Perimeter Institute Quantum Intelligence Lab (PIQuIL)
Perimeter Institute Quantum Intelligence Lab (PIQuIL) is a research centre and training hub for future leaders at the intersections of artificial intelligence (AI) and quantum systems. With partners in academia, industry, and government, PIQuIL leverages the world-class scientists and resources of Perimeter Institute to catalyze leading-edge, multidisciplinary research. The rapid advance in AI is attributed to decades of research in mathematics and computer science, with inspiration ranging from neuroscience to statistics and beyond. There is a strong relationship between foundational algorithms in machine learning and the fundamental physics of complex systems.
At PIQuIL, resident physicists gather to advance the use of AI algorithms in quantum physics. Residents come from a mix of academia, government, and industry, and co-exist in a unique research space designed to foster cross-disciplinary collaboration. PIQuIL promotes the free exchange of scientific ideas, algorithms, and open-source computer codes.
PIQuIL is also a training hub, plugged into the larger academic environment of Perimeter. Physicists, already in high demand in both academia and industry, are uniquely positioned to fill the gap for PhD-level talent in AI research. PIQuIL helps to meet this demand by providing young talent with research experience and training that will enable them to become leaders in this emerging field.
In PIQuIL’s constructive environment, researchers manage their own intellectual property. Partners are free to use and adopt ideas, technology, and other properties conceived and developed at PIQuIL to further research or in for-profit products and services. Software developed under the scientific direction of the Lab will typically be licensed as open-source (e.g., Apache 2.0).
Scientists are located at the PIQuIL headquarters at the Communitech Data Hub in Waterloo’s Quantum Valley, in close proximity to entrepreneurs and innovators, with access to the Perimeter Institute’s nearby facilities and resources. Work is also often done in collaboration with AI centres and experts in Montreal, Ottawa, Sherbrooke, Toronto, Edmonton, and Vancouver.
Research topics include
- Performing quantum state and process tomography on near-term quantum devices
- Using new neural network strategies
- Identification of unconventional order, including topological phases, many-body localized states, and their associated phase transitions, using machine learning techniques
- Using neural networks for decoding and quantum error correction
- Theoretically exploring the capability of shallow and deep neural networks for the efficient representation of quantum wavefunctions
- Developing a theoretical understanding of deep learning and its relationship to the information bottleneck, phase transitions, critical slowing down, and the renormalization group
PIQuIL includes a strategic mix of resident and visiting scientists and trainees from industry and academia, all participating in collaborative research that includes a program of graduate-level courses, seminars, and workshops.
Academic program manager:
PIQuIL is expanding and actively recruiting for new condensed matter theory and quantum information faculty.
The initiative is also actively recruiting for new postdoctoral positions. Exceptional candidates will be considered for five-year appointments.
Recent publications and links
- G. Torlai (Flatiron Institute, University of Waterloo, and Perimeter Institute) et al., “Integrating Neural Networks with a Quantum Simulator for State Reconstruction”
- M.J.S. Beach (Perimeter Institute and University of Waterloo) et al., “QuCumber: wavefunction reconstruction with neural networks”
- J. Carrasquilla (Vector Institute), G. Torlai (Flatiron Institute, University of Waterloo, and Perimeter Institute), R.G. Melko (University of Waterloo and Perimeter Institute), and L. Aolita (Federal University of Rio de Janeiro and ICTP South American Institute for Fundamental Research), “Reconstructing quantum states with generative models”
- G. Torlai (University of Waterloo and Perimeter Institute) and R.G. Melko (University of Waterloo and Perimeter Institute), “Latent Space Purification via Neural Density Operators”
- G. Torlai (University of Waterloo and Perimeter Institute) et al., “Neural-network quantum state tomography”
- C. Chamberland (Institute for Quantum Computing/University of Waterloo) and P. Ronagh (Institute for Quantum Computing/University of Waterloo, Perimeter Institute, and 1QBit), “Deep neural decoders for near term fault-tolerant experiments”
- Giacomo Torlai and Roger G. Melko, "Machine learning quantum states in the NISQ era"