Hi, I’m lead developer and author of PyQrack, a popular open-source quantum computer simulator for Python with over 1.1 million recorded downloads to date. I notice that (at least) the paid version of ChatGPT has the capability to use the Qiskit Aer simulator in its Python environment, indicating to me that OpenAI and your users have interest in quantum computing and its simulation. I’d like to ask that OpenAI please consider the potential benefits of including PyQrack simulation capabilities in ChatGPT’s Python environments. (PyQrack can also readily interoperate with Qiskit and many other open-source quantum computing APIs and frameworks in the ecosystem, like PennyLane and Catalyst.)
PyQrack can compile with the options of OpenCL support, CUDA support, or for purely CPU-based environments. I just packaged the first release of a (drop-in replacement) pyqrack-cpu variant of the module on PyPi, for purely CPU-based simulation, which might be the most appropriate build version for ChatGPT’s Python environment. I am very ready and willing to work with any engineering team to facilitate packaging needs, feature requests, and general support to integrate the package into their workflows. The project has been developed very actively, for 7 years, though the development project is now regarded as mature or nearly “complete” (barring the discovery of new techniques for classical simulation or emulation of quantum computers).
The module is extremely light-weight, at about 2.1 MB or less for the purely-CPU version compressed Python wheel files. Besides pure C++ language standard (backward-compatible to C++11 and fully forward-compatible), its only optional (or required) external dependency in standard release builds is OpenCL (or alternatively CUDA in custom source builds). The PyQrack wrapper on the C++ Qrack library is written in pure ctypes, with no Python module dependencies required alongside to install, making the module theoretically compatible back to Python version 2.3, while being fully forward-compatible to all Python v3.x versions.
It’s a very fast simulator and emulator, with a replete collection of novel optimizations and simulation techniques from across the peer-reviewed domain literature. Our own first report on the software was accepted to IEEE QCE’23 (“Quantum Week”) and can be read at arXiv:2304.14969. It is also designed to be extremely user-friendly and simple to use, as paramount design principle.
You can probably tell, Qrack is my “baby.” I work at Unitary Fund, a small, international, nonprofit R&D company for quantum computing technologies and open-source community building. We offer PyQrack under an MIT License (while the underlying C++ Qrack library is under the LGPL v3.0), requiring no practically-limiting restriction on the licensing or use of PyQrack in both open and proprietary software stacks.
Thank you for considering my proposal, and I’m very happy and eager to answer any questions, support the process of integration, and do anything else I can see PyQrack improve the capabilities of ChatGPT! (ChatGPT absolutely rocks, by the way, and thank you for many fulfilling conversations them/her/him!)