)
random_state /= np.linalg.norm(random_state)
# Initialize the encoder with the preferred circuit backend
encoder = Sequential(QiskitCircuit)
# Encode the MPS
encoded_circuit = encoder.prepare_state(
random_state,
num_layers=15,
bond_dimension=512,
num_sweeps=50
)
fidelity = np.dot(random_state.conj().T, encoded_circuit.get_statevector())
# 0.9867610785253651 fidelity, 223 depth
# Isometry requires 2027 depth
# Sequential provides a 88.99 percent reduction in depth
print(f"Fidelity: {fidelity}")
print(f"Depth: {encoded_circuit.get_depth()}")Run all tests with the command:
py -m pytest tests
Run lint checks with the commands:
mypy qmprs
ruff check qmprs
If you'd like to contribute to qmprs, please take a look at our contribution guidelines. By participating, you are expected to uphold our code of conduct.
We use GitHub issues for tracking requests and bugs.
If you wish to attribute/distribute our work, please cite as per the BibTex below:
@software{qmprs2025,
title={qmprs: {Q}uantum {M}atrix {P}roduct {R}educed {S}ynthesis},
author={Amir Ali Malekani Nezhad, Tushar Pandey},
year={2025},
publisher={Zenodo},
doi={10.5281/zenodo.15437417},
url={https://doi.org/10.5281/zenodo.15437417},
}This work would not have been possible without invaluable conversations with authors of the works. We would like to especially thank Manuel S. Rudolph for discussion and feedback on the optimization of MPS encoding.
Distributed under Apache v2.0 License. See LICENSE for details.