ucc 0.4.4

Unitary Compiler Collection: A quantum circuit interface and compiler for multiple quantum frameworks

ucc: Unitary Compiler Collection

The Unitary Compiler Collection (UCC) is a Python library for frontend-agnostic, high performance compilation of quantum circuits. UCC's goal is to gather together the best of open source compilation to make quantum programming simpler, faster, and more scalable.

By leveraging qBraid, UCC interfaces automatically with multiple quantum computing frameworks, including Qiskit, Cirq, and PyTKET and supports programs in OpenQASM 2 and OpenQASM 3. For a full list of the latest supported interfaces, just call ucc.supported_circuit_formats.

Want to know more?

• Check out our documentation, which you can build locally after installation by running make html in ucc/docs/source.
• Read the launch announcement to learn more on the current state of UCC, its capabilities and future direction.
• For code, repo, or theory questions, especially those requiring more detailed responses, submit a Discussion.
• For casual or time sensitive questions, chat with us on Discord.

Quickstart

Installation

Note: UCC requires Python version ≥ 3.12.

pip install ucc

If developing or running benchmarks, please install Poetry, which is used to managed dependencies. Then setup a dev version via:

git clone https://github.com/unitaryfund/ucc.git
cd ucc
poetry install

For any subsequent commands mentioned in the docs, we assume you either prefix each command with poetry run, or you first activate the poetry managed virtual environment by running the output of poetry env activate in your shell.

Example with Qiskit, Cirq, and PyTKET

Define a circuit with your preferred quantum SDK and compile it!

from ucc import compile

from pytket import Circuit as TketCircuit
from cirq import Circuit as CirqCircuit
from qiskit import QuantumCircuit as QiskitCircuit
from cirq import H, CNOT, LineQubit

def test_tket_compile():
    circuit = TketCircuit(2)
    circuit.H(0)
    circuit.CX(0, 1)
    compile(circuit)

def test_qiskit_compile():
    circuit = QiskitCircuit(2)
    circuit.h(0)
    circuit.cx(0, 1)
    compile(circuit)

def test_cirq_compile():
    qubits = LineQubit.range(2)
    circuit = CirqCircuit(
        H(qubits[0]),
        CNOT(qubits[0], qubits[1]))
    compile(circuit)

How does UCC stack up?

UCC seeks to provide an end-to-end compiler that works well for the majority of the users out of the box. Today, this is achieved by running a particular subset of Qiskit transpiler passes. To ensure we continue to improve performance and meet user needs, we regularly run benchmarks comparing UCC against the latest versions of leading quantum compiler frameworks across a range of circuits. Here’s the latest:

And here you can see progress over time, with new package versions labeled for each compiler: where pytket-peep indicates the FullPeepHoleOptimize function of PyTKET. Note that the compile times before 2024-12-10 may have been run on different classical compute instances, so their runtime is not reported here, but you can find this data in benchmarks/results. After 2024-12-10, all data present in this plot is on the same compute instance using our ucc-benchmarks GitHub Actions workflow.

To learn more about running these benchmarks, the overall benchmark philosophy, or how to contribute to improving the benchmarking methodology, check out the benchmarking section in the docs.

Contributing

We’re building UCC as a community-driven project. Your contributions help improve the tool for everyone! There are many ways you can contribute, such as

• Create a Custom Compiler Pass: Learn how in the User Guide
• Submit a bug report or feature request: Submit a bug report or feature request on GitHub.
• Contribute Code: Follow the Contribution Guide to submit new passes and improvements.

If you have questions about contributing please ask on the Unitary Foundation Discord.

License

UCC is distributed under GNU Affero General Public License version 3.0(AGPLv3). Parts of ucc contain code or modified code that is part of Qiskit or Qiskit Benchpress, which are distributed under the Apache 2.0 license.