Distributing Quantum Computations, Shot-wise

This repository contains the code used to distribute quantum computation workloads across multiple providers, along with calibration and analysis tooling for experiments.

Reproducing the environment

• Python: 3.10 or later is recommended.
• OS: Linux or macOS with Bash and python3 available.
• Virtual environment (recommended):

  python3 -m venv .venv
  source .venv/bin/activate

• Install packages:

  pip install --upgrade pip
  pip install -r requirements.txt

Explicit reproduction packages

The following Python packages (and pinned versions) are required to reproduce the experiments. They are also listed in requirements.txt for installation via pip install -r requirements.txt.

• cirq==1.3.0
• click==8.1.7
• mypy==1.8.0
• PennyLane==0.34.0
• pyquil==3.5.4
• pytket==1.24.0
• pytket-braket==0.34.1
• pytket-cirq==0.34.0
• pytket-pennylane==0.15.0
• pytket-pyquil==0.33.0
• pytket-qir==0.8.0
• pytket-qiskit==0.47.0
• qiskit==0.45.2
• requests==2.31.0
• mqt.bench==1.0.8

Configuration

Runtime settings live in config.ini. Set your access tokens under [TOKENS] (e.g., IonQ, IBMQ) and adjust experiment parameters such as qubits, calibration rounds, and policy selections under the corresponding sections.

Running and reproducing experiments

1. (Optional) If you have compressed experiment data checked in, decompress them before running analysis:

   ./decompress.sh

2. Run a single experiment with logging via the helper script:

   ./experiment.sh exp

   The script delegates to python3 -u run.py exp ... and writes logs to experiment.log.
3. To iterate over all configurations in configs/, use:

   ./exps.sh

4. Compress generated .csv and .json artifacts before committing or archiving:

   ./compress.sh

run.py also supports calibration (cal) and analysis phases directly; refer to the inline documentation in the script for additional modes and arguments.

License

This project is licensed under the Apache License, Version 2.0. See the LICENSE file for the full text.