Perceval
Through a simple object-oriented python API, Perceval provides tools for building a circuit with linear optics components, defining single-photon sources and their error model, manipulating Fock states, running simulations, reproducing published experimental papers results, and experimenting with a new generation of quantum algorithms.
It is interfaced with the available QPUs on the Quandela cloud, so it is possible to run computations on an actual photonic computer.
Perceval aims to be a companion tool for developing discrete-variable photonics circuits
- while simulating their design, modelling their ideal and real-life behaviour;
- and proposing a normalized interface to control photonic quantum computers;
- while using powerful simulation backends to get state-of-the-art simulation;
- and also allowing direct access to the QPUs of Quandela.
Perceval has been developed as a complete toolkit for physicists and quantum computational students, researchers and practitioners.
Key Features
- Powerful Circuit designer making use of predefined components
- Simple python API and powerful simulation backends optimized in C
- Misc technical utilities to manipulate State Vector, Unitary Matrices, and circuit Parameters
- Transversal tools for visualization compatible with notebooks or local development environments
- Numerical and symbolical computation
- Modular architecture welcoming contributions from the community
Installation
Perceval requires:
- Python between 3.10 and 3.14
PIP
We recommend installing it with pip:
pip install --upgrade pip
pip install perceval-quandela
Our qiskit, qutip, cqasm or myqlm bridges have been moved to a separate
interoperability package perceval-interop (https://perceval.quandela.net/interopdocs/).
GitHub
git clone https://github.com/quandela/Perceval
then to install Perceval:
pip install .
Or for developers:
pip install -e .
Running tests and benchmarks
Unit tests files are part of the repository in tests/ and can be run with:
pip install -r tests/requirements.txt
pytest
Additionally, you can see a coverage report with the command:
pytest --cov=perceval
Benchmark tests for computing-intensive functions are in benchmark/ and can be run with:
pytest benchmark/benchmark_*.py
Comparison benchmarks for different platforms are also committed in .benchmarks/ - see pytest-benchmark documentation for more information.
Documentation and Forum
- The documentation
- The Community Forum